1 OFF-LABEL PRESCRIBING OF ANTICONVULSANT DRUGS By AHUNNA JENNIFER ONYENWENYI A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2008
2 2008 Ahunna Jennifer Onyenwenyi
3 To everyone who made this possible. Thank you.
4 ACKNOWLEDGMENTS I gratefully acknowledg e the support of my supervisory committee: Almut Winterstein, Richard Segal, Earlene Lipowski, and Jonathan Shuster. I specially want to thank Almut Winterstein for all the guidance, support and encouragement given to me throughout the PhD program. I would forever be grateful. I would al so like to thank the faculty and staff of the Department of Pharmaceutical Outcomes and Poli cy. I also want to th ank my family, who has been a source of inspiration, guidance, and emotional support.
5 TABLE OF CONTENTS page ACKNOWLEDGMENTS...............................................................................................................4LIST OF TABLES................................................................................................................. ..........9LIST OF FIGURES.......................................................................................................................11LIST OF TERMS/SYMBOLS/ABBREVIATIONS..................................................................... 13ABSTRACT...................................................................................................................................14CHAPTER 1 INTRODUCTION..................................................................................................................16Background.............................................................................................................................16Need for Study........................................................................................................................18Purpose of Study.....................................................................................................................19Study Objectives.....................................................................................................................202 LITERATURE REVIEW.......................................................................................................21Regulation of Off-Label Activities......................................................................................... 21Existing Research on Off-Label Drug Use.............................................................................25Pediatric Off-Label Use Studies...................................................................................... 25Adult Off-Label Drug Use Studies.................................................................................. 29Off-label prescribing in oncology............................................................................ 29Off-label prescribing in dermatology.......................................................................30Off-label prescribing in other conditions................................................................. 30Studies Examining Factors Associated with Off-Label Use..................................................31Patient Characteristics.....................................................................................................31Physician Characteristics................................................................................................. 33Review Summary............................................................................................................34Approved Anticonvulsant Drugs............................................................................................35Mechanism of Action...................................................................................................... 36Spectrum of Clinical Efficacy......................................................................................... 37Side Effect Profile...........................................................................................................38Pharmacokinetics............................................................................................................. 39Non-Epileptic Uses of Anticonvulsant Drugs........................................................................41Use of Antiepileptic Drugs (AEDs) in Neurological Conditions Other than Epilepsy... 41Neuropathic pain......................................................................................................41Migraine...................................................................................................................42Movement disorders................................................................................................. 43Other nonepileptic neurological conditions............................................................. 43Use of Antiepileptic Drugs (AEDs) in Psychiatric Disorders......................................... 44
6 Bipolar disorder........................................................................................................44Schizophrenia........................................................................................................... 46Other psychiatric conditions.....................................................................................463 THEORETICAL FRAMEWORK..........................................................................................57Sociological Influences on Physician Decision Making........................................................ 58Patient Characteristics.....................................................................................................59Physician Characteristics................................................................................................. 61Physicians Interpersonal Rela tionship with the Patient:................................................ 62Physicians Interaction with the Medical Profession and the Health Care System......... 644 METHODOLOGY................................................................................................................. 68Research Questions............................................................................................................. ....68Data Sources...........................................................................................................................69Description of the National Ambulatory Medical Care Survey (NAMCS).................... 70Description of the National Hospital Ambulatory Medical Care Survey........................ 71Reliability of Survey Estimates....................................................................................... 72Weights for the Dataset................................................................................................... 72Intercontinental Marketing Services (IM S) Healths Office Promotion Reports Service (OPR).............................................................................................................. 73Intercontinental Marketing Services (IMS) Healths Hospital Promotion Reports Service (HPR).............................................................................................................. 73Intercontinental Marketing Services (IMS) Healths Total Sampling Report Service (TSR)............................................................................................................................73Off-Label Use Publications.............................................................................................74Study Design................................................................................................................... ........74Definition of Study Variables..........................................................................................75On-label drug visits..................................................................................................75Off-label drug visits.................................................................................................76Measurement of Study Variables.................................................................................... 77Off-label drug visit...................................................................................................77Independent variables...............................................................................................77Statistical Package for Complex Survey Design............................................................. 77Statement and Testing of Research Hypothesis......................................................................80Data Analysis...................................................................................................................82Time Series Analysis....................................................................................................... 83Logistic Regression......................................................................................................... 86Reference cell coding of independent variables....................................................... 87Variable selection..................................................................................................... 87Model building.........................................................................................................88
7 5 RESULTS...............................................................................................................................97National Ambulatory Medical Care Survey (NAMCS) Drug Visits...................................... 97National Hospital Ambulatory Medical Care Survey (NHAMCS) Drug Visits....................98National Ambulatory Medical Care Survey (NAMCS) Off-Label Anticonvulsant Drug Visits......................................................................................................................... ..........99Identification of off-Label Visits.....................................................................................99Trends in All Off-Label Anticonvulsant Drugs Visits....................................................99Trends in Off-Label Old Anticonvulsant Drugs Visits.................................................100Trends in Off-Label New Anticonvulsant Drugs Visits................................................ 100Off-Label Visits by Drug Types....................................................................................101Effect of Introduction of New Anticonvulsant Drugs on the Trend in All Medication Visits that ar e Off-Label AED Visits......................................................101Conditions Associated with Off-Label Visits................................................................ 101National Hospital Ambulatory Medical Care Survey (NHAMCS) Off-Label Anticonvulsant Drug Visits............................................................................................... 102Identification of Off-Label Visits.................................................................................. 102Trends in Off-Label Anticonvulsant Drug Visits.......................................................... 102Trends in Off-Label Old Anticonvulsant Drugs Visits.................................................103Trends in Off-Label New Anticonvulsants Drugs Visits.............................................. 103OffLabel Visits by Drug Types................................................................................... 103Effect of Introduction of New Anticonvulsa nt Drugs on Trend in All Medication Visits that are Off-Label Visits..................................................................................104Conditions Associated with Off-Label Visits................................................................ 104Effects of Pharmaceutical Promoti onal Activities on Off-Label Visits............................... 104Trends in Physician Detailing Contacts........................................................................104Association between Trends in Physician Detailing Contacts and Off-Label Visits.... 105Trends in Free Drug Sampling...................................................................................... 106Association between Trends in Free Drug Sampling and Off-Label Visits..................106Publications Concerning OffLabel Use of Anticonvulsants............................................... 107Predictors of Off-Label AEDs Visits....................................................................................108Sensitivity Analyses......................................................................................................109Sensitivity analysis with uncategorized vi sits classified as off-label visits...........109Sensitivity analysis with uncategorized visits classified as on-label visits............1106 DISCUSSION.......................................................................................................................135Overall Off-Label Drug Use................................................................................................. 135Off-Label Prescribing Trends............................................................................................... 138Off-Label Drug Use a nd Drug Promotions.......................................................................... 139Off-Label Publications......................................................................................................... .141Predictors of Off-Label Prescribing...................................................................................... 143Patient Characteristics...................................................................................................143Physician Characteristics............................................................................................... 145Visit Characteristics.......................................................................................................146
8 Limitations.................................................................................................................... ........147Policy Implications...............................................................................................................149Potential Solutions to the Various Pr oblems Posed by Off-Label Prescribing............. 150Conclusion.....................................................................................................................152APPENDIX A DESCRIPTION OF NAMCS AND NHAM CS DATA COLLECTION PROCEDURE.... 154B SEARCH STRATEGY FOR NEW ANTICONVULSANTSOFF-LABEL USE PUBLICATIONS .................................................................................................................162C ICD-9-CODES FOR ON-LABEL INDICATIONS OF AEDs ............................................ 163D SAS CODE FOR UNIVERSAL OFFLABEL I NDICATIONS OF AEDS........................ 167LIST OF REFERENCES.............................................................................................................171BIOGRAPHICAL SKETCH.......................................................................................................184
9 LIST OF TABLES Table page 2-1 Categories of Unlicen sed/Off-Label Drug Use ..................................................................482-2 Off-Label Drug Use Preval ence Studies in Pediatrics....................................................... 492-3 Off-Label Use Prevalen ce Studies in Adults..................................................................... 512-4 AEDs and their Approval Dates........................................................................................ 542-5 Main Mechanism of Actions of Old and New-Generation AEDs..................................... 542-6 Main Approved Indications of AEDs in the Treatment of Seizure Disorders................... 552-7 Other Approved and Non-A pproved Indications of AEDs................................................ 553-1 Clinical, Sociological a nd Psychological Attributes of Major Factors Affecting Physician Prescribing Behavior.........................................................................................674-1 NAMCS Sampling Frame.................................................................................................. 894-2 NHAMCS Sampling Frame...............................................................................................894-3 List of Older Anticonvulsants............................................................................................904-4 List of Newer Anticonvulsants.......................................................................................... 914-5 Pharmaceutical Promotions............................................................................................... 924-6 Journal Publication........................................................................................................ .....924-7 Patient, Physicians and Visit Variables............................................................................. 924-8 Independent Variable Referen ce Cell Coding for NAMCS Variables.............................. 955.1 Characteristics of All Anticonvulsan ts Drug Visits between 1993 and 2005 (NAMCS).........................................................................................................................1115-2 Characteristics of Ol d Anticonvulsant drug Visits between 1993 and 2005 (NAMCS).1125-3 Characteristics of New Anticonvulsant Drugs Visit between 1993 and 2005 (NAMCS).........................................................................................................................1135-4 Characteristics of All Anticonvuls ants Drug Visits between 1993 and 2005 (NHAMCS)...................................................................................................................... 114
10 5-5 Characteristics of Old Anticonvuls ant drug Visits between 1993 and 2005 (NHAMCS) ...................................................................................................................... 1155-6 Characteristics of New Anticonvulsant Drugs Visits between 1993 and 2005 (NHAMCS)...................................................................................................................... 1165-7 Characteristics of Off-Label and On-L abel Visits between 1993 and 2005 (NAMCS).. 1175-8 Categories of Off-Labe l Indications (NAMCS).............................................................. 1175-10 Characteristics of Off-Label and On-Label Visits between 1993 and 2005 (NHAMCS)...................................................................................................................... 1185-11 Categories of Off-Label Indications (NHAMCS)............................................................ 1185-13 Association between a 10% Increase in Drug Detailing Contacts and a Percent Increase in all Medication Off-Labe l AED Visits between 1994 and 2005.................... 1195-14 Association between a 10% Increase in Free Dtug Sampling and Percent Increase in all Medication Off-Label AED Visits between 1994 and 2005....................................... 1195-15 Association between a 10% Increase in Off-Label Use Publications and Percent Increase in Medication Off-Labe l AED Visits between 1994 and 2005.........................1195-16 Characteristics of NAMCS ALL AEDs Visits between 2001 and 2005.........................1195-17 Results of Logistic Regression Analysis to Determine Predictors of Off-Label Prescribing.......................................................................................................................121B-1 ICD-9 Codes for On-Label Indications from 1993 to 1999............................................. 163B-2 ICD-9 Codes for On-Label Indications from 2000 to 2005............................................. 165
11 LIST OF FIGURES page 5-1 Trends in the Proportions of All Medication Visits that are A EDs Visits between1993 and 2005 (NAMCS)................................................................................... 1225-2 Trends in the Proportions of All AED Visits that ar e Old or New AEDs Visits between 1993 and 2005 (NAMCS).................................................................................. 1225-3 Proportions of All AEDs Visits by Drug Type (NAMCS).............................................. 1235-4 Trends in the Proportions of All Medication Visits that are AEDs Visits between 1993 and 2005 (NHAMCS)............................................................................................. 1235-5 Trends in the Proportions of all AED Vi sits that are Old or New AEDs Visits between 1993 and 2005 (NHAMCS)............................................................................... 1245-6 Proportions of All AEDs Visits by Drug Type (NHAMCS)........................................... 1245-7 Trends in the Proportions of All Medication and All AEDs Visits that are Off-Label AEDs Visits between 1993 and 2005 (NAMCS)............................................................1255-8 Trends in the Proportions of All Off-Label AED Visits Attributable to Old or New AEDs between 1993 and 2005 (NAMCS).......................................................................1255-9 Trends in the Proportions of All Old a nd All New AEDs Visits that are Off-Label Visits between 1993 and 2005 (NAMCS).......................................................................1265-10 Proportions of All Off-Labe l Visits by Drug Type (NAMCS)....................................... 1265-11 Frequencies for On-label and Off-Label Visits by Drug Types (NAMCS).....................1275-12 Trends in the Proportions of All Medication and All AEDs Visits that are Off-Label AEDs Visits between 1993 and 2005 (NHAMCS)..........................................................1275-13 Trends in the Proportions of All Off-Label AED Visits Attributable to Old or New AEDs between 1993 and 2005 (NHAMCS)....................................................................1285-14 Trends in the Proportions of All Off-Label AED Visits Attributable to Old or New AEDs between 1993 and 2005 (NHAMCS)....................................................................1285-15 Proportions of Off-Label Visits by Drug Type (NHAMCS).......................................... 1295-16 Frequencies for On-label and Off-Label Visits by Drug Types (NHAMCS................... 1295-17 Trends in AEDs Detailing Contacts in Bo th Physician offices and Hospitals between 1994 and 2005..................................................................................................................130
12 5-18 Proportions of Detaili ng Contacts by Drug Types ........................................................... 1305-19 Trends in Old AED Detailing Contacts between 1994 and 2005....................................1315-20 Trends in GBP Detailing Contacts between 1994 and 2005............................................1315-21 Trends in Other New AEDs De tailing Contacts between 1994 and 2005.......................1325-22 Trends in All AED Detailing Cont acts by Settings between 1994 and 2005.................. 1325-23 Trends in AEDs Free Drug Sampling by Drug Category between 1994 and 2005......... 133
13 LIST OF TERMS/SYMBOLS/ABBREVIATIONS AED Anti-Epileptic Drugs CBZ Carbamazepine DS Divalproex ESM Ethosuximide FBM Felbamate FPHT Fosphenytoin GBP Gabapentin IMS Health Intercontinental Marketing Services LEV Levetiracetam LTG Lamotrigine NAMCS National Ambulatory Medical Care Survey NCHS National Center for Health Statistics NHAMCS National Hospital Ambulatory Medical Care Survey OXC Oxcarbazepine PB Phenobarbital PGB Pregabain PRM Primidone TGB Tiagabine TPM Topiramate VPA Valproic Acid ZNS Zonisamide
14 Abstract of Dissertation Pres ented to The Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy OFF-LABEL PRESCRIBING OF ANTICONVULSANT DRUGS By Ahunna Jennifer Onyenwenyi August 2008 Chair: Almut G. Winterstein Major: Pharmaceutical Sciences Anticonvulsant drugs are among the 10 top-se lling therapeutic drug classes and are frequently used for unapproved indications. The objective of this study was to examine national trends in off-label prescribing of anticonvulsa nt drugs in ambulatory care from 1993 to 2005, and to explore patient and physician characteristics, and other factors that might be associated with off-label prescribing of anticonvulsant drugs. Physician office and hospital outpatient depa rtmental visits mentioning anticonvulsant drug visits were identified from the National Ambulatory Medical Care Survey (NAMCS) and National Hospital Medical Care Survey ( NHAMCS) respectively. Anticonvulsants were classified as old anticonvulsa nt if approved prior to 1990 and new if approved after 1990. Onlabel visits were defined as visits where any type of convulsion related condition or epilepsy or other approved indications for an anticonvulsant drug was reported as th e visit diagnoses or reason for office or hospital visit. Off-label vis its were defined as visits at which no approved indications was reported and at least one previously reported off-label use of any of the anticonvulsants was reported. Vi sits unclassifiable based on thes e criteria were excluded from the study. Sampling weights were used to prov ide national estimate. OLS regression was employed in examining the association between pa tterns of off-label pres cribing and variations
15 in promotional activities of drug manufacturers (free drug sampling, a nd physician detailing), and the publication of off-label drug use studies. Logistic regression an alysis was employed in examining associations between patients, physic ian and visit characteristics with off-label prescribing. During the 13 years under review, about 62% (95% CI: 59%-64%) of all anticonvulsant drug visits to both physician offices and hospitals outpatient departments were visits at which at least one anticonvulsant drug wa s prescribed for an unapprov ed indication. There was a significant linear growth in the pr oportion of off-label anticonvulsa nt drug visits during the study period. Off-label use of new anticonvulsant drugs grew, while the off-label use of old anticonvulsant drugs remained over time, therefor e the introductions of the new anticonvulsant drugs had a significant effect on the s ecular trend in off-label visits Trends in promotional activities of manuf acturers and publication of off-label drug use studies were found to be associated with trends in off-label prescribi ng of anticonvulsant drugs, though not all association reached statistical significance. Fu rthermore, certain patient characteristics such age, physician characteri stics such as physicia n specialty and visit characteristics such as number of drugs prescrib ed were found to be associated with off-label prescribing of anticonvulsant drugs. Anticonvulsant drugs are increasingly used for off-label non-epileptic purposes with the magnitude of growth in off-label prescribing attr ibutable to the new agents. In light of the present political interest in the safe and cost eff ective use of prescription drugs, there is an urgent need for formulation of policies that will give providers the au tonomy to treat patients according to their best knowledge and judgment and at the same time ensure safe and cost effective use of prescription drugs.
16 CHAPTER 1 INTRODUCTION Background Off-label drug prescribing refers to the prac tice of prescribing drugs for a purpose outside the scope of the drugs approved labeled indica tion. In the U nited States, the Food and Drug Administration (FDA) has the responsibility of approving drugs fo r their labeled indications. Theoretically, a drug is prescribed off-label anytime the patient description, condition, drug dosage, or route of administrati on is different from what is sp ecifically indicated in the FDA approved labeling. The Kefauver-Harris Amendments to the US Dr ug Laws mandates that the FDA require all new drugs to pass through stringent testing to pr ove a drugs safety and efficacy in treating or managing a disease or condition. If satisfied that the drug is safe and efficacious, the manufacturer and the FDA agree on specific lang uage describing conditions, dosages, routes of administration, patient populations, adverse drug reactions, drug inte ractions and other information to be included on the drugs label. It is only for these approved indications that the risk has been systematically weighed against the benefit and accepted by the regulatory authorities. However, once a drug has been approved for a ny indication, it is common for physicians and researchers to discover othe r uses for the drug besides the FDA labeled uses. The FDA does not regulate the prescription of drugs by physicians. Physicians are generally free to prescribe drugs for any other purpose that in their professi onal judgment is both safe and effective, in other words, their prescribing authority is not limited to a drugs offi cial FDA approved indications. Therefore, off-label prescribing of most drugs is legal in the United States and in many other
17 countries. Exceptions to this are certain controll ed substances, such as opiates, which cannot be legally prescribed outside of their approved purpose.1, 2 The literature reports a high prevalence of off-label drug use. A 1991 study by the US General Accounting Office (GAO) found that one-thi rd of all drugs administered to cancer patients were for off-labeled purposes and that more than half of cancer patients received at least one drug for an off-label indication3. Another survey of AIDS patients found that 81% of patients received at least one drug off-label, and that 40% of all reported drug use was for offlabel purposes4. Finally, the American Academy of Pedi atrics (AAP) reports that about 75% of medications marketed in the US lack pediatric approval and are prescribed off-label in this population5. However, drug prescribing for off-label indi cations is not limited to cancer, AIDS, and pediatric patient populations. A study conducted in office based physicians using data from the 2001 Intercontinental Marketing Services (IMS) Healths National Disease and Therapeutic Index (NDTI) to describe prescribing patterns by diagnosis for over 100 commonly prescribed drugs reports that about 21% of prescription for the selected medications were for off-label purposes6. The authors report that off-label pr escribing was most common for cardiac medications and anticonvulsants. In fact, there ar e reports in the literature that some anticonvulsant drugs are used more frequently fo r off-label purposes than for their original approved indications6, 7. This study aims to explore the trends in offlabel use of anticonvulsa nts, explore patient and physician characteristics a nd other factors that might be associated with off-label anticonvulsant drug prescribing, a nd evaluate the effects of cer tain types of pharmaceutical promotions on off-label use of anticonvulsants.
18 Need for Study Technically, drug use for off-label indicati ons do not go through the sam e level of regulatory scrutiny as drug use for approved indications. Consequently, the effectiveness and safety of most drug/off-label indication combinati ons have not been critically examined. A well known off-label safety issue that came to light in the 90s was the combination of fenfluramine and phenteramine for weight loss. These diet drugs were never approved for long term weight loss or to be used in combination, and caused he art valve damage for many as a result of offlabel use8, 9. A more recent safety issue with offlabel drug use includes atypical and typical antipsychotic used in elderly patients with deme ntia, which has been associated with increased mortality10, 11. There is also some evidence in the literature that suggests that off-label prescribing is associated with an increased risk of medication errors, and adverse drug reactions among pediatric populations12-15. Therefore, it is important that studies are conducted to evaluate the magnitude and consequences of off-label drug use. The practice of off-label drug pr escribing also raises a number of ethical and legal issues. One frequently raised question is whether off-la bel prescribing should be viewed as a form of human experimentation, thereby tr iggering the application of safeguards established to protect human subjects16. Related controversies in clude: when off-label prescr ibing can be considered appropriate in that it confor ms to the standard of care17, whether failure to prescribe a drug for off-label use could leave a physic ian vulnerable to a malpractice suit, whether physicians are duty bound to inform patients that a product is being prescribed off-label, and whether manufacturers should be allowed to promote products for off-label purposes18. These issues have placed physicians, regulator y bodies, policy makers, and th e judicial system on opposing sides on several occasions19.
19 There is also the issue of increased cost that might be associated with off-label drug use. Off-label drug use with poor evidence of effectiveness can signifi cantly increase overall health care expenditure20. The use of drugs for such off-label indications might contribute significantly to increasing drug expenditures especially if th e drugs in question are very expensive. Examination of the prevalence and trends in off-label practices will enable regulatory bodies and drug manufacturers proj ect future development of offlabel drug use and to develop risk management plans to limit potentially unsafe and/or inappropriate o ff-label practices. Studies describing factors that contribute to offlabel prescribing will provide useful information to create profiles of patients and physicians likel y to receive and prescr ibe drugs for off-label indications. In addition, comprehensive studies on off-label drug use will provide the foundation for policy development and guidan ce on clinical decision-making. Purpose of Study Pharm aceutical promotions may play a role in off-label drug use21. When the GAO conducted its studies on off-label drug use in oncology in the early 90s; the primary issue surrounding off-label drug use was reimbursement. Doctors feared that prescriptions that were for off -label indications would not be covered by insurance comp anies. Since then, legislative action has made the issue of reimbursement le ss pressing. Todays major issue surrounding offlabel drug uses are effectiveness, safety, and drug marketing to physicians. The first two issues are more important because, even if marketing encourages off-label prescribing; it may be appropriate if the off-label use is sa fe, rational, and supported by evidence. Existing studies on off-label drug use have mostly been limited to single disease conditions, select therapeutic cl asses, and the pediatric population. Anticonvulsants were chosen in this study for three reasons: 1) they were among the top five therapeutic categories by sales volume in 2006, 2) the presence of numerous case reports of off-label use of drugs from this
20 therapeutic class in the treat ment and management of various conditions, 3) we found no published study that has evaluate d the off-label use of anticon vulsant drugs over time or examined what effects promotional activities might have on off-label prescr iption of these drugs. To begin to address these issues, an attempt was made in this study to describe national trends in off-label prescribing of anticonvulsa nts using the National Am bulatory Medical Care Survey (NAMCS) and the National Hospital Ambul atory Medical Care Survey (NHAMCS). A comparison of trends in off-label prescriptions of the new and old anticonvulsants was carried out. Anticonvulsants drugs were classified as old anticonvulsants if th ey were approved for marketing before 1990 and new anticonvulsants if they were approved for marketing during or after 1990. This study also examined patient and provider characteri stics that might be associated with of off-label use of anticonvulsant s. The study is unique in its attempt to examine the association between patterns of off-label pres cribing and variations in promotional activities of drug manufacturers (free drug sampling, and phys ician detailing), and in the publication of off-label drug use studies. Study Objectives 1. To exam ine annual trends in the prevalence of off-label prescribing of anticonvulsants in ambulatory care visits from 1993 to 2005. 2. To identify conditions for which off-label prescribing of anticonvulsants are most common. 3. To identify patient, physician, and visit characteristics that ar e associated with off-label prescribing of anticonvulsants 4. To examine the association between drug promotions to physicians and off-label prescribing behavior of physicians. 5. To examine the impact of publications of of f-label studies on off-label prescribing of anticonvulsants.
21 CHAPTER 2 LITERATURE REVIEW This chapter is divided into four parts: 1) An overview of the regulatory fram ework guiding off-label prescribing, 2) A review of existing studies on off-la bel drug use, 3) An overview of approved anticonvulsant drugs, and 4) A summary of non-epileptic uses of anticonvulsant drugs Regulation of Off-Label Activities Off-label p ractices can be classified into three categories: off-label drug use, off label prescribing and off-label promotions. Sin ce the FDA has no contro l over the practice of medicine by physicians, the focus of FDA regulat ory efforts on off-label practices has been focused on off-label promotion. Though off-label pres cribing is legal, off-label promotion in any manner has not always been considered legal. Formerly, the FDA prohibited manufacturers promotion of a product in any way for unapprov ed purposes. A Supplemental Drug Application had to be submitted, reviewed and approved before the promotion of a potentially beneficial additional indication of a drug. Ho wever, manufacturers have been reported to be reluctant in some cases to pursue a Supplemental Drug Application22, 23. There are several disincentives for ma nufacturers to pursue a Supplemental Drug Application. One disincentive frequently me ntioned is the lengthy time it takes to get a Supplemental Drug approval from the FDA22. It is reported that the supplemental approval process takes on the average more time than the original indication approval24, 25. The FDA has been reported to give supplemental drug request lower priority than original applications therefore, approval of additional uses takes longer than the original approval26. Thus, practitioners who may prescrib e drugs for off-label uses wi thout FDA approval often, find promising uses for drugs faster than the regulatory system can approve the uses. In fact, these
22 unapproved uses may become the predominant use, such that the package inserts no longer account for the dominant use of the medication. Other disincentives frequently mentioned are the high financial cost of conducti ng additional clinical trials to support an application to the FDA for label extension, and the rema ining short patent life of the drug23, 27. A significant proportion of the about 800 million dollars (in 2005 dollars) that has been estimated to be the cost of getting a new drug to the market is attrib utable to the cost of conducting phase III trials. Therefore, manufacturers may have little incentive to pursue supplemental drug applications, since the cost may outweigh the benefits. The passage of the Food and Drug Administ ration Modernization Act (FDAMA) in 1997 relaxed some of these restrictions on promo ting off-label use to physicians. The law now permits manufacturers to disseminate informa tion on unapproved uses of a drug or device to physicians, insurance issuers, group health plan s, federal and state agencies, and pharmacy benefit managers under certain restrictions26. Under FDAMA, manufacturers were allowed to disseminate information concerni ng new uses for a drug that al ready has been approved or cleared by the FDA for marketing according to the following rules. a) The manufacturer can disseminate inform ation regarding the new use only if the manufacturer has submitted a supplemental a pplication to FDA for approval of the new use. The studies to support the new use mu st have been completed or the manufacturer must submit a protocol and schedule for comp leting the studies no later than thirty-six months after the date of the initial dissemination. b) The information must appear in the form of an unabridged reprin t of a peer-reviewed article in a scientific or me dical journal that would be c onsidered scient ifically sound by experts in the area. c) The information must be accompanied by a disclosure statement indicating that the information concerns an unapproved use and that the information is being distributed at the manufacturers expense. d) The names of authors or consultants with a fi nancial interest or ties to the manufacturer must be disclosed.
23 e) The official labeling for the product must be included, along w ith information about other products that have been appr oved or cleared for the off-la bel use that the information describes. Recently the FDA issued a draft document titled:Guidance for Industry: Good Reprint Practices for the Distribution of Medical Journa l Articles and Medical or Scientific Reference Publication on Unapproved New Uses of Appr oved Drugs and Approved or Cleared Medical Devices28. This draft guidance document represents the Food and Drug Administrations current thinking on the i ssue of off-label prescribing and as such can only be viewed only as recommendations. A summary of the content of this draft guideline is given below. An Overview of FDA Recommendati ons for Good Reprint Practices The FDA guideline document on Good Reprint Pr actices addresses, the types of/articles or reference materials that can be disseminat ed and the manner in which these articles or references materials should be disseminated. a) Types of Reprints/Articles/Reference Publications According to the FDA draft guideline document, medical journal articles and scientific or medical reference publications that discuss unapproved new uses for approved drugs or approved or cleared medical devices marketed in the United States that is to be distributed by manufactures to healthcare professionals and hea lthcare entities, should: be peer reviewed and published in journals that would be considered scientifically s ound by experts in the area, should contain information that addresses adequate a nd well controlled clinical investigations, and should be generally accessible thro ugh independent distribution cha nnels, Such materials should not: be in form of special supplements, should not be materials edited or significantly influenced by a drug or device manufacturer, should not have false or misleading information or pose a significant health risk to the public health. b) Manner in which to Disseminate Scientific and Medical Information
24 The draft guideline document also addresses the manner by which scientific or medical information should be distributed to healthcare professionals and h ealthcare entities: According to the guideline such materials, a) should not be in the form of unabri dged reprint, copy of an article, or reference publication; b) shoul d not be marked, highlighted, summarized, or characterized by the manufacturer in any wa y, c) should be accompanied by the approved labeling for the drug or medical device; d) should be dissemi nated with a representative publication that reaches contrary or different conclusions re garding the unapproved use; e) should be accompanied by a prominently displayed and permanently affixed statement disclosing that the uses described in the information is fo r unapproved use, any significant safety concern concerning the unapproved use that is not discussed in the publica tion and any potential conflict of interest of the authors of the publication. Fina lly off-label journal reprint should be distributed separately from any promotional material. Generally there is an appreciati on of the potential benefit to pu blic health that might result from allowing manufacturers to disseminate truthful and non-misleading information on unapproved uses of approved drugs and approved or cleared me dical devices to healthcare professionals and healthcar e entities. However, the problem i nherent in the present situation and which is the fear expressed by some stakeholders is the potential for some manufacturers to have the inability to separate their goal of maximizing sales profits from the tended goal of the recent FDA policy which is, arming providers with info rmation so that they can make relevant prescribing decisions. Hence it is not surprising that there has been some push back on the FDA to rethink their current policy on off-label drug use promotion, since ther e is a fear that the current policy will only worsen a situation some stakeholders already consider objectionable
25 Existing Research on Off-Label Drug Use This section is an overview of existing research on off-label prescribing. T he primary focus of this review is to evaluate the incidence a nd prevalence of off-label prescribing. The Medline database was searched for articles publishe d in English language between 1990 and 2007 using the keywords; off-label, off-label drugs, o ff-label drug use, off-label prescribing, offlabel drug prescription, unapproved use, unapproved drug use, unapproved drug prescribing, or unapproved drug prescription. The resulting articles/abstracts were scanned for relevance; only studies regarding off-labe l drug use with documented a research methodology were selected for this review. The bibliogra phies/references of all studies identified were searched for additional relevant studies. The iden tified studies were then categorized as pediatric off-label drug use studies or adult off-label drug use studies. Fifty-four studies were identified and incl uded in this review; 31 in the pediatric population and 23 in the adult population. Table 2-2 lis ts the studies that we re included in this review. Pediatric Off-Label Use Studies Most studies about off-label us e in the pediatric p opulation can be said to evaluate two practices simultaneously; off-label drug use and unlicensed drug use. T h e characterization of what each practice entails varies with different authors. Broadly, off-label drug use was referred to in most studies as the use of drugs outside their licensed indications with respect to dosage, patient age, route of administ ration, indication and contra-indication, while unlicensed drug use was referred to the use of drugs whose form ulation have been modified, prepared as extemporaneous preparations, imported and used be fore a license was granted or chemicals used for therapeutic purposes.
26 The literature reports that unli censed and off-label drug use in children is widespread. This is mainly due to the fact that a large percenta ge of both old and newly marketed drugs are not licensed for pediatric use; therefore health care pr oviders who provide care to children are forced into situations where they have to use drugs ei ther in an unlicensed manner or off-label manner to ensure that children receive treatments. The lack of licen sing for the pediatric population is attributable to the fact that ve ry few controlled clinical trials are carried out in the pediatric population and therefore, th ere is quantitativel y and qualitatively insufficient data to successfully apply for drug licensing for pediatric use. Historically, drug manufacturers have been reluct ant to conduct clinical trials in children as such trials are costlier and logi stically more challenging. Furthe rmore, there is a potential for low profitability due to small population size. Cons idering the fact that the current drug licensing process was introduced in the US and European c ountries as a response to drug related tragedies affecting newborn babies, infants and children (chloramphenicol-induced gray baby syndrome, thalidomide-induced phocomelia and deaths followi ng diethylene glycol po isoning), it is ironic that few controlled trials are conducted in ch ildren. However, some initiatives aimed at increasing the number of pediatric studies have been introduced in the last few years. The voluntary Pediatric Exclusivity (PE) clause within the FDA modernization Act (FDAMA) of 1997, the Pediatric Rule (PR) in 1999, the Best Pharmaceuticals for Children Act (BPCA) of 2002, and the Pediatric Research Eq uity Act (PREA) of 2003 are various innovative steps taken to coax, stimulate and mandate dr ug manufacturers to undertake clini cal trials in children. These initiatives have recorded some me asure of success, but it is gene rally agreed that more needs to be done in order to generate more safety and efficacy data in the pediatric population.
27 Studies examining off-label and unlicensed drug use in the pediatric population have been conducted in different settings, th erapeutic areas, age ranges and c ountries, and all have revealed high off-label drug use especially in neonates. Of the 31 stud ies identified, four involved only neonatal hospital wards29-32, 18 studied pediatric hospital wards15, 33-45, and 11 were conducted in ambulatory care settings14, 46-54. The number of patients included in these studies ranged from 34 to 335,409, with an average of 25,133. About 50% of the studies were prospective studies. The criteria for assignment of off-label status of a prescription varied signif icantly between studies; some only considered one criterion of off-label use, while others considered up to six. The classification strategies used in these studies vari ed, yielding different off-l abel rates. However, when off-label and unlicensed prescriptions ar e considered together, the rates are more comparable since some drug uses that were considered off-label by some authors were considered unlicensed by other au thors. Table 2-1 shows the different criteria considered by the studies in assigning off-label dr ug use status. The most frequent category was age, followed by dose, and indication. The overall rates of prescriptions considered off-label ranged from 15% to 79% while that for unlicensed drug use ranged from 1% to 48%. The most common drugs mentioned in these studies were salbutamol, acetaminophen and caffeine. The four studies in neonatal wards were conduc ted in general, surgical, and intensive care units. Proportions of off-label/unlicensed drug use in neonatal wards were higher than in the other pediatric populations, and ranged from 47% to 79%. The studies by Turner and colleagues in 199915 and t Jong and colleagues in 200143 both examined and compared off-label/unlicensed drug use in different inpatient pe diatric settings and found the hi ghest proportion in the neonatal ward. The high proportion found in neonatal populat ion in these studies was mainly due to high numbers of unlicensed drug use. The most frequently encountered unlicensed/off-label
28 medicines were caffeine, antibiotics and vitamins. Off-label use was mainly attributable to use of different dose and frequency to that reco mmended on the product labe l while unlicensed drug use was mainly due to the manufacture or modifi cation of prescription drugs by pharmacy staff. The 16 studies from pediatric hospital wards included in this revi ew were conducted in various types of pediatric wards. The off-label/unlicensed drug use rate s resulting from these studies ranged from 43% to 100% of children (receiving at leas t one off-label or unlicensed drug) and from 15% to 61% of prescriptions. Tw o of the studies were conducted in pediatric patients with specific medical conditions; oncology35 and gastroenterology,37 while two other studies examined only specific dr ug classes; respiratory drugs51, 54 and antihypertensive drugs54. Bronchodilators, analgesics and antibiotics were the medicines most frequently prescribed unlicensed/off-label in the pedi atric inpatient population. Th e most common reasons for offlabel prescribing were patients age, followed by dose and indi cation, while unlicensed drug use occurred most often because of the manufactur ing or modification of prescription drugs by pharmacy staff. In the community setting, 12 studies evalua ted data from family practice clinics, pediatricians offices, and prescr iption databases. These stud ies had greater variations in classification of unlicensed/off-label drug use and the cut-off age for pediatric populations, therefore they were more difficult to compare. For example the study by Yoon and colleagues54 evaluated off-label with respec t to age only regardless of indication for use, and that by Buchleler and colleagues46 did not look at individual pres criptions, but evaluated whether information on the drug, its dose units or its fo rmulation was available for the age range for which the drug was prescribed. The off-label/unl icensed prescription rates ranged from 11% to 50%. Most studies considered onl y commonly prescribed drugs. In these studies in ambulatory
29 care settings topical drugs were re ported to be used more often th an systemic drugs for off-label purposes. The drugs appearing most often among the more common off-label/unlicensed drugs were antibacterial and anti-asthmatic drugs and the most common reason for off-label was dosage, age and different indications. Adult Off-Label Drug Use Studies Most studies in adult off-label drug use were conducted in specific disease conditions or certain th erapeutic classes. A majority of the studies employed cross-sectional designs to evaluate off-label prescribing. The selected studies consisted of six physician surveys3, 4, 6, 55-57, one patient survey58, five prospective59-63 and eleven retrospective studies20, 64-73. Off-label prescribing in oncology The survey by the US General Accounting O ffice (GAO) is one of the early em pirical evaluations of off-label dr ug use in the United States3. This survey study in which about 1,470 oncologists across the US participat ed gave an insight into the o ff-label prescribing patterns of oncologists. Physicians were asked to report data on patient demographics, cancer type and drugs prescribed for their last th ree cancer patients. The study evaluated physician reports of offlabel prescribing, reimbursement de nials, and prescribing changes due to reimbursement denials. One of the highlights of the reports was that more than a third of all drug administered to cancer patients were given for off-label indications and th at half of cancer patients received at least one drug in an off-label manner. In another oncology related study, Kocs and coll eagues reviewed the use of the anti-cancer drug Rituximab in a single academic center over a three year period20. They linked patient diagnoses to pharmacy records for each Rituxima b prescription and classified each prescription as either on-label or off-label according to FDA-approved indications. Out of the total 320
30 Rituximab prescriptions received by 101 patients, approximately 75% were used in an off-label manner. In another study, Poole and Dooley evaluate d 130 medication charts of hospitalized oncology patients on a single day, in a single tertiary cancer center63. Off-label drug use was assessed only in respect to approve d indications. About 85% of all patients received at least one drug for off-label indication, while 22% of all prescriptions were for off-label purposes. In summary off-label prescrib ing is reported to be common in oncology patients with more than half of these patients receiving at least one drug for off-label purposes. Off-label prescribing in dermatology Off-label prescribing is also reported to be common in dermatology. This has also been attributed to the la ck of well designed clinical trials. Sugarm an and colleagues used 8 years worth of data from the National Ambulatory Medical Care Survey to evaluate off-label medication prescribing by office based physician for dermatological diseases57. The average proportion of off-label prescribing was 32% during the study period. Li and colleagues administered a survey to de rmatologists affiliated with a large medical center (academic attending dermatologists, pr ivate practitioner fellows-in-training and residents)55. The questionnaire sought to access physicians knowledge of the approval status of drug used for specific dermatologic conditions. Dr ug-disease combinations that were off-label drug combinations were thought by more than 23% of respondents to be approved and about 5% of such combinations were believed by more than half of the respondents to be approved. Off-label prescribing in other conditions Other diseas e areas where off-label research ha s been gaining attenti on are in psychiatry, cardiology, obstetrics and gynecology. In recent years, the majority of studies that have examined off-label drug use have been in the area of psychiatry. Most of these studies report
31 that off-label use is common with estimates ranging from 20% to 70% of all psychotropic prescriptions being for off-label purposes. An tipsychotics and anticonvulsant drugs have been identified as the major psychotropic drug classes used in an off-label manner. In their study Radley and colleagues found that among the to p 100 most commonly prescribed drugs in ambulatory care, anticonvulsant drugs were amongst the top five most commonly used for offlabel purposes6. In addition, they report that gabape ntin and amtriptyline hydrochloride had the highest off-label use. A one day survey of psyc hotropic drug usage in the psychiatry wards in a state owned hospital found that one out of every three psychotr opic prescription was off-label and that off-label prescribing was particularly hi gh for anticonvulsants and to a lesser extent for mood stabilizers and anxiolytics70. Studies Examining Factors Associated with Off-Label Use Som e of the studies reviewed for determining the prevalence of off-label prescribing also document some characterization of the factors that might be associated with off-label prescribing. A review of these studies identified some factors which are consistently linked to off-label prescribing. These factors include patient characteristics, physician characteristics, drug characteristics, and hos pital characteristics. Patient Characteristics It is well docum ented that pr escription drug use and drug expens es increase with age. The study by Chen and colleagues reports that the odds of receiving off-label antidepressants, anticonvulsants or antipsychotic medications offlabel increases dramatic ally with advancing age66. Some authors have attributed this finding to the elderly tending to have a lower health status and more co-morbidity, therefore making them likely candidates to receive drugs off-label. However, in the pediatric population, the positiv e relationship between o ff-label prescribing and age observed in the adult population changes; ne onates and younger patients become more likely
32 to receive off-label medications than older pediat ric patients. This might be an indication that more fragile health state rather than age is the driver of the observed associations. The relationship between patient gender a nd off-label prescribing has also been investigated. Rijcken and colleagues in their exam ination of the sex difference in off-label use of antipsychotics in ambulatory care report that males were more often off-label users of antipsychotics than females72. The Medicaid study by Chen and colleagues found conflicting results as to the relationship betw een off-label drug use and gender66. Males were more likely to receive antidepressants for off-label purposes than females, while females were more likely to receive anticonvulsants and an tipsychotics for off-label purposes than males. Two empirical studies; one by Schrim a nd colleagues in a pediatric population49, and the other by Chen and colleagues in an adult Medicaid population66 examined the association between race and off-label drug use. Race was not found to be a significan t factor in off-label prescribing in the pediatric popula tion while Whites were reported to be more likely to receive antidepressants, anticonvulsant s and antipsychotic medications off-label than non-Whites. Severity of disease and diagnosis related como rbidites have also been reported to be important factors in off-label prescribing. The United States GAO study revealed that cancer patients with metastasized malignancies had a great er probability of receiving off-label therapy than patients with cancers in initial stages of development3. Off-label prescribing was also reported to be more prevalent for cancer types with no standard chemotherapeutic treatment. The study by Chen and colleagues reports that renal failure was associated with an increased risk of receiving off-label antidepressants, anticonvuls ants and antipsychotics drugs. In addition, a majority of the off-label antipsychotic prescriptions were written for persons with mental
33 disorders or substance abuse related conditions, while off-label anticonvulsants were used more often for pain problems associated w ith diabetes and joint diseases. Prescription drug plans can decline reimbursement of off-label prescriptions on the grounds that such uses are experimental, wh ich corresponds with standard policy on the exclusion of experimental inte rventions. The United States GAO study found that about 62% of doctors admitted cancer patients to avoid restrictions of off-label prescribing3. The GAO study also found that 23% of doctors re ported that they had been for ced to change their preferred treatment regimen in response to reimbursement restrictions. Though neither Medicare nor Medicaid specifically excludes cove rage of off-label uses backed by strong clinical data, offlabel drug coverage policy varies across insurance plans; as a result the influence of insurance on off-label prescribing is not well understood. No study has critically examined the relationship between off-label prescribing a nd prescription drug coverage, si nce most studies have been conducted in populations with one type of insurance. Physician Characteristics A few studies have evaluated the relationshi p between physician specialty and off-label prescribing. The study by Chen and colleagues found that the relations hip between off-label drug use and physician specialty was inconsistent across cohorts66. Antipsychotic recipients who were seeing a psychiatrist were more likely to receive antipsychotic medications off-label. However, for antidepressant a nd anticonvulsants recipients seeing a psychiatrist was associated with a lower likelihood of receiving these two drug categories off-label. Schrim and colleagues reported that in a pediatric ambulatory setting, sy stemic drug prescription s written by specialists were more likely to be off-label49. In contrast, Sugarman and colleagues in their study using NAMCS data to evaluate ambulatory prescrib ing for dermatologic conditions found that the
34 likelihood of off-label prescribing was signi ficantly higher for non-dermatologist when compared to dermatologists57. Review Summary In the cu rrent US health care system, there are numerous cases where there are few incentives for the pharmaceutical in dustry to conduct additional expensive clinical trials to expand the number of on-label indications. Th erefore, many drugs are prescribed extensively off-label. However, the extent to which these o ff-label uses are appropri ate or what factors are associated with off-label prescribing is not clear. The review demonstrated that off-label use is widespread ranging from 11 to 80 percent of all prescriptions. Previous off-label drug us e research has been focused mainly on oncology, pediatrics and dermatology and most recently on psychiatry and pain disorders. Furthermore, the majority of the studies documenting the prevalence of off-label drug use have been carried out in a single institution or practice setting limiting the generalizabilithy of the results. Patient age is the only variable that has been frequently asse ssed and consistently shown to affect off-label drug use. Other demographic variables such as patient race, drug coverage, and severity of illness that might contribute to off-label prescribi ng have also been examined for their influence on off-label prescribing with some conflicting results. In addition, th e inquiries into the association between physician spec ialty and off-label prescribing have also revealed conflicting results. In light of these findings, there is a need for more studies using nationwide data in order to better understand the relationships between these va riables and off-label prescribing behaviors of physicians. It is important to highlight th e fact that off-label is not synonymous with incorrect. Offlabel prescriptions might represen t the most rational, evidence based therapies so the problem is
35 not inappropriate drug use by physicians, but the in adequate evaluation and registration process. The problem behind the large extent of off-label and unlicensed use is ther efore also due to the lack of harmonization between evidence and drug license Approved Anticonvulsant Drugs Epilepsy af fects nearly 2 million people in the United States74 and represents the most common serious neurologic disorder75. The International Classification of Epileptic Seizures currently classifies seizure disorders into four major categories; partial seizures (seizures beginning locally, generalized seiz ures (seizures bilaterally sy mmetrically and without local onset), unilateral seizures (seizures that are predominantly unilateral) and unclassified epileptic seizures (seizures that are unclassifiable because of incomplete data). Despite the many recent surgical advances, medications commonly referred to as antiepileptic drugs (AEDs) also called anticonvulsants remain the mainstay of treatme nt. A drug must act on one or more target molecules in the brain before it can exhibit anti epileptic activity. Th ese targets include ion channels, neurotransmitter transporters, and neurotransmitter metabolic enzymes76. The fundamental action of AEDs is to modify the bursting properties of neurons, reduce synchronization in neuronal ensembles and inhibi t the spread of abnormal firing at distinct sites76. AEDs can be categorized according to their chemical groups into six categories. Barbiturates, used in the prophylac tic management of various type s of seizures, benzodiazepines, used mainly in the management of absence seizures and myclonic seizures, hydantoins, used mainly for the control of tonic-clonic se izures and partial seizures with complex symptomatology, oxazolidinedi ones, which were formerly the drugs of choice in the treatment of absence seizures, and are now used only in the tr eatment of partial seizures, succinimides, used mainly in the management of absence seizures, and other AEDs which include a wide variety of
36 agents. There were not more than a handf ul of AEDs introduced between 1912 and 1978; however, 11 new-generation anticonvul sant drugs have been introduced into the US market since 1993 and 2004. Some of these newer agents represent modifications of pre-existing compounds (e.g. oxcarbazepine, fosphenytoin), others were developed with the specific objective of modifying neurotransmitter function (e.g. tiagabine, vigabatrin), while others were found to be clinically useful in the treatment of epileps y (e.g. lamotrigine, topiramate, gabapentin, zonisamide and felbamate). Table 2-4 shows the various AEDs and their approval dates. Anticonvulsants are among the most commonl y prescribed centrally active agents77. The Caresmark drug trend report for 2006 reports that anticonvulsants are among the top five most prescribed drug by sales volume. A retrospectiv e analysis of a Danish outpatient prescription database revealed a prevalence of 1.1% anti convulsant drug use in a sample of about 480,000 dispensed drug prescriptions78. The use of AEDs has also b een reported to increase with increasing age78, 79. A study carried out in elderly nursing home residents in the US reports that about 11% of the residents had a prescription for AEDs79. Mechanism of Action AEDs are not easily classified into categories based on their m echanism of action. This is mainly because of the limited understanding of both the mechanism of action of most AEDs and the pathophysiology of epilepsy77. In addition, most AEDs have more than one mechanism of action, each of which may contribute to therapeutic efficacy to a variable extent depending on the type of epilepsy being treated. Nonetheless, a number of clearly important mechanisms have been identified and it is conveni ent to categorize the mechanis m of action of anticonvulsants according to those that involve a) modulati on of voltage-dependent ion channels, b) enhancement of synaptic inhibition/inhibition syna ptic excitation, c) stab ilization of thalamic neurons, d) combination of the above mechanisms, often in addition to other mechanisms76, 80.
37 Voltage dependent ion channels shape the subthreshold electrical behavior of the neuron, allow it to fire action potentials, and regulate it s responsiveness to synaptic signals. They are also integral to the generation of seizure discharges, and are critical elements in neurotransmitter release, which are required for synaptic transmission76. Consequently, modulation of these channels by some AEDs results in preventi on of high frequency re petitive neuronal firing thereby inhibiting the spr ead of seizure activity77. Blockade of voltage-dependent sodium channels is a primary mechanism of acti on of phenytoin (PHT), carbamazepine (CBZ), oxacarbamazepine OXC), lamotrigine (LTG), topiramate (TPM), zonisamide (ZNS) and felbamate (FBM). -aminobutyric acid (GABA) is the most important inhibitory transmitter in the mammalian brain and some AEDs owe their antic onvulsant effects to thei r ability to enhance GABA mediated inhibitions. Vigabatrin (VGB) and tiagabine (T GB) both increase the pool of GABA available at nerve terminals, th ereby potentiating GABAergic inhibition77. In addition, blockade of excitatory aminoreceptors also prot ects against seizures. The glutamate receptors mediate the bulk of fast excitatory neurotransmissi on in the central nervous system. Blockade of some classes of these receptors is a well recognized means of protecting against seizures. Felbamate is the best evidence among market ed anticonvulsants drugs for such glutamate receptor blocking actions76. Most AEDs have other actions in addition to those discussed above, e.g. lamotrigine (LTG) modulates sero toninergic transmission. Most AEDs inhibit seizures by utilizing one or a combination of these mechanisms. Spectrum of Clinical Efficacy Pharm acological studies have de monstrated wide variation in the spectrum of activities of various AEDs against different seizure types76. Currently marketed anticonvulsants can be
38 categorized into several broad groups according to th eir spectrum of clinical activity. AEDs that produce their anticonvulsant effect s mainly by their action on voltage dependent sodium channel blockers (e.g. PHT, CBZ, OXC, and LTG) are very effective in the control of partial and generalized tonic-clonic seizures but are ineffective in the tr eatment of generalized absence seizures. AEDs that potentiate GABAergic inhibition such as VGB and TGB are effective in partial seizures but may worsen absence seizures. Similarly barbiturates such as Phenobarbital (PB) which augment the function of GABAA receptors and have additional effects on calcium and other ion channels are effective in the cont rol of many seizure types but are ineffective in control of absence seizures. Benzodiazepines su ch as diazepam (DZP) that enhances only a subset of GABAA receptors are typically broad-spectrum agents, effective in the treatment of partial, generalized absence seizures and also myoclonic seizures. Ethosuximide (ESM) acts by affecting T-type calcium channels, and probably persistent sodium currents, are efficacious only against absence, continuous spike waves during slow sleep and possibly myoclonic seizures. Table 2-6 shows the main indications of AEDs in the treatment of seizures disorders. Side Effect Profile Because AEDs generally have a narrow therapeu tic ratio, safety problems may occur with all AEDs even at doses and serum concentrations within the recommended range, with the possible exception of GBP76. The side effect profiles of AE Ds differ substantially from one agent to another, and the likelihood of appearance of specific adverse eff ects represents one of the most important considerations in tailoring drug c hoices to the character istics of individual patients. Adverse effects can generally be categ orize as those that ar e reversible and dose dependent (e.g. ataxia, se dation, dizziness, cognitive dysfunc tion), those that are chronic and non-rapidly reversible (e.g. changes in body weight hirsutism, gingival hyperplasia), and those that are idiosyncratic (e .g. skin rashes, blood dyscrasias, liver toxicity)77. This classification is
39 however an oversimplification sin ce a precise distincti on between these categories is not always possible. All AEDs can produce central nervous syst em (CNS) side effects; this includes somnolence, drowsiness, fatigue, ataxia, irri tability, headache, restlessness, nystagmus, dizziness, vertigo, dysarthria, and mental slow ing. Most CNS effects are dose related and, usually occur if dosage is too high or increase d too rapidly, and frequently disappear during continued therapy. CBZ most often produces dizz iness, PHT ataxia and mental slowing or sedation; PB sedation and affective/behavioral/cognitive change; and valproic acid (VPA), tremor and at times sedation, primidone (PRM) causes dizziness and somnolence. Very serious side effects such as hypersensitivity reactions with rash ar e seen in about 5 to 10% of patients treated with most AEDs after start-up, and usually occur within the first 6 months of therapy76. VPA is the only old AED not known to cause these hypersensitivity reactions. Among the new AEDs, LTG has been asso ciated most frequently with idiosyncratic hypersensitivity reactions that have at times pr ogressed to Stevens-Johnson syndrome or toxic epidermal necrolysis syndrome76. Slow titration of the dose es pecially when co-administered with VPA minimizes this risk. CBZ is commonly associated with clinically unimportant leucopenia, hyponatremia and, rarely cardiac arrhythmias. All othe r older AEDs have also been associated with rare idiosyncratic hepatitis, va culitis and multi-organ failure and with some degree of decreased bone calcium and pathologic fractures. Pharmacokinetics Since pharm acokinetic variability is an importa nt determinant of the interand intraindividual differences in re sponse to AEDs, knowledge of pharmacokinetics properties is essential for a rational use of AEDs81. A marked inter-individual and intra-individual variability occurs in pharmacokinetics of AEDs under th e influence of genetic background, age related
40 factors (children, as a general rule, eliminate drugs faster compared with adults), other physiological influences (e.g. pregnancy) associated diseases (p articularly those affecting the liver and the kidney), and drug interactions. As a result patients receiving the same dosage exhibit a wide variation in serum drug concentrat ions, which, in turn, translates into important differences in clinical response. An especially important parameter is th e elimination half-life, not only because it determines the time to r each steady state after a dosage change, but also because it affects fluctuations in serum drug concentration during a dosing interval77. All of the older drugs are cleared par tly by hepatic metabolism followed by renal elimination of inactive metabolites76. As a consequence, drug interactions are complex and can result in changes in blood levels. PHT, PB, and PRM induce clearance of CBZ and VPA. Many drugs (including non-antiepileptic drugs) inhibit the metabolism of CBZ, resulting in increased serum levels thereby leading to side effects. Finally, all AEDs, with the exception of VPA, have the potential to induce the cleara nce of other drugs, such as warfarin sodium, and oral contraceptives, leading to less than expected outcomes76. Most new AEDs have pharmacokinetic propertie s that make management easier. Unlike older AEDs, most of which are eliminated by oxidative biotransformati on, many new AEDs are eliminated by renal excretion (e.g. GBP, LEV and VGB) or non-oxidative metabolism (e.g. LTG, TPM, and OXC)81. Only TGB is eliminated enti rely by oxidation. All new AEDs penetrate readily across the blood br ain barrier. Binding to plasma proteins is negligible for all new AEDs, except for TGB which is over 95% pr otein bound. The half-life of many the new AEDs are sufficiently long to allow twice daily dosing with the exception of GBP, TGB, and VGB76.
41 Non-Epileptic Uses of Anticonvulsant Drugs Non-epileptic uses of AEDs are m ost ofte n found in the areas of non-seizure related neurologic and psychiatric conditions. As far back as in the 1960s when it was observed that VPA and CBZ had mood stabilizing prosperiti es in people with recurrent mood disorders particularly those with bipolar disorder, AEDs have been used for a variety of non-epileptic conditions. The potential of AEDs for having an effect on non-epileptic conditions is not unexpected since AEDs are neur o-modulators as are, antidep ressants, and antipsychotic medications. In addition, there is evidence that epilepsy, pain syndromes, and affective disorders have common pathophysiological mechanisms82. However, the fact that a drug has antiepileptic properties should not be enough to make such drug el igible for treatment of psychiatry or other non-epileptic neurologic disorders. It is almost a given that nearly a ll AEDs approved for the management of epilepsy will be at one time be i nvestigated for their use in other non-epileptic conditions. Evidence for the benefits of the use of AEDs in these non-epileptic conditions varies widely among different drugs a nd is mostly based on case reports uncontrolled studies and few small RCTs, and thus cannot be used to determin e the efficacy and safety of these medications. Actually, the United States FDA has approved only a few AEDs for some select non-epileptic disorders: CBA for trigeminal neuralgia and acute mania; VPA for mania and migraine prophylaxis; GBP for postherptic neuralgia; and TPM for migraine prophylaxis. The section below gives a brief overview of the use of AEDs in nonepileptic indications (non epileptic neurological and psychiatry conditions). Use of Antiepileptic Drugs (AEDs) in Ne urological Conditions Other than Epilepsy Neuropathic pain Neuropathic pain refers to pain initiated or caused by a prim ary lesion or dysfunction in the nervous system. The term encompasses a heterogeneous group of disorders for which
42 standard therapeutic approaches for pain are generally disappointing. Several AEDs, including CBZ, PHT, and GBP are used in neuropathic pain and in paroxysmal nerve pain states such as trigeminal neuralgia. Multiple studies have looked at the efficacy of various AEDs in treating neuropathic pain83-87. CBZ is approved as a first line agent in trigeminal neuralgia, and GBP is approved for the treatment of postherpetic neuralgi a and is used in many other neuropathic pain conditions. Assorted painful neuropathies have also reportedly been treated with LTG, LEV, TPM, OXC, and ZNS76. Migraine Drug treatment of m igraine may be acute (abortive, symptomatic) or preventive (prophylactic)88. Triptans and ergotamine derivatives are the primary agents for migraine attacks, while drugs such as beta-blockers, calcium channel antagonists, antidepressants, serotonin antagonists and nonsteriodal anti-inf lammatory drugs are currently used for the preventive purposes. In recent years, AEDs have been investigated for use in migraine based on their action on the metabolism of GABA and glutamic acid89. In August of 2004, TPM became the second AEDs along with VPA, to be approved for migraine prevention in adults. GBP has been investigated for its possible use in the prevention of migraine with modest success90, 91. However, to date there are no convincing data to support the routine use of GBP for the prophylaxis of migraine, rather the data supports use of GBP in migraine management only after failure of standard prophylaxis regimens92 There are conflicting data on the usefulness of LTG for migraine prevention, While the results of tw o open studies suggest that lamotrigine might be useful in preventing aura associated with migraines93, 94, the results of two small placebo controlled trial indicates that lamotrigine might not be eff ective for migraine prophylaxis95.
43 Movement disorders Movem ent disorders are neurolog ical conditions that affect the speed, fluency, quality, and ease of movement. Movement disorders may be classified as dyskinesi as (abnormal fluency or speed of movement), hyperkinesia (excessive or involuntary movement), and hypokinesia (slowed or absent voluntary movement). Although of variable efficacy, AEDs are used to treat a number of movement disorders a nd neuromuscular syndrome such as essential tremor, restless leg syndrome (RLS), periodic limb movement di sorder of sleep (PLMD) and spasticity. Essential tremor is a postural and kinetic tremor caused by contraction of agonist and antagonist muscles. Some AEDs block sustained repetitive firing in neurons and this might be a reason for their efficacy in this neurological condition. PRM has been investigated in randomized clinical trials for its efficacy in essential tremors with some degree of success96. The efficacy of primidone in the treatment of essential tremor has been reported to be on par with that of propranolol97. Both are considered first line drugs in the management of essential tremors. Two small studies have evaluated the efficacy of GBP in essential tremors with conflicting results98, 99. However, because of the small number of patients involved (16 and 18 respectively) the trials possess low statistical power. TPM ha s been found in a small placebo controlled study to control tremors in patients with essential tremor100. However, larger st udies are required to define the place of TPM in treating this condition. Other nonepileptic neurological conditions In addition to the previously m entioned categorie s, AEDs are also used in the treatment of a variety of other nonepileptic neurological conditions, including none pileptic myoclonus, myotonia, amyotrophic lateral scle rosis, and multiple sclerosis. However, evidence for the efficacy of AEDs in these disorder s is at best inconclusive.
44 Use of Antiepileptic Drugs (AED s) in Psychiatric Disorders In the last th irty years, AEDs have become part of the pharmacological treatment of many psychiatric conditions such as bi polar disorder, impulse control disorders, aggressive behavior, substance use disorder, and refractory anxiet y disorder. Although CBZ, VPA and LTG are the only AEDs with FDA approval fo r a psychiatric disorder, virt ually every other new AED has been tested in trials or received claims of e fficacy for some psychiatry symptoms or disorders82. The use of AEDs in psychiatric disorders is ba sed on the belief that there are shared biological mechanism involved in epilepsy and these disorders. The role of assorted neurotransmitters in both AEDs and psychotropic effects has become increasingly documented and may explain the potential utility of AEDs fo r psychiatric indications. Bipolar disorder Bipolar disorder is a severe m ood disorder characterized by manic (type I) or hypomanic (type II) states with or without depression101. Lithium has been for many years the treatment of choice for bipolar disorder102. The growing awareness of the limitations of lithium treatment prompted the search for alternat ive treatment options. In this regard, the discovery of the mood stabilizing properties of some AE Ds has significantly broadened the array of treatment options for bipolar disorder88. Today the most important use of anticonvulsants in psychiatry is preventing relapse and treating depression and mani a in bipolar disorder. AEDs are generally regarded as mood stabilizers in the treatment of bipol ar disorders, like lithium. But unlike lithium, none of them has proved effective for both manic and depression. Psychiatrists sometimes combine two anticonvulsants or l ithium and an anticonvulsant, to increase effectiveness or minimize side effects by using smaller doses of each drug. Old AEDs such as CBZ, VPA have been approv ed as standard brief term treatment of manic episodes and mixed states and long term prevention of relapses of bipolar disorders. The
45 availability of a new generation of AEDs has br oadened the therapeutic choices for the treatment of bipolar patients who are resistant or intolerant to traditional mood stabilizers. In general, these drugs have more favorable tolerability prof ile and a lower potential for drug interactions as compared to traditional mood stabilizers, and this may significantly improve compliance with treatment. Among the new AEDs, only LTG has been approved for the treatment of bipolar disorder. Other new AEDs such as OXC, GBP, TPM, TGB and ZNS have been tested for their efficacy in the monotherapy treatment of refractor y bipolar disorder as well as in combination with traditional mod stabilizers103. While numerous case reports and open label studies generally suggest that GBP is efficacious in the treatment of bipolar disorder 104-108, results from uncontrolled studies tend to contradict these findings88. Review of the literature indicates that current literature on GBP is primarily based on open-label trials that eval uate small numbers of patients. The few randomized controlled trials designed to investigate the efficacy of GB P in treating bipolar disorder have concluded that ther e is no significant diffe rence in the effects of the drug compared with placebo92, 109, 110. The use of TPM in the treatment of bipolar disorder has been reviewed111. TPM has also shown some promise as a mood stabilizer after open studies112-114. However, the results from a couple of double blind randomized placebo controlled trials acute mania studies indicate that the drug is inefficacious in bipolar disease115. TGB and ZNS have been examined in few small open label studies. Available evidence appears to indicate a limited e fficacy of TGB for acute mania116-118. Moreover, unfavorable sideeffects profile is reported, in par ticular at high doses. ZNS appears to have moderate antimanic and mood-stabilizing efficacy in open label studies119.
46 Schizophrenia The first choice drugs used in the m anagement schizophrenia are the neuroleptic drugs. However, the neuroleptics are not effective in all patients and with every symptom of schizophrenia. Some AEDs are prescribed along with neurolptic drugs in the management of schizophrenia and especially schizoaffective disorders, which has features of both mood disorders and schizophrenia. Although there is no evidence from controlled studies, the American Psychiatric Associations Practice Gu idelines recommend the use of some AEDs for hallucinations and delusions as well as mood symptoms in sc hizophrenic patients who dont respond to neuroleptic drugs. Several AEDs, including CBZ, BZP, LTG and particularly VPA have been reported to be efficacious in the treat ment of narcoleptic resistant schizophrenia. VPA is commonly applied in combined therapy of schizophrenia, especially in hostile and aggressive patients120. However, there are limited number of trials that supports the use of VPA in schizophrenia, especially la rge placebo controlled trials121. Most of the studies, both uncont rolled and controlled trials of CBZ in the management of schizophrenia indicate that it decreases aggressive behavior, a nxiety and depression, but failed to improve psychosis symptoms such as hallucinations and delusions122. Therefore it is generally administered in combination therapy with neuroleptics. LTG has also been investigated for its u tility in the manageme nt of schizophrenia. However, most of the evidence of the benefit of LTG as an adj unctive agent in patients with schizophrenia is only anecdotal observations and case reports123, 124. At present there is only partial consensus on the use of AEDs in schi zophrenia resistant to the first line therapy120. Other psychiatric conditions AEDs have also been utilized in a variety of other psychiatric c onditions such as panic disorder, binge eating disorder, anxiety disorder drug and alcohol withdr aw al symptoms. CBZ
47 has been reported in small uncontrolled trials to be useful in the treatment of aggressive behavior and to facilitate sedative withdraw al, but no controlled trial have be en carried out to establish its establish its efficacy in managing these conditions125. TGB has been reported as beneficial in the management and treatment of generalized anxiet y disorder, post-traumatic stress disorder, insomnia, cocaine addiction, and impulse control disorder125. However, reliable data on the utility of TGB in managing these conditions ar e scarce. In the same vein, though there are studies that suggest that GBP might be of some benefit in attent ion deficit hyperact ivity disorder (ADHD), social anxiety di sorder, insomnia and post-traumatic stress disorder, data from well controlled trials are lacking92. Finally there is some preliminar y evidence that indicates that the PGB might be effective fo r the treatment of generalized anxiety disorder126 In summary, most information on the widening spectrum of indication for most AEDs is based on case series, open studies an d small controlled trials and therefore should be considered preliminary.
48 Table 2-1. Categories of Un licensed/Off-Label Drug Use Unlicensed Drug Use Off-label Drug Use Modification to a licensed Drug Outside Licensed Age Range Drugs Manufactured Under Special License Outside Dose/Frequency Recommendations Chemicals Used as Medicines Us ed for Non-licensed Indication Imported Unlicensed Drugs Administered by Unlicensed Route Special Formulation of Licensed Drug Drugs Contraindicated in Certain Patients
49Table 2-2. Off-Label Drug Use Pr evalence Studies in Pediatrics Studies Author/Year Drugs/Disease Setting/Country Study Design/Duration # of Patients/# of Prescriptions % offlabel/Unlic ensed Prevalence Turner et al 199644 All Drugs 1 PICU/UK Prospective Cohort Study /4 months 166/862 31 McKenzie et al 199740 All Drugs 1 ED/ USA Secondary Analysis of Medical Charts /30 days 521 34 Turner et al 199845 All Drugs 2 Pediatric Wards/UK Prospective Cohort Study /13 weeks 609/2013 18 Turner et al 199915 All Drugs 5 Pediatric Specialty Wards/ UK Prospective Cohort Study/13 weeks 936/4455 35 Conroy et al 199930 All Drugs 1 NICU/UK Prospective Cohort Study /13 weeks 79/455 55/9.9 Wilton et al 199953 63 newly marketed drugs General Practitioners/UK Secondary Data Analysis 24337 15 Conroy et al 200034 All Drugs 5 Pediatric Wards/5 European Countries Prospective Cohort Study/4 weeks 624/2262 39/7 Gavrilov et al 200039 All Drugs 1 Ambulatory Pediatric Unit/Israel Secondary Analysis of Medical Charts/2 months 132/222 26/8 Chalumeau et al 200047 All Drugs 77 Office based Pediatricians, France Cross-sectional/ Physician Survey/1 day 989/2522 29/4 Mclntyre et al 2000127 All Drugs 1 General Practice/UK Seconda ry Data Analysis/1 year 1175/3347 11/1 Craig et al, 200136 All Drugs Single Pediatric ward/N. Ireland Prospective Cohort study/1 day a week for 2 months 74/237 19/3.4 t Jong et al 200143 All Drugs 4 Pediatric Specialty wards/Netherlands Prospective Cohort Study/5 weeks 237/2139 18/48 Pandolfini et al 200241 All Drugs 9 General Pediatric wards/ Italy Prospective Cohort Study/12 weeks 1325/4265 60 PICU=Pediatric Intensive Care Unit. ED = Emergency Department, NICU= Neonatal In tensive Care Unit, UK= United Kingdom, USA= United States of America
50Table2-2. continued Studies Author/Year Drugs/Disease Setting/Country Study Design/Duration # of Patients/ # of Prescriptions % offlabel/Unlic ensed Prevalence ODonnell et al 200232 All Drugs 1 NICU/Australia Prospec tive Cohort Study/10 weeks 101/1442 47/11 t Jong et al 2002 All Drugs 1Pediatric+1 Neonatal Ward/Netherlands Prospective Cohort Study/5 months 293/1017 44/28 t Jong et al 200250 All Drugs 150 GP/Netherlands Secondary Analysis of Prescription Database/1 year 8271/17453 13/6 Buchelor et al 200246 All Drugs Office based Physicians /Germany Secondary Data Analysis/3 months /1.5 million /13 Barr et al 200229 All Drugs 1 NICU/Israel Prospectiv e Cohort Study/4 months 105/525 59 Horen et al 200214 All Drugs Office Based Physicians/France Prospective Cohort Study/4 months 1419 42 Carvalho et al 200333 All Drugs 1PICU/Brazil Prospective Cohort Study/1 day a week for 6 weeks 51/747 50/11 Dick et al 200337 All Drugs/ Gastroenterology Three Pediatric Gastroenterologist /UK Secondary Analysis of Prescription Database/ 1 year 308/777 37/12 Ufer et 200352 317 Drugs 1Prescription Data for County/Sweden Secondary Analysis of Prescription Database/ 1 year 183219/57552 6 21 Conroy et al 200335 All Drugs/Oncology 1 single Hospital Both Inpatients And Outpatients/UK Prospective Cohort Study/4 weeks 51/569 26/19 Schrim et al 200349 All Drugs Ambulatory Care/ Netherlands Secondary Analysis of Prescription Database/1 year 18943/66222 17/21 Neubert et al 2004128 All Drugs 1 Pediatric ward/Germany Pros pective cohort Study/8 month 156/740 26/0.4 t Jong et al 200451 Respiratory Drugs 150 GP/Netherlands Secondary Analysis of Random Sample from a Prescription Database/1 year 2502/5253 20/17 NICU= Neonatal Intensive Care Unit, PI CU=Pediatric Intensive Care Unit. GP= General Practitioner, UK= United Kingdom
51Table 2-2. continued Studies Author/Year Drugs/Disease Setting/Country Study Design/Duration # of Patients/# of Prescriptions % offlabel/Unlic ensed Prevalence Ekins-Dankes et al 200448 160 most commonly used drugs 161 GP/Scotland Secondary Data Analysis/1 year 167865 26 Eiland et al 200638 All Drugs 1Hospital /USA Secondary Analysis of Medical Charts /6 months 403/1383 31 Shah et al 200742 Most Commonly used Drugs 31 Tertiary Hospitals/USA Secondary Analysis of Prescription Databases/1 year 335409 79 Dellera M et al 200731 All Drugs 1 NICU/Italy Prospectiv e Cohort Study/2 months 34/176 51/12 Yoon et al 200754 Antihypertensi ve drugs Privately Insured Population/USA Secondary Data Analysis/1 year 3660 50 GP= General Practitioners, NICU= Neonatal Intensive Care Unit, USA= United States of America Table 2-3. Off-Label Use Prev alence Studies in Adults Studies Author/Year Drugs/Disease Setting/Country Study Design/Duration # of Patients/# of Prescriptions % offlabel/Unlic ensed Prevalence GAO 19913 Oncology Drugs 1470 Inpatient and Outpatient Oncologist USA Cross-Sectional/Physician Survey 56 Rayburn et al 199571 All Drugs/ Pregnant Females 1Tertiary Center/USA Secondary Analysis of Medical Charts /5 months 731 23 Brosgart et al 19964 All Drugs/HIV 690 Primary Care Providers for HIV patients/USA Cross-Sectional/ Physician Survey/Last 3 Prescriptions /4790 40 GAO= Government Accountability Office, HIV= Human Immunodeficiency Virus, USA= United States of America
522-3 continued Studies Author/Year Drugs/Disease Setting/Country Study Design/Duration # of Patients/# of Prescriptions % offlabel/Unlic ensed Prevalence Li et al 199855 22 Drugs/ Dermatology indications 55 Dermatologist/USA Cross-Sectional/Physician Survey 100 Atkinson et al 199959 All drugs/ Oncology 1 Malignancy Palliative Care Unit UK Prospective Cohort Study/4 months 76/689 32 Weiss et al 200058 Antipsychotics Patients from 3 Pharmacies/Austria Prospective Cohort Study/9 months 173 67 Chen et al 200060 Intravenous Immunoglobulin Multiple Hospitals/USA Prospec tive Cohort Study/9 months 251 52 Sugarman et al 200157 Top 10 Dermatologic Conditions Ambulatory Medical Care/USA Secondary Analysis of Annual Physician Survey Data//8 years 32 Kocs D et al 200320 Rituximab/ Oncology 1 Academic Center USA Secondary Analysis of Medical Charts /34 months 101/428 75 Rijcken et al 200372 Antipsychotics 1 General Practice Group/Netherlands Secondary Analysis of Medical Charts / 2 years 192 21 Poole et al 200463 Anticancer Drugs 1 Tertiary Cancer Center/Australia Cross-Sectional/1 day 130/1351 18/4 Barbui et al 200465 Antipsychotics 1 Psychiatric Hospital/Italy Secondary Analysis of Medical Charts /2 years 311/378 31 Loder et al 200462 Headache 1 Tertiary Care Headache Center/USA Prospective Cohort Study/30 days /379 47 Haw et al 200569 Mood Stabilizers (Lithium and Antiepileptics) 1 Tertiary Psychiatric Center/UK Secondary Analysis of Medical Charts / 1 month 249 29 Trifiro et al 200573 Antipsychotics Primary Care/Italy Secondary Data Analysis/4 years 222240 59 USA= United States of Am erica, UK= United Kingdom
53Table 2-3 continued Studies Author/Year Drugs/Disease Setting/Country Study Design/Duration # of Patients/# of Prescriptions % offlabel/Unlic ensed Prevalence Haw et al 200568 Intellectual Disability/ Psychotropics 1 Tertiary Psychiatric Center/UK Pediatric (11-23) and Adult Patients Secondary Analysis of Medical Charts 38/86 37 Chen et al 200666 Antidepressants, Anticonvulsants, Antipsychotics 1 State Medicaid Program/USA Secondary Data Analysis/1 year 76 80 64 Radley et al 20066 100 Most Commonly Prescribed drugs Ambulatory Medical Care/USA Secondary Analysis of Annual Physician Survey Data/ 1 year 21 Lin et al 200656 Beta-Blockers Ambulatory Medical Care/ USA Secondary Analysis of Annual Physician Survey Data /4 years 52 Assion et al 200764 Psychotropic Drugs 1 Hospital/Germany Secondary Analysis of Medical Charts /1 year 233/1282 12 Cras et al 200761 All Drugs 1 General Hospital/ France Pr ospective Cohort Study/ 1 day 192/1341 23 Martin-Latry et al 200770 Psychotropic Drugs 1 Psychiatric Hospital,/France Secondary Analysis of Medical Charts /1 day 75/261 39 Gleason et al 200767 Telithromycin / Infectious Disease Privately Insured/USA Prescription claims Analysis/4 months 507 48 UK= United Kingdom, USA= United States of America
54 Table 2-4. Antiepileptic Drugs (AEDs) and their Approval Dates Old AEDs Year Approved* New AEDs Year Approved* Benzodiazepines (BZD) Clonazepam (CLZ) Clorezepate (CDP) Diazepam (DZP) Lorazepam (LRZ) 1975 1972 1963 19977 Felbamate (FBM) 1993 Carbamazepine (CBZ) 1968 Fosphenytoin (FPHT) 1996 Divalproex Sodium (DS) 1983 Gabapentin (GBP) 1993 Ethosuximide (ESM) 1960 Lamotrigine (LTG) 1994 Methosuximide (MSM) 1957 Levetiracetam (LEV) 1999 Phenobarbital (PB) 1912 Oxcarbazepine (OXC) 2000 Phenytoin (PHT) 1939 Pregabalin (PGB) 2004 Primidone (PRM) 1954 Tiagabine (TGB) 1997 Valproic Acid (VPA) 1978 Topiramate (TPM) 1996 Vigabatrin (VGB) Not yet approved in the US Zonisamide (ZNS) 2000 *Approval dates for epilepsy Table 2-5. Main Mechanism of Acti ons of Old and New-Generation AEDs OLD AEDs Na+ Channel Ca+ Channel GABA receptor GABA Transaminase GABA Transporter GABAB Receptor NMDA Receptor Other Actions BZD + CBZ + + DS ESM + MSM PB + + + PHT + + PRM VPA + + + + + NEW AEDs FBM + + + + FPHT + + GBP + + + LTG + + + LEV + + + OXC + + PGB + TGB + TPM + + + + VGB + ZNS + + +
55 Table 2-6. Main Approved Indi cations of AEDs in the Trea tment of Seizure Disorders OLD AEDs Main Indication in Seizure Disorders BZD Status Epilepticus, Partia l and Generalized Seizures CBZ Partial Seizures, (with and without secondary generaliza tion) and primarily generalized tonic-clonic seizures DS Partial and generalized seizures ESM Absence Seizures, Continuous Spikes-waves during sleep (CSWS) MSM PB Partial and generalized se izures, Status epilepticus PHT Partial seizures (with and without secondary generalization) and primarily generalized tonic-clonic se izures, Status epilepticus PRM Grand mal, psychomotor, a nd focal epileptic seizures VPA Partial and generalized seizures FBM Severe epilepsies, particularly LennoxGastaut syndrome, refractory to all other AEDs FPHT Same as Phenytoin GBP Partial Seizures (with and w ithout secondary generalization) LTG Partial and generalized seizures (may aggravate severe myclonic epilepsy of infancy) LEV Partial and probably generalized seizures OXC Partial seizures (with and without secondary generalization) and primarily generalized tonic-clonic seizures PGB Partial seizures (with and w ithout secondary generalization) TGB Partial seizures (with and w ithout secondary generalization) TPM Partial and generalized seizures effi cacy against absence seizures not proven) VGB Infantile spasms (West syndrome) Par tial seizures (with a nd without secondary generalization) refractory to all other AEDs ZNS Partial and probably, generalized seizures Table 2-7. Other Approved and Non-Approved Indications of AEDs Drug FDA Approved Neurological Indications FDA Approved Psychiatric Indications Other Proposed non FDA Uses BDZ Anxiety Disorders, Ethanol Withdrawal, skeletal Muscle Relaxant Neuropathic Pain, Trigeminal Neuralgia, Non-epileptic Myoclonus, Essential Tremor, RLS, Dystonia, Insomnia, Anxiety Disorders, Alcohol Withdrawal, Tetanus CBZ Trigeminal Neuralgia, Glossopharygngeal Neuralgia Bipolar I Disorder: acute and mixed episodes, Schizophrenia Agitation, Alcohol Withdrawal Syndrome, Mental Retardation, Benzodiazepine Withdrawal, Chorea, Cocaine Dependence, Depression, Diabetes Insipidus, Neurologic pain, Pain, Psychotic disorder, Restless Legs Syndrome
56 Table 2-7 continued Drug FDA Approved Neurological Indications FDA Approved Psychiatric Indications Other Proposed non FDA Uses ESM Basic Learning Problems PB Anxiolytic Essential Tremor, Hyperbilirubinemia PHT Neuropathic pain, Trigeminal Neuralgia, Myotonia, Dystonia, Bipolar disorders VPA Bipolar Disorder Acute mania, rapid cycling bipolar disorder and mixed states, agitation, explosive aggression and impulsiveness, alcohol withdrawal FPHT Similar to PHT GBP Postherpetic Neuralgia Mild to moderate bipolar depression, Cocaine Dependence, Diabetic Peripheral Neuropathy, Fibromyalgia, Generalized Seizures, Hot Sweats, Intracranial Tumor,-Seizures, Migraine Prophylaxis, Multiple Sclerosis Complications, Neuropathic Pain, Neuropathy due to HIV, Nystagmus, Orthostatic Tremor, Partial Seizures, Restless Legs Syndrome, Social Phobia, Tardive Dyskinesia LTG Lennox-Gastaut Syndrome Bipolar I Disorder, Migrai ne, Trigeminal Neuralgia LEV Manic Bipolar I Disorder, Migraine Prophylaxis OXC Similar to CBZ PGB Diabetic Peripheral Neuropathy, Postherpetic Neuralgia, Primary Fibromyalgia Syndrome Generalized Anxiety Disorder TGB Generalized Anxiety Disorder, Spasticity TPM Lennox-Gastaut Syndrome, Migraine Prophylaxis Alcohol Dependence, Alcoholism, Bipolar Disorder, Bulimia Nervosa, Essential Tremor, Obesity, West Syndrome ZNS Parkinsons Disease
57 CHAPTER 3 THEORETICAL FRAMEWORK The aim of this chapter is to provide a review of the studies that have evaluated the effects of various factors on prescribing behaviors of physic ians. The ultimate goal of this review is to provide empirical support for the selection of vari ables that would be evaluated in this study for their potential effects on off-label prescribing behavior of physicians. Research efforts in this area can generally be said to be focu sed on the relationship between prescribing behaviors and specific characteristics of patients, physicians and the health care environment/organization. These factors have been categorized as cl inical, sociological or psychological factors129. Table 3-1 lists the different clinical, psyc hological and sociological attributes of patients, physicians and the h ealth care system that might affect physician prescribing behavior. Models that have been developed to descri be or evaluate factors influencing physician prescribing behaviors can be br oadly divided into two: sociol ogical and psychological models. The sociological models try to describe how be liefs and values may be formed and how they might reach providers, while the psychological m odels describe how providers process beliefs and values to make prescribing decisions. Bo th perspectives provide complementary views on factors that influence physician prescribing behavior. The two approaches interact as the prescribing decision enters the re al world and becomes subject to environmental constraints and patient preferences. Evaluating psychological/cognitive models that might best explain o ff-label prescribing behaviors are beyond the scope of this study and are therefore not addressed in this review.
58 Sociological Influences on Ph ysician Decision Making Investigatio n examining the relationship between socio-demographic variables and prescribing patterns can be broadly di vided into three lines of research: a) Evaluation of how information passes thr ough the social network of physicians and attempts to predict prescribing behavior based on the physicia ns position in the social network b) Evaluation of how several sociological factor s, such as age, type of practice, and geographic location influence phys ician prescribing behaviors. c) Evaluation of sources of drug information and how these information sources may influence prescribing decisions. Various theoretical models have been develope d in an effort to e xplain the sociology of physician decision making, be it prescribing decision, decision to adopt guideli nes, or decision to use electronic prescribing me thods. Arguably the most developed sociological model of physician behavior is the Diffusi on of Innovation model (DoI), first proposed by EM Rogers in 1983130. The model was initially developed to st udy how new agricultura l techniques were adopted by farmers. However, due to its utility in explaining barriers to the adoption of new innovations, it has found utility in many other fi elds including health care. The DoI model basically describes the process by which new information or ideas spread through a social network from person to person over time. The di ffusion process involves fa ctors that affect the innovation itself as well as channels used to communicate the innovation and the characteristics of the systems or environment in which this pr ocess of diffusion takes place. However, this model is better at explaining ba rriers to the uptake of innova tions, and does not fully address other sociological factors that mi ght predict physician behavior. Another model that has been used to expl ain prescribing behavior is the Eisenberg framework of physician decision making. Accord ing to Eisenberg, physic ian clinical decision making is as a result of the interplay between phy sicians, patients and socio-cultural interactions
59 as well as some biomedical considerations131. The framework proposes that sociological influences on physician medical decision making can be categorized into four sets of: a) the characteristics of the patient, b) the sociol ogical characteristics of the physician, 3) the physicians interpersonal relationshi p with the patient and 4) the physicians interaction with his profession and health care system. Patient Characteristics Physicians are influenced by soci olo gical characteristics of their patients, including patient race, gender, age, physical appearance, income cl ass, social status, and insurance status. The availability of large administrative datasets rich in information on various patient characteristics has made it possible to conduct studies evaluatin g the effects of patient factors on physician prescribing behavior. Patient age : In their review of the literature, Clarke and colleagues came to the conclusion that in cases where age should make no diffe rence, younger people are given better prognoses and less treatment than older people.132 Physicians have been reported to prefer younger patients, and have negative images of the elderly. Older patients are often regarded as sicker, but less easily or successfully treated. Linn and coll eagues reported an association between age and medication prescribing patterns among prim ary Veterans Affair (VA) physicians133. Older patients were reported to receiv e significantly more hypertensive drugs, although they were only slightly more likely to be diagnosed with hypert ension. In addition, certain drugs are prescribed more often for elderly patients than younger patien ts. Drugs that fall in this category include digitalis, tranquilizer s and analgesics. Patient gender: Some studies have found differences be tween the health care services that males and females receive after controlling for age and seriousness of disease. These differences however are not consistent. Sometimes women are reported to receive better health care
60 services, whereas in some other occasions males s eemed to get better care. For example in one report, males were more likely to get optimal care for cardiovascular diseases than females, while in another study, females were found to be more likely treated with mood altering drugs. In a study conducted using pharmacy claim data of a large Pharmacy Benefit Management (PBM) company, Roe and colleague found differences in medication utilization patterns amongst males and females across all age groups134. They reported the greates t gender differences among young adults and in patients older than 65 year s of age. A study conducted in Australia found that physicians considered health complaints no rmal for women and theref ore considered health problems more consistent with and less of a threat to the social role of women135. Patient race/ethnicity : Analyses of physician practice patt erns provide evidence to show that there are qualitative and qua ntitative differences in health care services amongst the different ethnic groups136-140. In general, Caucasians have been reported to receive better health care services than other ethnic groups. In a review of medical charts produced in three primary care groups, White patients were referred to sp ecialists more often than Black patients141. Within the psychiatric settings, after contro lling for clinical findings, non-white s have also been reported to be more likely to be involuntarily hospita lized, secluded, and pres cribed antipsychotic medications142. In addition, data from Na tional Center for Health Stat istics (NCHS) indicate that the rate of cardiac catherizati on among Blacks is significantly less than the rates in Whites. Insurance type/drug coverage : Recent studies conducted in the United States have indicated that insurance status can influence physician prescrib ing. In general, self-paying patients and patients with no health insurance tend to receive le ss prescriptions/interventions on the average than patients with some form of h ealth insurance. Medica id and Medicare patients have been reported to be less likely refe rred to specialists by primary care physicians141. In the
61 emergency room report by Perkhoff and colleague s, emergency room patients with private insurance were more likely to be admitte d than patients without health insurance142. Another study found that lack of insurance was asso ciated with transfer to public hospitals143. In other countries, with universal hea lth insurance for all citizens, costs of medication/treatments and therefore insurance status have less in fluence on physician prescribing behavior. Physician Characteristics Whereas patient ch aracteristics have been a major focus of much research on the factors influencing physician behavior, studies documenti ng the influence of physic ian characteristics on decision making have received relatively less attention. Physician age/recency of professional training : Few studies have examined the influence of physician age on prescribing behavior. Physic ians age or more specifically number of years in practice has been shown to be associated to both the volume and the quality of prescribing. Generally there are mixed reports about the dire ction of this associa tion between physician age and prescribing behaviors. Two studies have found that younger physicians prescribed antibiotics more appropri ately than older physicians144, 145. The study by Stolley and colleagues also report that physician reporte d choice of antihypertensive drugs was associated with the age of the prescribing physician; older physicians we re found to be more conservative in their drug prescribing patterns144. Older physicians were also found to be less disposed to proceeding with certain treatments without consultation with a specialist. Physicia n age has also been investigated and found to influence the c hoice of contraception for female patients146. Physician gender : The few studies that have evaluate d the potential effect of physicians gender on physicians decision making have found no differences in prescribing or hospitalization decisions of physicians, but have found gender differences in the way physicians relate to their
62 patients147-149. One study examining physicians cont raceptive prescribing behaviors, found females physicians to have more positive atti tude towards the diaphragm as a method of contraception146. Another study of attitudes and practi ces of physicians regarding osteoporosis prevention reports that female physicians were more likely to prescribe hormone replacement therapy (HRT) than male physicians. However, in evaluating gender differences in prescribing behavior, the authors have noted that female physicians tend to see more female patients who have different health problems than male patients. Physician specialty : Considerably more research on phys ician characteristics that might explain prescribing behaviors have been focuse d on the dimensions of physicians professional background. These include: physicians area of sp ecialization, the quality of and how recent the training is, and their ongoing involvement in their professional community. Most of the previous research on prescribing by specialty suggest that specialists provide care that is more consistent with published guidelines than general practitioners150-152. In their study to determine factors associated with non-adoption of diabetic treatment guidelines, Pugh and colleagues report that primary care physicians were less likely to give novel medications to diabetic patients when compared to specialists152. Most authors have attributed th e differences reported between care by general practitioners and specialists to be related to the preferre d manner of information acquisition. Specialists tend to ra te research articles and informa tion from professional meetings higher than general practitioners153. Physicians Interpersonal Rela tionship w ith the Patient: Generally, most of the literatu re on the effect of physician-pa tient interactions on physician decision making is exploratory in nature. In their book preventing medication errors and improving drug therapy outco mes Hepler and Segal154 identify three types of relationship between physicians and patients:
63 1. Paternal relationship (in which the balan ce of power is towards the physician) 2. Consumer relationship, (in which the balanc e of power is towards the patient), 3. Therapeutic relationship, (in which power is shared between physician and patient). Several patients and physicians characteristic s, such as social class, appearance, and personality, appear to affect clinical decisions th rough the interaction between physicians and patients. The authority of the physician, the activ e role of patients, the communicative skills of both physician and patient may influence medical decision making, but no formal studies have evaluated such allegation. Perceived patient demand : Physicians have been reported to change patients drug prescriptions in response to thei r perception of patients demand. With the advent of direct to consumer advertising (DTCA), patients have b ecome more informed about their drug options and have been reported to put pressure on physic ians to prescribe certain drugs based on the information they get from DTCA155-157. There are studies that have shown changes in physician prescribing patterns that are correlated with raising pharmaceu tical industry spending on DTCA158-160. However, the evidence about the effect of DTCA is weak because this conclusion is based on correlation and survey that examine physicians percep tion of the effects of DTCA on their prescribing behaviors. Past relationship between physician and patient: Prior experiences wi th patients might influence physician prescribing behavior. Prior experiences that might influence prescribing behaviors include prior drug pr escription history, prior adve rse drug reactions and prior compliance history of the patien ts. Considerable research ha s been aimed at understanding and improving patient compliance. Improving compliance is usually a reason for prescribing more user-friendly drugs, be it in term s of dosage forms or dose regimen.
64 Physicians Interaction with the Medical Pr ofession and the Health Care System It is a widely appreciated fact that physicians do not prac tice m edicine in a vacuum, but that physicians practice within ne tworks of intraand inter-orga nizational arrangements that can influence their behaviors. Health care system factors that may influence physician prescribing behavior include practice set tings, drug policies guiding drug use, laws guiding drug approval and regulation, payment structures for drug expenditures, education and regulation of health professionals, and activities of the pharmaceutic al industry. Most importantly, drug regulatory agencies and drug manufacturers ha ve direct impact on the number, type and quality of the drugs that can be prescribed in any health care syst em. Differences in laws guiding drug use and drug availability have contributed to considerable drug use differences in different countries across the world. Measures related to the payment for drugs, such as formularies or reimbursement constraints, have also been known to influence drug prescribing. A review of the studies that have examined the effects of physicians interact ions with the healthcare system indicates that most of these studies did not control for other factors that might infl uence physician behavior, while examining the effects of physician interaction w ith the health system. It is therefore difficult to evaluate the true effects of these f actors. However, most of these have reported consistent results in terms of the direction of the effects th ese factors have on prescribing behavior. Practice setting : A recent study in ambulatory patients in Taiwan reports th at accreditation of practice setting (classified as medical center, regional center, district hospital and clinics), ownership of practice setting (private or public), patient volume (low, medium and high), urbanization of practice loca tion, and geographic location of practice location were all significantly associated with antibiot ic prescribing pattern of physicians148. Williamson also reports that physicians in teaching hospitals tended to prescribe newer drugs than physicians with
65 no affiliation to a teaching hospital161. Another study reports that physicians affiliated with or working for a teaching hospital generally provided better quality of care, and that government hospitals provided better care than private hospitals162. Rosenblatt and colleagues in their study report that physicians in solo practice had higher hos pitalization rates than physicians in group practice149. Sources of information : The way physicians interact with their health care environment also determines to some extent the way profession al information gets to them, which in turn has been shown to have a major influence on physicians behavior163. Studies of methods on information transmission are a major focus in th e sociology of prescribi ng. Studies on physician information sources have mainly addressed the following issues: 1. What sources of information are available to physicians? 2. What sources do physician prefer? 3. Are choices on preferred sour ces dependent on whether the information in question is initial information or whether it is further information that is needed to prescribe? 4. How influential are the sources of informa tion in influencing prescribing decisions? 5. Do preferences for particular sources vary according to the characteristics of the provider? 6. What is the relationship between adoption sp eed and type of information sources, e.g. scientific, professional and commercial? Generally sources of physician professional information can be broadly grouped into two sources, scientific and commercial. Scientific sources include journal ar ticles, and professional associations guidelines. Commercial sources incl ude but are not limited to drug advertisements, detail men, and direct mailings from manufacturers. Othe r sources include advice from colleagues and continuing education. Studies that have surveyed physicians on the importance of several sources of information in influenci ng prescribing behavior mostly indicate that
66 physicians rank scientific sources of informa tion more highly than commercial information sources163-166. Avorn and colleagues, based on a study they conducted to determine the sources of physicians information and the effects of such information on physician behavior, suggest that physicians tend to downplay the influence of commercial sources of information on their prescribing behavior163. This might indicate that physicia ns are not aware of the strength of commercial sources of information on their behavior. It is impor tant to study and understand the effects of different sources of physician inform ation, because of the potential for information bias. Commercial channels of information are usually one directional, normally overstating the effectiveness and safety of drugs while scientific sources have the reputation of giving more balanced information. Health sector policies/regulations : The education and regulation of health professionals play an important role in determining types of tr eatments that are obtainable in any health care system. Standards set by health authorities affect the number and quality of health care professions available in a region, which in turn affects the number of interventions that can be performed or prescribed. In addition, availabi lity and professionalism of other health care workers, such as pharmacists, have also been reported to influence physician prescribing behavior. Studies have documented that pharm acists have become more important in advising physicians, counseling patient s, and monitoring drug use167, 168. In summary, various personal, organizational, and environmental factors have been shown to influence prescribing behaviors. As a result of this, any potentially successful effort to influence physician behavior must be made w ith an understanding of the organizational and environmental context within which physicians practice.
67 Table 3-1. Clinical, Sociologi cal and Psychological Attributes of Major Factors Affecting Physician Prescribing Behavior Level Clinical Sociological Psychological Patient Severity of disease, Co-morbidities, Genetic/biological factors Socio-economic characteristics, Patientphysician relationship Beliefs regarding health and illness, Perception about therapy, Adherence behaviors, Expectation and demands Physician Specialization, Knowledge, Experience Socio-demographic characteristics, Patient-Physician relationship, Education and training, Physician-health system interaction Guideline adherence Self-efficacy & motivation, Beliefs regarding illness and therapy, Attitudes and working styles, Practice philosophy & orientation Health System Disease expertise or excellence, Scope of care, Continuum of care Health care and drug policies, Health care financing, and accessibility of health care, Activities of Pharmaceutical Industry Political factors Beliefs regarding health and illness, Peer relationship & support
68 CHAPTER 4 METHODOLOGY This chapter presents the m ethodology that wa s employed to examine the objectives of this inquiry. This study utilized data from the Na tional Ambulatory Medical Care Survey (NAMCS), the National Hospital Ambulatory Medical Care Survey (NHAMCS) and Intercontinental Marketing Services (IMS) healths Integrated Promotional Services (IPS). The research questions, description of the data sources, hypotheses, and statistical methods that were used to achieve the objectives of this study are presented below. The chapter is divided into four sections: 1. Statement of research questions 2. Data sources 3. Study design/methods 4. Statement and testing of hypotheses Research Questions 1) What are the trends in off-labe l prescribing of anticonvulsants? I. Was there a change in the prevalence of off-label prescribing of older anticonvulsant drugs over time? II. Was there a change in the prevalence of off-label prescribing of newer anticonvulsants over time? III. What is the impact of the introductio n of the newer anticonvulsants on the prevalence of off-label pres cribing of anticonvulsants? 2) What conditions are frequently associated wi th off-label prescribing of anticonvulsant drugs? 3) What is the relationship between offlabel prescribing of anticonvulsants and pharmaceutical promotional activities to physicians/physician information sources? I. Is there a relationship between off-la bel prescribing of anticonvulsants and physician detailing contacts by pharmaceutical companies?
69 II. Is there a relationship between off-label prescribing of antic onvulsants and free drug sampling by pharmaceutical companies? III. Is there a relationship between off-la bel prescribing of anticonvulsants and publications describing offlabel drug indications? 4) What patient characteristics, physician char acteristics, and physician office visit characteristics are associated with o ff-label prescribing of anticonvulsants? Data Sources Data for this study were obtained from tw o m ain sources: two annual ambulatory crosssectional physician surveys: NAMCS and outpa tient department (OPD) visits portion of NHAMCS, both from the National Center for He alth Statistics (NCHS), and pharmaceutical promotion data from IMS health. Both the NAMCS and the NHAMCS are designed to meet the need for objective, reliable information about the provision and use of ambu latory medical care se rvices in the United States. Data collection from the phys ician, rather than from the patient provides an analytic base that expands information on ambulatory care coll ected through other NCHS surveys such as National Health Interview Survey (NHIS). IMS Health is a one of the largest medical in formation companies, and is reported to be one of the global sources of pharmaceutical mark et intelligence. The main objective of the Integrated Promotional Services (IPS) data fr om IMS Health is to provide the pharmaceutical industry with an understanding of what, when, a nd how much promotional activity is occurring for pharmaceutical products. However, research ers can make use of IPS data to measure, evaluate, and understand the promotional activi ties of pharmaceutical repr esentatives that are directed to office-based physicians, front-office personnel, hospital-based physicians, and directors of pharmacy. With IPS information, resear chers are also able to evaluate the extent and effectiveness of the different pharm aceutical marketing strategies.
70 Data included under the IPS banner include: 1. Office Promotion Reports Service (OPR), 2. Total Sampling Report Service (TSRS), 3. Hospital Promotion Reports Service (HPR), and 4. National Journal Audit Service. This study w ill utilize data from OPR, HPR and TSRS. A brief description of all databases used in this study is given below, with more detailed information about the NAMCS and NHAMCS provided in appendix A. Description of the National Ambulat ory Medical Care Survey (NAMCS) The NAMCS sam ples a nationally representati ve sample of visits to non-federally employed office-based physicians who are princi pally engaged in direct outpatient care activities. Non-patient care specialties, mo st notably anesthesiology, pathology, and radiology, are excluded from the survey. The survey has been conducted annually from 1973 to 1981, in 1985, and annually since 1989. Specially trained interv iewers visit the physicians prior to their participation in the survey in or der to provide them with survey materials and instruct them on how to complete the forms. The basic samp ling unit for the NAMCS is the physician-patient encounter or visit. Only visi ts to offices of non-federally employed physicians classified by the American Medical Association or the American Osteopathic Association as office-based, patient care are included in the physician universe. The NAMCS utilizes a 3-stage probability sample design to obtain probability samples of primary sampling units (PSUs), physician practices within PSUs and patient visits within practices. PSUs are geographic segments and could be counties, groups of counties, county equivalents (such as parishes or independent cities) or towns and townships within the 50 states and the District Columbia. The final stage of the selection of patient vi sits within the annual practices of sample physicians involves two steps. Fi rst, the total physician sample is divided into 52 random sub-
71 samples of approximately equal size and each su b-sample is randomly assigned to 1 of the 52 weeks in the survey year. Second, a systematic random sample of 30 visits is selected by the physician during the reporting week. For each vis it record, the NCHS calculates a visit weight using physician and visit sampling rates adjusted fo r non response. Visit weights can be used to extrapolate to national practice patt erns for office-based physicians. Description of the National Hospit al Ambulatory Medical Care Survey The NHAM CS just like the NAMCS is design ed to collect data on utilization and provision of ambulatory care services in a nationally representative sample of visits to hospitalbased emergency departments and outpatient clinic s. This survey has been conducted annually since 1992. The sampling frame for the NHAMCS is general and short-stay hospitals in the United States, excluding federal, military, and Veterans Affairs hos pitals. The survey uses a 4stage probability design with sample from geographically de fined areas, hospital within these areas, clinics within the hospital, and patient visits within clinics. The final stage is similar to the NAMCS. Hospital staff record patient demogr aphics, symptoms, procedures, diagnoses, prescribed medications, and hospita l characteristics for a systema tic random sample of patient visits during a randomly assigned 4-week reporting period. The sample data are then weighted, depending on the total number of annual visits This study will only use data from hospital outpatient department visit of the NHAMCS, since outpatient departments represent an alternative site for care that is similar to that provided in physician offices, while emergency departments provide a di fferent level of care. The diversity of the NAMCS and NHAMCS make s it an ideal data source to examine the national prevalence and secular tren ds in off-label prescribing in ambulatory care physicians in the United States. Data from 1993 to 2005 from both surveys will be used for the study.
72 Reliability of Survey Estimates The standard error is a m easure of estimating variability that occurs by chance, as only a sample is surveyed rather than taking into acc ount the whole universe. The reliability of survey estimates has to be considered as the NAMCS and NHAMCS are sample surveys. The relative standard error (RSE) is a measure of an estima tes reliability. The RSE of an estimate is obtained by dividing the standard error of the estimate (SE ( )) by the estimate itself ( ). This quantity is expressed as a percent of the estimate and is calculated as follows: RSE = 100 (SE ( ))/ ( ) Estimates with large RSE are considered unreliable. According to the NCHS, the relative standard erro r is considered to be reliable if it is 30% or less (this means a standard error of not more than 30% of the estimates). Estimates are considered to be unreliable if they are based on fewer than 30 records, regardless of the magnitude of the relative standard error. Collapsing of multiple categories into fewer numbers of groups was used as a means to increase the reliability of estimates. Weights for the Dataset Visits weights are assigned for each observa tion in both datasets in order to obtain nationally representative cross sectional estimates More specific information on the derivation of case weights for the sample years 1993 throug h 2005 can be viewed on the NCHS website. For the survey years 1993 to 2005, between 734.5 and 963.6 million annual physician office visits were made in the United States based on the weighted NAMCS estimates, with response rates between 62% and 74%. Table 41 lists details of the office visits from 1993 through 2005.
73 For the survey years 1993 to 2005, be tween 56.7 and 94.6 million annual hospital outpatient office visits were made in the United States, based on the weighted NHAMCS estimates, with response rates between 90% and 99 %. Table 4-2 lists details of the outpatient office visits from 1993 through 2005. Intercontinental Marketing Serv ices (IMS) Healths Office Promotio n Reports Service (OPR) This database contains information on prom otional activities directed to office-based physicians in the continental U.S., specifically physicians contacts w ith pharmaceutical sales representatives. To estimate the number of detailing contacts made to physicians by pharmaceutical companies, IMS Health recruits a national panel of office based physicians, (office promotions panel), who document th e amount of time spent with pharmaceuticalcompany sales representative for 1 month. Participating physicians are selected us ing a stratified random sample design and are said to be representative of practicing office physicians in the continental United States. Intercontinental Marketing Serv ices (IMS) Healths Hosp ital Promotion Reports Service (HPR) This database contains the same information as the OPR service, but the information is collected from hospital-based physicians and hosp ital directors of pharmacy. The same sampling and data collection methods are used for both th e OPR and the HPR. Since the same type of information is collected for both services, the data from both databases can be added together to give a total picture of personal pr omotion for a product or company. Intercontinental Marketing Serv ices (IMS) Healths Tota l Sampling Report Service (TSR) This database contains information on the quantity of product samples provided to office based physicians through various delivery methods. This quantity is reported by a separate panel of front-office personnel from that of the OPR pa nel. Therefore, the data on free drug samples
74 are representative of samples given to office based physicians only. Panel members report the quantity of all drug samples either left at th e office by pharmaceutical representatives or mailed to the office during a one month period. The TSR compliments the OPR service to give a fuller picture of promotion to officebased physicians. Off-Label Use Publications The Medlin e database was searched for controll ed trials (CTs) or case reports/case series articles published in English language between 1993vand 2005 using as keywords the generic name of each of the anticonvulsant drugs include d in the this study and all identified possible off-label indications of the anticonvulsant dr ug (see Appendix B). The abstracts and full text articles (for those publications without abstracts) were reviewed to determine if the conclusion were in support of off-label drug use. The numbers of publications with positive conclusions for off-label use obtained for each dr ug were aggregated by annual quarters of publication. The search strategy for the literature search is given in appendix B Study Design A tim e series study design was used to answ er research questions 1 and 3. A crosssectional study design was employed to answer research question 2 and 4. Visits of interest were all visits in both the NAMCS and NHAMCS at which at least one of the anticonvulsant study drugs was recorded as being prescribed, administered, or provided during a physician-patient encounter. Thes e visits were termed drug visits. Tables 4-3 and 4-4 list the an ticonvulsants that were studie d in this inquiry and their specific drug and generic codes in the NAMCS and NHAMCS datasets.
75 Definition of Study Variables For the purpose of this study, drug visits were grouped into two categories. The two categories w ere on-label drug visits, and off-la bel drug visits. Though the off-label use of prescription drugs can include drug use for an indi cation, in a dosage form or dose regimen, or in a particular population not stated in the approved labeling, this study considered only off-label use as characterized by use in unapproved indications The label status of an anticonvulsant drug visit was assessed using the reasons for physicia n visit (up to 3 recorded in the NAMCS and NHAMCS) and physician diagnoses (up to 3 re corded in the NAMCS and NHAMCS), which were reported for each patient visit. The reas ons for physician visits in both the NAMCS and NHAMCS are coded using the NAMCS/NHAMCS reason for visit codes. Physicians diagnoses are coded using the ICD-9-CM diagnosis classification system. The classification of the drug visits to either on label drug visits or off label drug visits were determined by an assessment of the recorded ICD-9-codes and the r ecorded NAMCS reason for visits codes for the FDA approved indications for these drugs. Si nce the major objective of this study was to evaluate off-label drug prescribing, patient visits with the following characteristics were excluded from the study in order to be able to categorize visits as on-label or off-label visits: a) All drug visits with blank primary diagnosis codes b) All drug visits with codes in the primary di agnosis field, which indicate encounters for administrative purposes and general medical examinations. On-label drug visits On-label visits were defined as drug visits where at least one approved indication for the antiepileptic drug had been repor ted. The prim ary source for determining labeled indications was the FDA website169. All labeled indications, label upda tes and dates of update during the study period were ascertained from the FDA we bsite. The approved indications were then
76 translated into corresponding IC D-9-CM codes. Appendix C lists the ICD-9-CM codes by year for each of the labeled indications for all the anticonvulsant drugs studied. Off-label drug visits Though up to 3 diagnoses and 5, 6, or 8 m edications (depending on year) can be listed for each visit, the listed medications were not tied or matched to th e listed diagnoses. For this study, a drug visit was categorized as off-label drug visi t only if the following two criteria were met: (a) None of the NAMCS/NHAMCS specific reasons or recorded ICD-9-CM codes was for a labeled indication for the drug, and (b) At least one of the reported ICD9-CM diagnosis codes was for one of the previously reported off-label uses of any of the anticonvulsant s drugs. Three official compendia were used in identifying off-label uses of anticonvulsant drugs. The compendia are: a) The American Hospital Formulary Servic e Drug Information (AHFS DI) which is published annually by the American Society of Health System Pharmacists and provides information about on-label and off-label uses supported by clinical studies, scientific meetings, and other professional communicatio ns. AHFS DI does not provide reference of the clinical studies. b) American Medical Association Drug Evalua tion (AMA DE) which is published on an annual basis and lists both labeled and off-labeled uses of drugs. AMA DE provides references of the review articles and clinical articles and clinical st udies for all drug uses. c) US Pharmacopoeia Dispensing Information (USP DI) is published by the US Pharmacopeias Convention and provides information on labeled and off-label uses for more than 11,000 US and Canadian drug products Reviews of off-label uses describe those currently medically acceptable as we ll as off-label uses which are unproven or considered inappropriate by the USP Advisory Panels. The potential off-label indications for these drugs were translated into corresponding ICD9-CM codes. The SAS code for identifying all possible off-label indicatio n of all anticonvulsant drugs can be found in appendix D. Drug visits which could not be classified as either on-la bel drug visits or off-label drugs visits based on the identified on-label reason fo r visit codes and ICD-9-CM codes and off-label ICD-9-codes, were excluded from the study. However, in evaluating the potential predictors of
77 off-label drug visits, a sensitivity analysis wa s conducted by including these excluded visits to the on-label visits category in or der to determine their effects on the model used in examining the potential predictors of off-label drug use. Measurement of Study Variables Off-label drug visit The dependent variables for this study included a) individual off-label drug visits, b) the quarterly proportion of off-label dr ug visits, and c) the annual propor tion of off-label drug visits. The denom inator for computing the proportion of of f-label drug visits was all medication visits (NAMCS OR NHAMCS visits at which any drug was prescribed or given). Independent variables The independent variables in this study can be classified into six groups: Tim e, Pharmaceutical Promotions, Scientific Publi cations, Patient Characteristics, Physician Characteristics, and Visit Characteristics. These va riables were selected ba sed on a review of the literature and/or their av ailability in the datasets employed for this study. Descriptions of the variables for this study are presente d in the tables 4-5. 4-6 and 4-7. Statistical Package for Complex Survey Design In analyzing data collected using complex survey design, disregarding the design effects such as stratification and/or clustering m ight lead to misleading results. The NAMCS and NHAMCS utilize a multistage probab ility sample design. It is relatively difficult to properly analyze survey data that have this type of st ratified multistage design using the usual commands in commonly available statistical software packag es. This is because most of these packages assume that the data have been selected by simple random sampling method. This assumption can result in estimation and inference errors, when used in analyzing data from complex survey data. Error in inferences can result from grossl y underestimated standard errors of estimates,
78 which lead to artificially small confidence intervals and anticonserva tive hypothesis testing, namely rejecting the null hypothesis when it is in fact true. Estimation errors occur mainly through ignoring the role of weights in the es timation of means, proportions, and regression coefficients, resulting in biased estimates. Therefore, when unequal probability samples are drawn, results will be biased by underestimated standard errors in unweighted analysis. Most commonly used commands in most statisti cal computing packages have the ability to compute weighted analysis. However, the samp ling variances are still computed as though the sample was selected by means of simple a random sample. Programs such SUDAAN (Research Triangle In stitute), SAS (SAS Institute, Cary, NC), CSAMPLE (Centers for Disease Control and Prevention Atlanta GA), STATA (STATA Corp. College Station TX) and SPSS (SPSS Inc. Chicago, IL) have incorporated complex survey design into their software and have made it possi ble to estimate population parameters and their associated standard errors using design based inference. This approach assumes simply that the population is fixed. The randomness of the observation results from the random numbers that are selected during sampling process rather than from a presumed underlying model. When sample designs use stratification, clustering, post stratification to known totals, or other procedures, the estimates of means, totals, pr oportions, ratios, and regression coefficients may not be linear or even known functions of th e population parameters. To estimate variance in these estimates, statistical packag es use linearization. This techni que is based on a Taylor series expansion, and can be used to construct an approximation of the functional form of the original observations that is subsequently amenable to construction of a variance estimator. The Taylor Series (also called the delta method) is one of the most widely used variance estimation methods used in survey analysis. The Tayl or series linearization is used in statistical analysis as a way of
79 converting complicated exponential functions into more simplifie d polynomial expressions that can be more easily manipulated by the given statistical program. SAS was used for data management in this study and STATA version 9.0 (College Station, Texas) was used for all analysis. STATAs co mmands for analysis of complex design require specific design variable information in the form of PSUs, weights and Strata. Survey commands allows for multiple levels of clustering but only the first level i.e. the PSU needs to be specified (STATA 9 Manual). After these design variables are set into the STATA program, the survey estimation commands are used in a manner that is essentia lly identical to that of the corresponding nonsurvey commands. These commands compute sta ndard errors using the linearization variance estimator and are based on the first order Ta ylor series linear approximation (STATA Corporation 2005). The variance estimator that is used by th e STATA program make minimal assumptions about the nature of the sample and allow for correlations within the PSU. Therefore, the observations within the sampling unit do not have to be inde pendent, that is, there can be secondary clustering. As a result, survey analysis can be completed effectively by the STATA program and will generally produce varian ce estimates that are either approximately unbiased or biased in the directi on of more conservative estimates, i.e. larger standard errors (STATA Corporation 2005). In addition, STAT A has the option for subpopulation analysis (domain analysis) that allows for specifying only a subset of the data analysis without losing survey design information from the whole data sets (produces correct variance estimates for subpopulation). Therefore, results from such anal ysis are considered to be more robust than results obtained from using by statement wh ich does not take into account survey design information for the all sampled subjects.
80 Statement and Testing of Research Hypothesis This section is presented in the form of statements of th e research hypotheses in their alternative forms (Ha). Research hypotheses relate d to research questions 1-3 and the analytical method used for their testing is given in the first segment of this section, while research hypotheses related to research question 4 and its analytical method is given in the second segment of this section. RQ1: What are the trends in off-la bel prescribing of anticonvulsants? o Ha-1a: During the study period, there was a si gnificant increase in the quarterly proportion of all off-label an ticonvulsant drug visits. o Ha-1b: During the study period, there was a significant increase in the quarterly proportion of off-label old anticonvulsant drug visits. o Ha-1c: During the study period, there was a significant increase in the quarterly proportion of off-label new an ticonvulsant drug visits. o Ha-1d: During the study period, the increase in the quarterly proportion of off-label anticonvulsant drug visits was mostly attri butable to the increase in the quarterly proportion of off-label ne w anticonvulsant visits. RQ2: What types of conditions are frequently as sociated with off-la bel prescribing of anticonvulsant drugs? o Ha-2: During the study period, psychiatric conditions were more likely than neurologic conditions to be associat ed with off-label prescribing of anticonvulsants. RQ3: What is the relationship between indus trial promotion to physicians/physician information sources and off-labe l prescribing of anticonvulsants? Physician Detailing o Ha-3a: During the study period, there was a positive association between physician detailing contacts for all anticonvulsants and all off-label anticonvulsant drugs visits. o Ha-3b: During the study period, there wa s a positive association between physician detailing contacts for all ol d anticonvulsant drugs and off-label old anticonvulsant drugs visits.
81 o Ha-3c: During the study period, there was a positive association between physician detailing contacts for all new antic onvulsant drugs and off-label new anticonvulsant drugs visits. o Ha-3d: During the study period, there wa s a positive association between physician detailing contacts for gabapentin and off-label gabapentin drug visits. o Ha-3e: During the study period, there was a positive association between physician detailing contacts for other new anticonvul sant drugs and off-label other new anticonvulsant drug visits. Free Drug Samples o Ha-3f: During the study period, there wa s a positive association between distributed free drug sampling units for a ll anticonvulsants drugs and all off-label anticonvulsant drugs visits. o Ha-3g: During the study period, there wa s a positive association between distributed free drug sampling units for old anticonvulsants and off-label old anticonvulsant drugs visits. o Ha-3h: During the study period, there wa s a positive association between distributed free drug sampling units for a ll new anticonvulsant drugs and off-label new anticonvulsant drugs visits. o Ha-3i: During the study period, there wa s a positive association between distributed free drug sampling units for gabapentin and off-label gabapentin drug visits. o Ha-3j: During the study period, there wa s a positive association between distributed free drug sampling units for other new anticonvulsant drugs and offlabel other new anticonv ulsant drug visits. Off-label Publications o Ha-3k: During the study period, there was a positive association between cumulative publications describing successf ul use of new anticonvulsant for offlabel indications and off-label new anticonvulsant drugs visits. o Ha-3l: During the study period, there wa s a positive association between cumulative publications describing successf ul use of gabapentin for off-label indications and off-labe l gabapentin visits. o Ha-3m: During the study period, there was a positive association between cumulative publications describing successf ul use of other new anticonvulsants (excluding gabapentin) for off-label indications and off-label other new anticonvulsants drug visits (excluding gabapentin)
82 Data Analysis Data from both sources (NAMCS and NHAMCS ) were analyzed separately. The NAMCS and NHAMCS were aggregated to give quarterly off-label proportion estim ates. The association between each of the following variables: 1) To tal quarterly number of physician detailing contacts, 2) Total quarterly number of free drug samples 3) Cumulative quarterly off-label use publications, and the quarterly pr oportion of off-label anticonvul sant visits were examined separately using time series regression models. Quarterly estimates of the proportion of of f-label anticonvulsant vi sits and their 95% confidence Interval (CI) were cal culated for all anticonvulsants, a nd then separately for all old, all new anticonvulsants, gabapentin and other new anticonvulsant drugs (excluding gabapentin). To estimate the quarterly proportion of off-labe l anticonvulsant visits, th e total number of all medication visits by quarters was used as the de nominator. In additi on, the total quarterly number of physician detailing contacts was calculated for all anticonvulsants and separately for old, new and Gabapentin from the IPS database from IMS Health. The same calculation was carried out on free drug sampling. Previous studies on drug promotion have reported that drug promotional effects tend to last beyond the period during which marketing was carried out170. These effects have also been reported to diminish over time. Therefore, there was a need to construct a cumulative measure of detailing to physicians that takes into consider ation the diminishing effects of detailing efforts from previous quarters in evaluating the promoti on effects of the present quarter. A method used in previous studies was used in assigning decreas ing weights to previous promotional efforts in examining the effects of physician deta iling activities on off-label practices158, 170. Present quarter physician detailing stock P(t) was defined as a weighted sum of previous quarterly
83 physician detailing efforts and present physician detailing efforts accord ing to the perpetual inventory method, P (t) = ((1) P (t-1) +Q (t)) Where is the quarterly depr eciation rate and Q (t) is the new physician detailing effort during the quarter. A quarterly depreciation rate of 30% was applied158, 170. Therefore, the physician detailing stock at the end of quarter t is the depreciation adjusted effort from previous period plus the quarters new physician detailing efforts. Time Series Analysis In tim e series regression model, as well as in most regression models, the aim is to model a stochastic relationship (i.e. includ es error terms in the specification of the model). The simplest form of a time series model between tw o variables Y and time (X) is given by: Yt = + Xt + et Where: Yt = quarterly proportion of off-labe l visits (dependent variable) Xt = time, promotional activities, and number of publica tions (independent variables), et = random disturbance term. and = unknown parameters, And the subscript t indicates that Xt and Yt are a series of observations through time. The following assumptions are made in a simple time series regression model: I. Linearity-the relationship between the Y and X variable is linear; II. Non-random independent variables III. Zero mean: E[et] = 0
84 IV. Constant variance: E [et 2] = e 2 V. Non-autoregression or the errors at differe nt time points are not correlated i.e. E[etet-m] = 0 (m 0) Ordinary Least Squares (OLS) estimators can be used to estimate and in the equation above if the five assumptions above are not viol ated. A sixth assumption; normality (i.e. error terms are normally distributed) is needed for statis tical testing of the coefficients of the model ( and ). However, the problem with most time series da ta is that disturbances which are supposedly random, that enter into the relations hip under review are likely to car ry over into subsequent time periods, that is the error terms become correlated over time. As a result of the violation of the assumption of independence of errors, the es timated variances will underestimate the true variance. Therefore, conventional formulas used in carrying out hypothesis tests and/or constructing confidence intervals ar e likely to lead to incorrect inferences. Furthermore, because the autocorrelation is likely to be positive, the calculated acceptance regions or confidence bounds may well be much narrower than the true regions or bounds. Hence one can erroneously conclude that a variable exerts a significant causal influence if th e model exhibits positive serial correlation. However, since the two data sets used in this study are annual survey with observations grouped by quarters, it is logical to expect that th ere might be some dependency between the quarters within a given year but not between quarters across years. In other words, there is statistical independence between years, but possible correlation between quarters in any given year. Though OLS re gression can be used to model the relationships under review, there was a need to employ a technique that would adju st the variance estimate to account for potential clustering within years. Test for Hypothesis 1-3
85 To test hypothesis 1-3, ordina ry least square (OLS) regre ssion model with clustering option that specifies that obser vations are independent across y ears but not necessarily within years (clusters) was used to examine the rela tionship between the depe ndent and independent variables. The null hypothesis of no relationship between the i ndependent (physician detailing, free drug sampling and off-label pub lication) and the dependent vari able off-label prescribing of anticonvulsants were tested. An a priori level of significance of 0.05 was used to evaluate significance of model estimates. RQ4: What patient characteristics, physicia n characteristics, physician office visit characteristics are associated with o ff-label prescribing of anticonvulsants? Patient Characteristics o Ha-4a: Controlling for other factors, drug visits by older patients more likely to be off-label visits o Ha-4b: Controlling for other factors, drug visi ts by whites are more likely to be off-label visits. o Ha-4c: Controlling for other factors, drug visits by males are more likely to be offlabel visits. o Ha-4d: Controlling for other factors, drug visi ts by patients with prescription drug coverage are more likely to be off-label visits. o Ha.4e: Controlling for other factors, drug visi ts by non-referred patients are more likely to be off-label visits. o Ha.4f: Controlling for other factors, drug visits by patients with chronic conditions are more likely to be off-label visits. Physician Characteristics o Ha-4g: Controlling for other factors, drug visi ts to physicians practicing in urban areas are more likely to be off-label visits. o Ha-4h: Controlling for other factors, drug visits to psychiatrists ar e more likely to be off-label visits. Visit Characteristics
86 o Ha-4i: Controlling for other factor s, drug visits that are noninitial visits are more likely to be off-label visits. o Ha-4q: Controlling for other factors, drug visits with increasing number of prescribed medications are more likely to be off-label visits. Test for hypothesis 4 The examination of the association be tween patient, physician and office visit characteristics with off-label anticonvulsant prescription was examined using logistic regression analysis. The complex survey commands in STATA was used in conducting this analysis, which takes into account both the stratific ation of the data in the dataset, visit weights assigned to each of the visits in the dataset, as well as the secondary clustering within physicians for physicians that reported more than one drug visits. Logistic Regression Logistic regression is a varia tion of ordinary regression and is used when the dependent (response) variable is dichotom ous (i.e. takes only two values, which usually represent the occurrence or non-occurrenc e of some outcome event, usuall y coded as 0 or 1). Logistic regression estimates the probability of a certain event occurring. The form of the model is; Log (p/1-p) = 0 + 1X1 + 2X2 +...+ KXK where p is the probability that Y=1 (off-label visit) X1, X2 Xk are the independent variables (p redictors of off-label visits) .0, 1, 2 k are known as the regression coefficients, which are estimated from the data. Logistic regression produces Odds Ratios (OR) associated with each predictor value. The odds of an event is defined as the probability of the event occurring divided by the probability of the event not occurring. The odds ratio is one set of odds (given presence or absence of an independent variable) divided by an other given the alternatives.
87 Reference cell coding of independent variables For the purpose of model building us ing logistic regression, it is nece ssary to convert som e of the independent variables into a coding scheme that uses reference coding method. Tables 5-7, 5-8, 5-9, 5-10, 1-11 and 5.12 lists the reference codi ng schemes used in analyzing the data. Variable selection For catego rized independent variables, a two way contingency was created reflecting the relationship between the dependent and the independent variable, and the chi-square statistic using weighted data and accounting for complex sampling design was computed to determine the significance of any association be tween the dependent variable and the categorical variable. Upon completing these univariate an alyses, the variables were listed according to p-values of the tests. The statistical criterion for including a va riable in the multivaria te model was set at pvalue 0.3. In addition, an effort was made to ascer tain that the variable(s) chosen in this manner appeared logically meaningful or that they make sense in the model. Variables that had p-value > 0.3 were also considered or added to the model if such variables were considered to be important variables that needed to be in the models. An exam ination was also carried out to ensure that none of the cate gories of any of the discrete variables had zero cells. For continuous scaled variables, the assumpti on of the linear relationship between the logit of the dependent variable and the continuous variable was examined. A plot of the logit of the dependent variable against the inde pendent continuous variable was used to assess if the linearity assumption was not violated. In addition, correlations betw een predictor variables were examined using Pearsons correlation. If the correlation between any two variables was 0.8, only one of the variables was added to the model and used in the subsequent multivariate building procedure.
88 Model building A predictive m odel was built to test the hypothe sis that the different independent variables are associated with off-label prescribing of anticonvulsant dr ugs. A multivariate logistic regression model, using visit weights and NAMCS study design variables with all predictors identified from the variable selection step were added to the model at on ce. Following the fit of the multivariate model, the importance of each variable included in the model was verified by comparing each estimated coefficient with the co efficient from the univariate model containing only that variable. Variables that did not contribute to the model were eliminated and a new model fitted. The estimated coefficients for the remaining variables were compared to those from the full model. This was done to evaluate potential large change s in coefficient estimates, which might indicate that one or more of the excluded variables were important in the sense of providing a needed adjustment of the effects of the variable remaining in the model. The process of deleting, refitting, and verifying was carried out until it appeared that all of the important variables were included in the model and those excluded were eith er logically or statistically unimportant.
89 Table 4-1. National Ambulatory Medical Care Survey (NAMCS) Sampling Frame Year No of Patient Records Estimated Patient visits (millions) No. of Physicians Sampled No. of Eligible Physicians* No. of Responding Physicians % Response 1993 35,978 717.2 3,400 2,464 1,802 73 1994 33,598 681.5 3,499 2,426 1,704 70 1995 36,875 697.1 3,724 2,587 1,883 73 1996 29,805 734.5 3,173 2,142 1,500 70 1997 24,715 787.4 2,498 1,801 1,247 69 1998 23,339 829.3 2,500 1,806 1,226 68 1999 20,760 756.7 2,499 1,728 1,087 63 2000 27,369 823.5 3,000 2,049 1,388 68 2001 24,281 880.5 2,744 1,910 1,230 64 2002 28,738 890.0 3,150 2,095 1,474 70 2003 25,288 906.8 3,000 2,007 1,407 67 2004 25,286 910.9 3,000 1,961 1,372 65 2005 25,665 963.6 3,000 1,936 1,191 62 Note; Data collected from public use files of the NAMCS dataset from 1993 through 2005. *Number of physicians meeting survey inclusion criteria Table 4-2. National Hospital Ambulatory Me dical Care Survey (NHAMCS) Sampling Frame Year No of Patient Records Estimated Patient visits (millions) No. of Hospitals Sampled No. of Eligible Hospitals* No. of Responding Hospital % Response 1993 28,357 62.5 489 445 419 94 1994 29,095 66.3 489 443 419 95 1995 28,393 67.2 487 437 409 94 1996 29,806 67.2 486 438 415 95 1997 30,107 77.0 486 434 416 96 1998 29,402 75.4 488 433 423 98 1999 29,487 84.6 489 427 405 95 2000 27,510 83.3 488 413 397 96 2001 33,567 83.7 479 407 391 97 2002 35,586 83.3 481 409 390 95 2003 34,492 94.6 546 462 428 92 2004 31,783 85.0 464 411 376 91 2005 29,975 90.4 458 402 367 90 Note: Data collected from public use file s of the NHAMCS dataset from 1993 through 2005. *Number of hospitals meeti ng survey inclusion criteria
90 Table 4-3. List of Older Anticonvulsants AED Therapeutic Class/code DEA Status Generic Codes Drug Name Drug codes Date Added Carbamazepine (CBZ) Anticonvulsant /1374 IV 50880 Carbamazepine Epitol Tegretol Carbatrol Tegretol XR 05680 11549 30730 98114 01063 No date No date No date 4/6/1999 8/12/2002 Divalproex Sodium (VPA) Anticonvulsant /1374 IV 51927 Depakote Sprinkle Depakote Divalproex Sodium 03012 08836 94081 4/25/2004 No date No date Ethosuximide (ESM) Anticonvulsant /1374 IV 52130 Ethosuxamide Zarontin 11898 35150 No date No date Methosuximide (MSM) Anticonvulsant /1374 53605 Celontin 06030 No date Phenobarbital (PB) 54395 Phenytoin (PHT) Anticonvulsant /1374 VI 54470 Dilantin Infatabs Phenytek Di Phen Dilantin Diphenylan Sodium Diphenylhydan toin Sodium Phenytoin Diphen 03013 03207 09195 09585 09885 09890 24045 93049 4/25/2004 9/8/2004 No date No date No date No date No date No date Primidone Anticonvulsant /1374 VI 54795 Mysoline Primidone Primoline Sterasoline 20135 25055 25060 29450 No date No date No date No date Valproic Acid (VPA) Anticonvulsant /1374 VI 56145 Depacon Depakote ER Depakene Deproic Valproic Acid Valproate 01264 02099 08835 08902 33573 93249 No date 5/19/2003 No date No date No date No date
91 Table 4-4. List of Newer Anticonvulsants Anticonvulsant Therapeutic class/code DEA Status Generic Codes Drug Name Drug codes Date Added Felbamate (FBM) Anticonvulsant /1374 VI 56415 Felbamate Felbatol 93221 93341 No date No date Fosphenytoin (FPHT) Anticonvulsant /1374 VI 59701 Cerebyx Phosphenytoin Fosphenytoin 97005 01023 97162 9/10/1997 7/28/2002 3/28/1998 Gabapentin (GBP) Anticonvulsant /1374 VI 57102 Gabapentin Neurontin 94114 94099 1/20/1995 1/12/1995 Lamotrigine (LTG) Anticonvulsant /1374 VI 57220 Lamotrigine Lamictal 97136 95181 3/8/1998 4/28/1996 Levetiracetam (LEV) Anticonvulsant /1374 VI 70160 Keppra Levetiracetam 00184 02037 5/29/2001 4/7/2003 Oxcarbazepine (OXC) Anticonvulsant /1374 VI 70128 Oxacarbazepine Trileptal 01216 00076 No date 3/30/2001 Pregabalin (PGB) Anticonvulsant /1374 VI 50000 Lyrica 00084 4 4/2/2001 Tiagabine (TGB) Anticonvulsant /1374 VI 59830 Tiagabine Gabitril 01319 98116 98116 8/22/2005 4/6/1999 Topiramate (TPM) Anticonvulsant /1374 VI 59744 Topiramate Topamax 98131 97049 4/26/1999 12/4/1997 Zonisamide (ZNS) Anticonvulsant /1374 VI 70225 Zoniszmide Zonegran 01275 01008 No date 7/25/2002 Definition of Variables: Drug code : A unique NCHS assigned, 5 digit code applied to each entry name mentioned in NAMCS and NHAMCS. Drug Name : The specific drug name identifier entered by the physician. It corres ponds with the entry made on any prescription or medical order. It may be either or trade/brand/proprietary or generic name, but not a drug class Generic code : A unique NCHS assigned, 5-digit code applied to each generic/nonproprietary/active ingredient name Generic Name: the assigned to every drug entity by the United States Pharmacopeia/United States Adopted or other responsible authorities. If the drug lis ted is a combination medication, this entry will read combination or fixed combination and will contain up to five active ingredients. DEA Status: Controlled medications, because of their significant potential for depe ndence or abuse and their pos sible diversion into illicit channels, are regulated under Federal Law by the Department of Justice, Drug Enforcement Agency (DEA). The controlled Substance Act of 1970 charact erizes each controlled drug into one of five schedules. Schedule I drugs, like heroin and LSD, have a higher potential for abuse and no current accepte d medical usefulness for treatment in the United States. Each successive schedule, II V, re flects a decreasing degree of dependence and potential for abuse. Date added : Date the drug was first mentioned in the NAMCS/NHAMCS datasets.
92 Table 4-5. Pharmaceutical Promotions Variable Format Total free drug sampling units for all AEDs Continuous Total free drug sampling units for old AEDs Continuous Total free drug sampling units for new AEDs Continuous Total free drug sampling units for Gabapentin Continuous Total free drug sampling units for other new AEDs Continuous Total physician detailing contacts for all AEDs Continuous Total physician detailing contacts for all old AEDs Continuous Total physician detailing contacts for all new AEDs Continuous Total physician detailing contac ts for Gabapentin Continuous Total physician detailing contacts for other new AEDs Continuous Table 4-6. Journal Publications Variable Format Cumulative number of publications de scribing successful off-label use of all new AEDs Continuous Cumulative number of publications describing successful off-label use of Gabapentin Continuous Cumulative number of publications describing succe ssful off-label use of other new AEDs Continuous Table 4-7. Patient, Physicia ns and Visit Variables Variable Description ITEM NAME Category Levels Year of Visit VYEAR Categorical a) 2001 b) 2002 c) 2003 d) 2004 e) 2005 Patients age AGE Continuous Age in years Patient Gender SEX Dichotomous a) Female b) Male Patient Race RACE Categorical a) White b) Black/African American c) Other Expected source of payment PAYTYPE Categorical a) Private Insurance b) Medicare c) Medicaid/SCHIP d) Workers Compensation e) Self Pay f) others Are you the patients primary care physician? PRIMCARE Dichotomous a) Yes b) No Was patient referred by another physician? REFER Dichotomous a) Yes b) No
93 Table 4-7 continued Variable Description ITEM NAME Category Levels Has the physician seen patient before? SENBEFOR Dichotomous a) Yes b) No How many past visits in the last 12 months? PASTVIS Categorical a) None b) 1-2 c) 3-5 d) 6 or more Do other physicians share patients care for this problem? SHARE Dichotomous a) Yes b) No Major reason for visit MAJOR Categorical a) Acute Problem b) Chronic Problem (routine) c) Chronic Problem (flare up) d) Pre-or post Surgery/injury follow up e) Non-illness care Patient reason(s) for visit a) reason 1 b) reason 2 c) reason 3 RFV1 RFV2 RFV3 Categorical a) Labeled Reasons b) Un-labeled Reasons Physician Diagnoses a) 1st Diagnosis b) 2nd Diagnosis c) 3rd Diagnosis DIAG1 DIAG2 DIAG3 Categorical a) Labeled Diagnosis b) Un-labeled Diagnosis Medication provided MED1 MED2 MED3 MED4 MED5 MED6 MED7 MED8 Categorical Number of medications coded NUMMED Continuous Additional drug characteristics for each medication coded a) Generic name code GEN1 GEN 2 GEN3 GEN4 GEN5 GEN6 GEN7 Categorical
94 Table 4-7 continued GEN8 a) Patient visit weight PATWT NA NA Geographic region of visit REGION Categorical a) Northwest a) Midwest b) South c) West Metropolitan Statistical Area (MSA) or nonMSA location of the visit MSA Dichotomous b) MSA c) Non-MSA Patient-physician linking code PHYCODE PATCODE NA NA Physician specialty SPEC Categorical a) Neurologist b) Psychiatrist c) General and family Practitioner d) Internal Medicine e) Others Type of setting for this visit RETYPOFF Categorical a) Private solo or Group Practice b) Free Standing Clinic c) Federally Qualified Health Center d) Mental health Center e) Non-Federal government Clinic f) Family Planning Clinic g) HMO or other prepaid practice h) Faculty Practice Plan i) Other Solo practice? SOLO Dichotomous a) Yes b) No Employment status of physician EMPSTAT Categorical a) Owner b) Employee c) Contractor Who owns this office? OWNS Categorical a) Physician or Physician group b) HM O c) Medical/Academic Health Center d) Other Health Care Corporation e) Other
95 Table 4-7. continued Age in days for patients less than one year AGEDAYS Continuous Single or multispecialty practice Dichotomous a) Single b) Multipartite Table 4-8. Independent Va riable Reference Cell Coding for NAMCS Variables Variable Variable Code Description Reference Sex Female If female, then 1 else 0 male Race WHITE If white, then 1, else 0 Non-Whites Age AGE Continuous NA Doctor Specialty DOCTYPE2 If Psychiatri st then 1, else 0 General Practitioner Doctor Specialty DOCTYPE3 If Neurologist then 1, else 0 General Practitioner Doctor Specialty DOCTYPE4 If Internal Medicine, then 1, else 0 General Practitioner Doctor Specialty DOCTYPE5 If others, then 1, else 0 General Practitioner Primary Care Physician Pri_Prov If Primary care provider then 1, else 0 Non-Primary Care provider Established Patient Est_Patient If established patient then 1,else 0 New Patient Primary Reason for Visit REA_2 If Psychiatry related, then 1, else 0 Neurological Reason Primary Reason for Visit REA_3 If others, then 1, else 0 Neurological Reason Diagnosis Dia2 If Psychiatry only, then 1, else 0 Neurological Diagnosis Diagnosis Dia3 If Pain only then 1, else 0 Neurological Diagnosis Diagnosis Dia4 If Neurology + Psychiatry, then 1, else 0 Neurological Diagnosis Diagnosis Dia5 If Neurology + Pain, then 1, else 0 Neurological Diagnosis Diagnosis Dia6 If Psychiatry + Pain, then 1, else 0 Neurological Diagnosis Diagnosis Dia7 If Neurology + Psychiatry + Pain, then 1, else 0 Neurological Diagnosis Major Reason for Visit CHRONIC If Chronic Condition then 1, else 0 Acute Condition Drug Type DRUGTYPE2 If GBP, then 1, else 0 Old AEDs Drug Type DRUGTYPE2 If Other New AEDs, then 1, else 0 Old AEDs Number of Medications NUMMED Continuous NA
96 Table 4-8. continued Drug Coverage INSUR If(Medicaid or Workers Comp then 1,else 0 Non-Medicaid Insurance Number of Visit in last 12 months PAST_V If >5 visits, then 1, else 0 Less than 6 Visits Geographical Region REGION2 If Mi dwest, then 1,else 0 Northeast Geographical Region REGION3 If South, then 1, else 0 Northeast Geographical Region REGION4 If We st, then 1, else 0 Northeast Metropolitan Statistical Area URBAN If MSA then 1, else 0 Non-MSA NAMCS Year Y02 If Year =2002, then 1, else 0 2001 NAMCS Year Y03 If Year =2003, then 1 else 0 2001 NAMCS Year Y04 If Year =2004, then 1, else 0 2001 NAMCS Year Y05 If Year =2005, then 1, else 0 2001
97 CHAPTER 5 RESULTS The results of this study are pr esented in this chapter. The first section describes antiepileptic drug (AED) visits reported in the NAMCS and NHAMCS, and also describes the trends in off-label AED visits. The second sectio n presents the trends in promotional activities (physician detailing contacts and free drug sampli ng), trends in publication of off-label label studies for the anticonvulsant drugs of interest and also presents the results of the time series analysis to determine the association between the variations in trends of off-label visits and trends in these variables. The last section pres ents the results of the regression model fitted to evaluate possible patient and physician characteristic s that might predict of off-label drug visits. National Ambulatory Medical Ca re Survey (NAMCS) Drug Visits A total of 361,697 physician office visits was re ported in the NAMCS dat aset from 1993 to 2005. This represented a sample from 10.6 billion physician office visits in the United States during this period. At about 65% (6.9 billion) of these visits, patient s were prescribed or provided with at least one medication (medication visits). The fraction of all NAMCS visits that were medications visits were similar across th e years during the study period and ranged from 62% to 71%. At about 1.7% (140.3 million) of all medication visits, patients were prescribed at least one of the studys AEDs (range: 1.0% to 3.3%). The numb er of AED visits increased during the period under review, but the proportion of all medication visits that were AED visits did not change significantly across the years. AE D visit patients tended to be older than all medication visit patients (49 vs. 44, P<0.0001), they also tended to have more drugs mentioned than medication visit patients (3.8 vs. 2.3, P< 0.0001). Tables 5-1, 5-2, and 5-3 provide annual descriptions of NAMCS AEDs visits. The onl y significant trend observe d across the years in
98 AED visits characteristics was an increase in the average number of medicat ions per AED visits. No other significant trend was obs erved in the characte ristics of AED visits across the years. There were no drug visits attrib utable to new AEDs reported at the beginning of the study period. However, by 1999, the new AED visits accounted for about 0.5% of all medication visits and more than 20% of all AEDs visits, and by the end of the study period, they accounted for about 2% of all medication visits and more than 70% of all AED visits. No longitudinal change was observed in the proportion of all medication visits attributab le to the old AEDs, but there was a downward trend in the proportion of all AED vis its attributable to old AEDs due to the fact that the new AEDs gained market share. Figures 5-1 and 5-2 show the trends in AED visits over time. Overall, the old AEDs accounted for about 59% (83 million) of all AED visits while the new AEDs accounted for about 41% (57.3 million) of all AED visits. The top four anticonvulsant drugs most freque ntly associated with AED visits were: Gabapentin (27%), Divalproex (19%), Carbamazep ine (15%), and Phenytoin (14%). (Figure 5-3) National Hospital Ambulatory Medical Care Survey (NHAMCS) Dr ug Visits A total of 397,560 hospital outpatient departme nt (OPD) visits were reported in the NHAMCS dataset between 1993 and 2005. This represented a total of about 1.02 billion hospital OPD visits. At about 66% (655 million) of these visits, at least one medication was prescribed or provided to patient s (medication visits). At l east one of the studys AEDs was prescribed or provided at about 3% (21 million) of these drug visits. Tables 5-4, 5-5 and 5-6 provide a descriptive summary of various characteristics of NHAMCS AED visits. No significant trend was observed in the characteristics of the AED visits across the years. As in the NAMCS dataset, there were no new anticonvulsant medication or AED visits at the beginning of the study period (Figure 5-4), however by 1998, they accounted for more than 30% of all AED visits and up to 60% by the end of study period. The proportion of all AED medication visits
99 rose slightly and peaked in 1999, after which there was a gradual downward trend. As with NAMCS visits, there was a downwar d trend in the proportion of a ll AED visits attributable to old AEDs over time (Figure 5-5). Sixty-five pe rcent (13.7 million) of al l AED visits were old AED visits, and 35% (7.4 million) of all AED visits were new AED visits. The top four most frequently mentioned anti convulsant drugs were : Gabapentin (24%), Divalproex (18%), Phenytoin (18%), and Carbamazepine (15%), (Figure 5-6) National Ambulatory Medical Care Survey ( NAMCS) Off-Label An ticonvulsant Drug Visits Identification of off-Label Visits A total of 158 AED visits had no recorded reas on for visit or diagnos is codes or had only ICD-9 codes that reflected enc ounters for adm inistrative purpo ses, leaving 7,481 AED visits. This represented a total of about 136 million physici ans office visits. After identification of onlabel visits and off-label visits, 635 visits could not be categorized as on or off-label visits and were excluded from the study. Examination of o ff-label trends were carried out on the remaining 6,846 observations (118.1 million AED visits). Ther e was no difference in the proportion of old and new AED visits between the original datasets (before study exclusions) and the proportion of old and new AED visits in the dataset used for th e trend analysis after ex clusion of visits that could not be categorized as on or off-label vi sits (old AEDs =59%, 70 million and new AEDs =41%, 48 million.) Trends in All Off-Label Anticonvulsant Drugs Visits Trends in off-label AED visits ranged from 51% in 1993 of all AED visits to 67% in 2005 (Figure 5-7). The overall propor tion of off-label AED visits was 65% (95% CI: 63%-67%) during the study period, which is an estimated 76 million off label visits nationwide. There was a significant linear growth in the proportion of all medication (0.04% per quarter, 95% CI: 0.03-
100 0.05) and all AED visits ( =0.7% per quarter, 95% CI: 0.3-1.0) that were off-label AED visits during the study period. Compared to on-label drug visits, off-label visits were more likely attributable to female visits. (Table 5-7). Off-label visit patients we re also older and had more drugs mentioned than on-label visits patients. Neurol ogists were more frequently responsible for off-label visits. New AEDs were more frequently associated with off-label visits than old AEDs (P<0.0001). Trends in Off-Label Old Anti convulsant Drugs Visits A significant downward trend in the proportion of off-label A ED visits attributable to old AEDs ( = -0.62% per quarter, 95% CI: -0.91-0 .33) was observed during the period under observation (Figure 5-8), howev er, there was no significant ch ange in the proportion of medication visits that were o ff-label old AED visits. At th e beginning of the period under observation, about 51% of all offlabel AED visits (all off-label visits) were old AEDs off-label visits. This proportion decrease d to about 31% at the end of the study period. Overall, about 29% of all AED visits and 49% of all old AED visits were off-label old AED visits. No longitudinal change was observed in the proportion of all old AED visits that were off-label visits during the study period (Figure 5-9). Trends in Off-Label New Anti convulsant Drugs Visits A significant upward trend in the proportion of all off-label A ED visits and all off-label medication visits attributable to new AEDs was observed ( =1.31% per quarter, 95% CI: 1.031.58) during the period under review (Figure 5.8). The new AEDs accounted for 0% of all offlabel AED visits at the beginning of the study pe riod, the proportion of of f-label new AED visits peaked in 2003 (56%) and was about 43% at th e end of the study period (Figure 5-9). About 35% of all AED visits and 87% of all new AED visits were offlabel new AED visits.
101 Off-Label Visits by Drug Types Gabapentin, Divalproex and Carbamazepine were the ind ividual drugs most frequently associated with off-label visits. 94% of all Gabapentin visits were off-label visits (Figure 5-11) and about 40% of all off-label AED visits were Gabapentin visits. (Fig ure 5-10). Phenytoin and all other old AEDs (ESM, MSM PB, PRM and VP A) combined as a group had more on-label visits than off-label visits (Figure 5-11). Effect of Introduction of Ne w Anticonvulsant Drugs on the Trend in All Medication Visits that are Off-Label AED Visits Since the trends in off-label visit is a composite of the trend in off-label old AED visits and off-label new AED visits, an ordinary least square regression line (controlling for potential clustering between quarters within a year) wa s fitted on the quarterly differences in the proportion of all medication visits that were off-label new AED vi sits and all medication visits that were off-label old anticonvulsant visits. The intercept at the beginni ng of the series was negative indicating that at the beginning of the period under revi ew, the proportion of off-label old AED visits was higher than the proportion of off-label new AED visits. The slope of the fitted regression line was significantly greater than zero ( =0.0003% per quarter, 95% CI: 0.0002-0.0004) indicating that the observed increase in the proporti on of off-label AED visits over time was attributable to the increase in off-label new AED visits. Conditions Associated with Off-Label Visits The conditions for which AEDs were prescribed off-label were identified and grouped into 4 groups: N eurological Conditions, Psychological Condition, Conditions associated with Pain, and Other Conditions (conditions th at do not fit into any of thr ee previously listed groups). Table 5-8 shows the proportion of off-label visits associated with the four condition categories.
102 Off-label visits were mostly classified off-label for psychological conditions. Conditions categorized as others contributed less than 0.5% of the total off-label indications. National Hospital Ambulatory Medical Care Survey (NHAMCS) Off-Label Anticonvulsant Drug Visits Identification of Off-Label Visits A total of 695 AED visits had no recorded di agnosis codes or had only ICD-9 codes that reflected encounters for adm inistrative purposes leaving a total of 9,749 AED visits. This represented a total of about 20.5 million anticonvulsant OPD visits. After identification of onlabel visits and off-label visits 949 visits could not be categori zed as on or off-label visits. These visits were further excluded from the final population of drug visits. Therefore examination of off-label trends was carried out on the remaining 8800 observations (16.5 million) after all exclusions were carried out. There was no significant difference in the proportion of AED visits that were old or new AED visits between the original dataset (before all study exclusion) and the dataset us ed for the trend analysis (after all study exclusion), (old AEDs =65%,10.7 million and new AEDs = 35%, 5.8 million). Trends in Off-Label Anti convulsant Drug Visits A signif icant linear growth in th e proportion of all medication ( =0.03% per quarter, 95% CI: 0.03-0.04) and all AED ( =0.62% per quarter, 95% CI: 0.45-0.71) visits that were off-label AED OPD visits was observed during the 13 years studied. Off-label visits ranged from 18% (lowest value) in the first quart er of 1993, to 78% (highest value) in the fourth quarter of 2000 (Figure 5-12). The overall proport ion of off-label visits was 58% (95 CI: 55%-61%), which is an estimated 9.6 million OPD AED visits. There were significant differences observed in the characteristics of on-label and off-label visits. Off-label visits patients were older, mostly likely to be females, and on the average were
103 prescribed more medications than on-label visit patients. Ne w AED visits more frequently associated with off-label visits than ol d AED visits (P<0.0001). (Table 5-10) Trends in Off-Label Old Anticonvulsant Drugs Visits There was no significant change in the proportion of all m edi cation visits that were offlabel old AED visits, however there was a downward trend in the proportion of all anticonvulsants visits that were off-label old AED visits ( =-0.7% per quarter, 95% CI: -0.80.5), (Figure 5-13). About 28% of all AED visits were off-label old AED visits and about 44% of all old AED visits were off-label visits. Th e quarterly proportion of a ll old AED visits that were classified as off-label visits did not change significantly during the period under observation (Figure 5-14) Trends in Off-Label New An ticonvulsants Drugs Visits A significant upward trend in the proportions of all m edication visits that were off-label new anticonvulsants ( =1.3% per quarter, 95% CI 1.1-1.5) and in the proportions of all anticonvulsants visits that were new AED visits (P<0.0001) was observed (Figure 5.13). Thirty percent of all AED visits were off-label new AE D visits and 84% of all new AED visits were off-label visits. The proportion of new AED vis its that were classified as off-label visits remained above 60% during the en tire study period (Figure 5.14). OffLabel Visits by Drug Types In the NHAMCS dataset, Gabapentin visits were m ost frequently associated with off-label visits. About 96% of all Gabapen tin visits were off-label visits (Figure 5-15) and 39% of all offlabel visits were Gabapentin visits (Figure 5-16).
104 Effect of Introduction of Ne w Anticonvulsant Drugs on Trend in All Medication Visits that are Off-Label Visits The ordinary least square regression model fitted on the quarterly differences in the proportion of all medication visits that were off-label new AED vi sits and all medication visits that were off-label old anticonvulsant visits to evaluate the effects of the new anticonvulsants on trends in proportion of NHAMCS off-label AED visits had a pos itive linear slope ( =0.0004% per quarter, 95% CI: 0.03-0.05), wh ich indicates that on the aver age the difference between the proportion of all medication visits that are new and old off-la bel AED visits increased by 0.3% each quarter.. Conditions Associated with Off-Label Visits Table 5-11 shows the proportion of off-label visits associated w ith the three m ajor condition categories in the NHAMCS datasets. Ju st as in the NAMCS, off-label visits were mostly classified off-label for psychological conditions. Effects of Pharmaceutical Promotiona l Activities on Off-L abel Visits Trends in Physician Detailing Contacts The trends in physician detailing contact s m ade for AEDs between 1994 and 2005 are shown in Figure 5-17. In general, there was an increase in the total number of AED detailing contacts. The increase in detai ling contacts started in the first quarter of 2000 and prior to this quarter, physician detailing contacts for both ol d and new AEDs remained relatively constant. The rise in detailing contacts observed was due to the increasing number of new AED detailing contacts, as no significant longi tudinal change was observed in the number of old AED detailing visits. Sixty-eight percent of all AED detaili ng contacts during the pe riod under review were new AED detailing contacts (17% for GBP and 51% for all other new AEDs, Figure 5-18). Most of the new AED contacts occurred after 1999, as only 20% of all new AED detailing visits
105 occurred prior to the year 2000. In addition, prior to 2000, new AED contacts accounted for 46% of all AED contacts. However, from 2000 to 2005, 76% of all AED detailing contacts were attributable to new AEDs. Figur es 5-19, 5-20 and 5-21, show the tr ends in AED contract visits by drug types. For the old AEDs, divalproex was th e only one that showed a significant increase in detailing visits during the study period. For ga bapentin, a sharp increase in detailing visits was observed in the 3rd quarter of 2002, this in crease peaked in the 3rd quarter of 2004 but declined to virtually zero at th e end of the study period. Detailing visits for other new AEDs showed no longitudinal change up until 2000, after which there was a sustained increase in detailing visits. Figure 5-22 shows the trends in detailing visit by sett ing (physician office visits or hospital visits). Trend in detailing visi ts to office-based physicians and hospital based physicians were generally similar during the study period, except that the increase obs erved in office-based physician detailing visits was higher than th e increase observed in hospital-based physician detailing visits. Association between Trends in Physician Detailing Contacts and Off-Label Visits Since longitudinal change obser ved in off-label visits was mostly a ttributable to new AEDs, only the effects of promotional activities of the new AEDs on off-label use of new AEDs were examined. The results of the time series analyses carried out to determine if the trends in physician detailing contacts is a ssociated with the trend in offlabel AEDs visits are shown in Table 5-13. The regression coeffi cient indicates that on the av erage, a 10% increase in the number of all AED detailing cont acts made to office based phys icians was associated with a 0.008% increase in the proportion of all physician o ffice medication visits that were off-label new AED visits, while every 10% increase in detailing contacts made to hospital based physicians was associated with a 0.026% increase in th e proportion of all OPD medication visits
106 that were off-label new AED vis its. Physician office detailing contacts had the most effect on off-label visits for gabapentin, while hospital de tailing contacts had the mo st effect on off-label visits for other new AEDs (minus gabapentin). Only the association between OPD detailing contacts for all other new antic onvulsants (minus gabapentin) a nd hospital OPD off-label visits for these other new anticonvulsants reached st atistical significance, although the association between hospital OPD detailing visits for gabapentin and off-label gabapentin hospital visits tended towards significance. Trends in Free Drug Sampling Only data on free drug sam pling for office base d physicians were available, therefore only examination of association between trends in new free drug sampling for new AEDs and trends in off-label physician office visits were conducte d. There was no significant longitudinal change observed in free old anticonvulsant drug sampling units distributed to office based physicians. The number of old AED samples distributed during the study period ranged from 1,303,000 to 7,192,000 per quarter (Figure 5-23). An increas e in the number of new AED sampling units distributed to office base physicia ns was observed in 2000, and was ma intained for the rest of the period under observation. New AED sampling units distributed in 2005 was about 600% more than sampling units distributed in 1999. The number f new AED sampling units distributed during the study period ranged from 1,151,000 to 31,999,000. About 73% of all distributed AED sampling units were new AEDs, with about 17% of all distributed AEDs sampling units reported to be gabapentin samples. Association between Trends in Free Drug Sampling and Off-Label Visits The results of the tim e series analyses carried out to determine if there is an association between trends in free drug sampling to physicians and trends in off-label drug visits are shown in Table 5-14. Total free drug sampling for all new anticonvulsants as a whole, for gabapentin
107 and other new anticonvulsants separately were a ll significantly and positively associated with off-label all new AEDs visits, ga bapentin visits and other new AE Ds visits respectively. A 10% increase in the number of all new AED free sample s distributed per quarter was associated with a 0.0003% increase in the proportion of all office medication visits that were off-label visits. Publications Concerning Off-Label Use of Anticonvulsants The rela tionship between the trends in the cu mulative number of controlled trials (CTs) and case report/case series publications repor ting positive results for the use of the new anticonvulsants in off-label indications and off-label visits attributable to the new anticonvulsants in both office and hospital se ttings between 1993 through 2005 are shown in figures 5-24, and 5-25. In general, gabapentin had fewer numbers of CT publications and case reports than all the other anticonvulsant drugs as a whole. Individually gabapentin had the highest number of RCTs and case reports with la motrigine and topiramate having the second and third highest number of both RCTs and case reports publications. Trends in cumulative number of case reports were consistently associated with trends in off-label new anticonvulsant drug visits in both hospital and office settings (Table 5-15), while trends in cumulative number controlled trials appeared not to be associated with the trends in offlabel new anticonvulsant drugs visits. A 10% increase in the cumulative number of case report/case series publications for all new AEDs was associated with a 0.042 % per quarter and 0.051% per quarter increase in the proportion of all medication visits that were off-label visits in physician office and hospital OPD settings respec tively. All observed association between case reports/case series and off-label visits reached statistical significance at the 0.05 level of significance. With CT publica tions, a 10% increase in th e cumulative number of CT publications for all new AEDs was associated with about a 0.001% per quarter decrease and 0.013% per quarter increase in the proportion of all medication visits that were off-label visits in
108 physician office and hospital OPD settings respec tively though none of the association between CT publications and off-label visits reached statistical significance. Predictors of Off-Label AEDs Visits The NAMCS dataset from 2001 to 2005 contains reports on 129,258 physician office visits, which represents about 4.6 billion am bulatory care visits na tionwide. At 2% (4,216) of these visits, at least one of the studys AEDs was prescribed and this represents an estimate of about 81 million AED visits nationwide. Only 3,884 of these visits were included in the study after exclusion of all visits with no primary dia gnosis or with administrative codes in the primary diagnosis field and visits that co uld not be classified as on-label or off-label. The rest of the analysis was carried out on this population of patients. A total of 2315 AEDs visits were identified as off-label visits. There were differences in off-label proportions observed between the 2001-2005 subpopulation identified to evaluate the predic tors of off-label visi ts and the overall study population used in conducting off-label trend analys is. The overall proportion of off-label visits was significantly higher in the 2001-2005 subpopul ation (71%, 95% CI 68-73 vs. 65%, 95% CI 63-67). In addition, a lower proporti on of AED visits were attributab le to off-label old AEDs in the 2001-2005 subpopulation (22% 95% CI: 20-25, vs. 29 95% CI: 28-32), while a significantly higher proportion of AED visits were attributab le to off-label new AE Ds (48%, 95% CI; 45-51 vs. 35%, 95% CI: 33-38). Table 5-16 gives a genera l description of visits that were included in this section of the study The result of the logistic regression model evaluating the predic tors of off-label visits are shown in Table 5-17. The probability of receivi ng an off-label prescription was significantly higher for gabapentin and other new anticonvulsant drug visits than for old anticonvulsant drug visits. Compared to family practice practi tioners, there was a doubling in risk of off-label
109 prescribing of anticonvulsant drugs with physicians with specialties classified as others. In addition, patients with diagnosis related to pain only, neurology plus pain psychology plus pain, and neurology plus psychology plus pain all had more than a 50% reducti on in the risk of receiving anticonvulsant drugs for off-label indications when compared to patients with a diagnosis for neurological conditions only. Increasing age and numb er of medications prescribed at a visit were also positively associ ated with an increase in the odds of off-label visits. There was a 1% increase and a 22% increase in the risk of off-labe l visits with each unit increase in age (in years) and number of medications prescribed at drug visits respectively. Finally, compared to visits from 2001, visits from all the other four years included in the study (2002, 2003, 2004 and 2005) were less likely to be offlabel visits; with visits in 2005 having the least likelihood of off-labe l prescribing of AEDs. Sensitivity Analyses The results of the sensitivity analyses conducte d to determ ine the eff ects of the exclusion of visits that could not be categ orized as on-label or off-label visits on the observed associations between off-label prescribing and study covariat es, showed that some of the associations observed were sensitive to the inclusion of these visits. Sensitivity analysis with uncategorized visits classified as off-label visits The associations between study covariates and of f-label prescribing in the original analysis (exclusion of uncategorized visits) and the new anal ysis with the unca tegorized visits classified as off-label visits were very similar. There was no reversal in the direction of the observed associations between off-label prescribing and any of the examined covariates. The only difference noted was that two associations: be tween visits to neurol ogists and off-label prescribing and between visits to psychiatrists and off-label prescribing reached statistical significance in the new analysis.
110 Sensitivity analysis with uncategorized visits classified as on-label visits There were notable differences observed in the re sults obtained from the two analyses: original analysis (excluding uncategorized visits ) and new analysis with uncategorized visits classified as on-label visits. There was a revers al of direction in five associations. These associations were between off-labe l prescribing and: vi sits to primary care pr oviders, visits with only pain related diagnosis, visits with both neurologic and pain related diagnosis, visits made to neurologist and visits made to psychiatrist. All reversed a ssociations were from negative association to positive associations, with two of the reversed associations reaching statistical significance (visit with only pain related diagnosis and visits with both neurological and pain related diagnosis).
111 Table 5.1. Characteristics of All Anticonvulsants Drug Visits between 1993 and 2005 (NAMCS) Variables 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 Med. Visits X108 4.9 4.4 4.5 4.7 5 5.4 5 5.5 5.5 5.8 6 5.9 6.7 All AED Visits X106 6.2 4.3 5.2 5.1 7.0 10.2 11 11.5 12.8 18.2 18.2 18.8 22.5 Drug Category (Old AED, % of all AEDs) 100 97 99 91 85 81 74 67 52 57 38 40 38 Sex (Female, %) 53 55 54 59 58 60 57 61 64 55 58 62 62 Race 88 79 88 83 88 89 87 85 95 91 90 87 88 Av. Age (SD) 47.9 (2.0) 49.4 (1.7) 46.1 (1.8) 49.3 (1.8) 46.7 (1.9) 51.3 (1.7) 48.6 (1.2) 47.7 (1.4) 48.4 (1.5) 46.0 (1.4) 46.7 (1.7) 49.7 (1.1) 51.9 (1.1) Region (%): Northeast Midwest South West 30 21 23 26 22 19 33 27 15 24 34 26 16 19 42 22 24 20 35 21 22 18 31 29 28 13 29 30 24 23 31 22 23 17 36 24 21 18 40 22 27 19 37 17 15 22 37 26 19 25 40 16 Av # of Meds/Visit (SD) 2.9 (0.1) 3 (0.1) 3 (0.1) 3.1 (0.2) 3.2 (0.2) 3.4 (0.2) 3.5 (0.1) 3.4 (0.1) 3.2 (0.1) 3.6 (0.1) 4.1 (0.2) 4.5 (0.2) 5 (0.2) Primary Reason for Visit Categories (%): Neurologic Psychiatry 24 20 25 14 27 14 27 18 17 20 22 16 18 23 16 22 18 21 16 18 21 18 16 21 12 12 Diagnosis Category Neurology Psychiatry Pain 81 32 45 69 28 41 75 26 47 80 38 52 72 40 35 71 32 33 68 37 32 79 40 36 76 42 34 79 40 35 78 40 29 78 39 30 61 29 34 Physician Specialty (%) Neurologist Psychiatrist Family Med Internal Med Others 21 19 20 17 19 14 25 22 16 26 25 24 14 25 15 29 20 32 17 28 19 29 16 27 14 20
112 Table 5-2. Characteristics of Old Anticonvulsant drug Visits between 1993 and 2005 (NAMCS) Variables 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 Drug Types (%) CBZ DS PHT Others 31 13 35 21 33 16 27 24 35 15 33 17 31 22 26 21 26 37 21 16 32 35 22 11 28 36 22 14 24 40 23 13 28 44 17 11 20 40 24 16 19 43 20 18 16 42 25 17 19 43 21 17 Sex (Female, %) 53 54 53 57 57 56 55 57 56 49 53 52 50 Race (%) Whites Blacks Others 88 10 2 79 18 3 88 7 4 82 7 4 87 11 2 89 8 3 88 10 3 84 13 3 95 3 2 89 9 11 90 8 2 86 10 3 87 10 4 Av. Age (SD) 47.9 (2) 49.2 (1.8) 46 (1.8) 48.3 (1.9) 44.7 (1.8) 50.7 (2.1) 47.5 (1.5) 45.4 (1.6) 45.6 (2.1) 43.5 (1.7) 42.2 (2.8) 48.7 (2.2) 50.1 (1.4) Region (%): Northeast Midwest South West 30 21 23 26 21 19 33 28 15 24 34 26 16 21 41 22 23 19 36 22 18 20 30 32 29 35 28 28 26 24 28 22 24 20 29 27 23 16 41 19 24 22 36 18 14 22 31 32 18 27 37 18 Av # of Meds/Visit (SD) 2.9 (0.1) 3.1 (0.1) 3 (0.1) 3.2 (0.2) 3.1 (0.1) 3.4 (0.2) 3.4 (0.2) 3.2 (0.2) 3.1 (0.2) 3.5 (0.1) 4.2 (0.3) 4.7 (0.4) 4.8 (0.2) Primary Reason for Visit Categories (%): Neurologic Psychiatry Others 24 20 53 25 15 59 27 14 57 28 20 50 19 23 54 24 16 59 20 23 56 17 22 58 20 21 58 16 20 64 302 34 55 19 19 59 13 13 71 Diagnosis Category (%) Neurology Psychiatry Pain 81 31 45 69 29 39 75 26 47 82 40 50 75 42 32 75 33 32 73 39 30 67 40 32 83 48 36 77 44 33 79 34 35 82 38 36 65 36 33 Physician Specialty (%) Neurologist Psychiatrist Family Med Internal Med Others 21 19 22 21 21 19 18 17 26 21 18 14 32 17 19 23 24 25 9 19 18 30 23 12 17 22 25 20 19 15 13 27 22 25 13 15 29 16 21 19 21 32 15 12 20 17 28 25 17 13 28 22 34 21 15 17 20 16 12 35 14 23 23 19 20
113 Table 5-3. Characteristics of New Anticonvuls ant Drugs Visit between 1993 and 2005 (NAMCS) Variables 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 Drug Types (%) GBP Others 0 100 5 95 46 64 66 34 72 28 77 23 80 20 74 26 68 32 61 39 55 45 51 49 54 46 Sex (Female, %) 64 56 35 73 62 74 65 66 73 60 61 67 69 Race (%) Whites Blacks Others 100 0 0 91 0 9 97 0 3 95 1 4 90 10 0 91 8 1 84 13 2 89 8 3 95 4 2 94 4 2 91 7 2 88 10 3 87 11 2 Av. Age (SD) 36.1 (4.8) 49.6 (5) 43 (3.3) 57.5 (3.4) 55.1 (3.2) 53 (2.4) 50.9 (2) 50.8 (2.1) 50.4 (1.6) 48.1 (1.5) 48.7 (1.5) 50.1 (1) 52.5 (1.4) Region (%): Northeast Midwest South West 0 88 0 12 39 24 23 14 26 31 33 10 17 23 41 19 34 25 27 14 33 17 34 16 24 8 36 32 20 10 39 21 24 13 44 19 19 21 37 23 28 18 36 17 15 22 40 22 19 25 42 14 Av # of Meds/Visit (SD) 2.8 (0.1) 2.7 (0.2) 3.5 (0.5) 2.6 (0.3) 3.6 (0.4) 3.4 (0.3) 3.7 (0.2) 3.7 (0.2) 3.3 (0.2) 3.8 (0.2) 4.2 (0.2) 4.4 (0.2) 5 (0.2) Primary Reason for Visit Categories (%): Neurologic Psychiatry Others 85 15 0 59 0 41 51 4 45 32 4 63 12 2 85 20 16 63 16 24 60 17 20 57 16 21 62 20 15 64 19 20 60 15 21 63 13 12 73 Diagnosis Category (%) Neurology Psychiatry Pain 100 0 85 88 3 80 90 12 53 69 9 66 59 20 53 67 22 39 68 34 36 70 40 44 75 38 30 77 35 37 80 43 26 75 40 28 60 26 35 Physician Specialty (%) Neurologist Psychiatrist Family Med Internal Med Others 100 0 0 0 0 76 0 24 0 0 53 5 42 0 0 43 3 34 12 8 14 5 51 7 22 40 18 15 12 16 19 24 15 18 29 17 25 14 19 25 22 33 11 19 16 21 26 20 13 20 17 32 17 11 21 17 30 20 13 20 15 18 18 20 28
114 Table 5-4. Characteristics of All Anticonvulsants Drug Visits between 1993 and 2005 (NHAMCS) Variables 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 Med. Visits X108 0.4 0.4 0.4 0.4 0.5 0.5 0.6 0.5 0.54 0.5 0.6 0.6 0.7 All AED Visits X106 0.6 0.8 0.9 1.1 1.2 1.2 1.7 1.6 2.0 1.9 2,7 2.2 3.1 Drug Category (Old AED, % of all AEDS) 100 99 98 97 94 84 82 69 58 56 49 49 40 Sex (Female, %) 56 59 44 50 54 48 48 59 56 60 60 60 59 Race (%) Whites Blacks Others 74 24 2 74 25 0 79 19 2 72 26 2 68 31 2 72 25 8 80 18 2 72 24 4 74 21 5 69 29 2 74 20 6 76 20 5 70 27 3 Av. Age (SD) 34.4 (2.4) 39.7 (2.9) 40.2 (2.7) 41 (2.5) 40.3 (3) 43.7 (1.5) 41 (2.4) 41.5 (2.6) 42.3 (2.7) 43.5 (1.7) 37.8 (3.2) 44.5 (1.3) 44.5 (1.4) Region (%): Northeast Midwest South West 41 23 21 14 35 26 30 9 35 27 21 16 24 50 14 12 40 21 30 10 34 28 24 15 43 19 28 11 45 20 26 10 35 18 31 16 30 24 33 12 35 21 36 9 43 22 29 11 31 35 24 10 Av # of Meds/Visit (SD) 2.8 (0.2) 2.5 (0.1) 3.2 (0.2) 3.2 (0.2) 3 (0.2) 3.3 (0.2) 3.7 (0.2) 3.7 (0.2) 3.6 (0.1) 3.9 (0.2) 4.2 (0.3) 4/6 (0.1) 4.7 (0.2) Primary Reason for Visit Categories (%): Neurologic Psychiatry Others 29 5 62 18 9 70 19 12 68 14 5 79 16 9 73 21 19 65 16 13 69 13 11 74 17 10 71 14 8 76 23 10 66 14 15 70 13 11 76 Diagnosis Category (%) Neurology Psychiatry Pain 74 25 55 66 25 52 68 28 38 63 24 26 74 33 45 65 33 45 62 29 40 63 30 41 70 35 33 63 25 40 66 28 37 66 37 32 60 29 31 Physician Specialty (%) General Med Surgery Pediatrics Others 48 25 14 14 44 16 17 31 54 11 10 25 48 15 13 23 47 9 7 38 53 6 6 35 42 21 8 33 50 4 10 37 41 8 5 46 53 4 8 36 36 7 16 42 48 7 5 41 31 9 5 34
115 Table 5-5. Characteristics of Old Anticonvulsant drug Visits between 1993 and 2005 (NHAMCS) Variables 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 Drug Types (%) CBZ DS PHT Others 30 7 35 28 30 11 35 24 37 8 32 23 25 25 30 20 27 28 30 15 20 28 32 20 22 29 33 16 27 29 29 15 23 39 20 18 17 34 25 24 22 33 26 19 19 36 24 21 21 38 25 16 Sex (Female, %) 56 58 45 51 54 48 48 53 55 51 51 54 45 Race (%) Whites Blacks Others 74 24 2 73 26 1 79 19 2 71 27 2 67 31 2 73 26 2 80 18 2 69 27 4 71 23 6 70 29 1 68 25 7 70 25 5 65 32 3 Av. Age (SD) 34.4 (2.4) 40 (2.9) 40.5 (2.7) 40.7 (2.5) 40.1 (2.8) 42.6 (1.7) 39.6 (2.8) 37.9 (3.1) 39.3 (3) 40 (2.7) 34.9 (3.1) 40 (1.7) 40.1 (2.2) Region (%): Northeast Midwest South West 42 23 21 14 35 26 30 10 36 27 20 17 24 51 12 13 40 21 30 9 30 30 24 16 44 19 27 10 43 22 25 10 33 18 34 15 26 28 34 12 30 18 43 9 53 20 17 10 37 31 24 8 Av # of Meds/Visit (SD) 2.8 (0.2) 2.5 (0.1) 3.3 (0.2) 3.2 (0.2) 3 (0.2) 3.5 (0.2) 3.4 (0.2) 3.6 (0.2) 3.4 (0.2) 3.7 (0.3) 3.8 (0.3) 4.3 (0.2) 4.1 (0.3) Primary Reason for Visit Categories (%): Neurologic Psychiatry Others 29 5 62 18 9 69 18 12 69 14 4 79 15 9 73 24 14 62 18 14 60 15 7 75 19 9 71 16 8 75 26 8 65 15 18 67 18 11 71 Diagnosis Category (%) Neurology Psychiatry Pain 74 25 55 66 25 52 67 27 38 63 25 26 74 35 44 69 38 42 64 30 44 66 29 40 74 37 35 60 27 34 73 29 39 70 42 27 72 38 30 Clinic Type (%) General Med Surgery Pediatrics Others 48 25 14 14 44 16 6 33 54 11 10 26 48 16 13 23 48 9 7 35 53 5 7 35 38 24 3 34 48 4 13 35 38 8 7 46 55 3 11 31 37 5 20 38 37 6 8 48 41 9 8 40
116 Table 5-6. Characteristics of New Anticonvuls ant Drugs Visits between 1993 and 2005 (NHAMCS) Variables 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 Drug Types (%) GBP Others 0 100 26 74 74 26 68 32 55 45 72 23 63 27 81 19 68 32 70 30 51 49 58 42 52 48 Sex (Female, %) 40 54 19 48 54 47 46 68 56 63 68 65 66 Race (%) Whites Blacks Others 100 0 0 89 11 0 85 15 0 93 3 3 86 14 0 78 16 6 86 11 3 77 20 3 77 18 5 71 27 3 80 16 4 79 16 5 73 25 2 Av. Age (SD) 26.8 (3.6) 19.1 (5.5) 28.2 (3.6) 46.1 (5.5) 40.3 (5.9) 44.4 (3) 46.3 (2.8) 48.3 (2.6) 45.2 (2.9) 46.3 (1.4) 40 (3.5) 46.4 (1.4) 46.2 (1.6) Region (%): Northeast West South West 0 32 4 64 36 41 23 1 17 10 59 14 37 12 38 12 40 21 22 17 37 40 18 6 43 15 30 12 50 15 27 9 35 18 27 21 35 23 31 12 40 23 30 7 37 25 26 2 27 38 23 11 Av # of Meds/Visit (SD) 2.1 (0.1) 2.8 (0.6) 2.7 (0.7) 3.3 (0.3) 3.8 (0.5) 2.9 (0.3) 3.6 (0.3) 4.0 (0.3) 3.9 (0.2) 4.2 (0.2) 4.6 (0.4) 4.9 (0.2) 5 (0.3) Primary Reason for Visit Categories (%): Neurologic Psychiatry Others 52 8 32 10 0 90 56 0 44 21 17 59 24 0 75 28 0 66 6 11 82 11 17 70 17 10 69 15 8 77 23 12 64 14 13 72 12 9 78 Diagnosis Category (%) Neurology Psychiatry Pain 100 12 44 76 17 68 96 43 38 72 11 53 71 8 49 58 19 64 57 24 29 58 27 43 67 29 33 64 23 44 63 27 39 65 34 35 54 24 32 Clinic type (%) General Med Surgery Pediatrics Others 64 4 0 32 15 42 27 16 51 0 29 20 61 11 3 24 31 16 4 50 43 6 7 44 49 18 2 31 54 5 5 36 44 8 3 43 51 4 6 38 33 7 13 46 52 6 4 39 55 9 4 31
117 Table 5-7. Characteristics of Off-Label and On-Label Visi ts between 1993 and 2005 (NAMCS) Off label Visits (%)On-label Visits (%) P-value Sex (Female) 62,2 54.4 <0.0001 Age (SD) 50.8 (0.5) 42.1 (0.6) <0.0001 Drug Category (New AEDs) 54.8 15.1 <0.0001 Average Meds/Visits (SD) 4.1 (0.10 2.9 (0.1) <0.0001 Race White Black Others 89.3 8.3 2.4 86.9 10.7 2.4 0.23 Primary Reason for Visit Neurologic Psychiatric Others 11 21.7 45.3 38 20.1 40.1 <0.0001 All Recorded Diagnosis* Neurologic Psychiatric Pain 79.6 43.8 35.5 98 40.8 50.8 <0.0001 0.13 <0.0001 Physician Specialty Neurologist Psychiatrist 33.1 26 12.9 30.7 <0.0001 Region Northeast Midwest South West 21.7 18.7 37.3 22.2 21.7 21.4 33.6 23.6 0.23 *Percent adds up to more than 100%, because all possible threee recorded diagnosis were examined Table 5-8. Categories of O ff-Label Indications (NAMCS) Estimate (million) Proportion (%) (95% CI) Neurology 24.9 32.7 (30.0-35.3) Psychiatry 29.8 39.0 (35.9-42.3) Pain 21.6 28.3 (25.4-31.1) Table 5-9. Off-Label Diagnosis Groups by Drug Types (NAMCS) Neurology (95% CI) Psychology (95% CI) Pain (95% CI) CBZ (%) 24(20-31) 50 (42-57) 26 (20-34) DS (%) 23(18-30) 60 (53-67) 17 (13-22) PHT (%) 43(33-53) 16(9-25) 40(32-51) Other OLD AEDs (%) 49(40-59) 21 (13-32) 30(21-39) GBP (%) 41(36-46) 23(20-27) 36(32-41) Other NEW AEDs (%) 18 ( 14-23) 70(60-72) 16(11-22)
118 Table 5-10. Characteristics of Off-Label and On-Label Visits between 1993 and 2005 (NHAMCS) Off-label Visits (%) On-label Visits (%) P-value Sex (Female) 58.8 53.4 0.01 Age (SD) 45.5 (0.7) 34 (1.3) Drug Category (New AEDs) 51.2 13.0 <0.0001 Average Meds Visits (SD) 4.3 (0.1) 2.8 (0.1) Race (%) White Black Others 75.5 21.6 2.9 70.1 26.3 3.6 0.03 Primary Reason for Visit (%) Neurologic Psychiatric Pain 7.3 15.3 75.9 36 10.1 52.5 <0.0001 All Recorded Diagnosis* (%) Neurologic Psychiatrist Pain 73.4 42.6 39.3 95.2 31.2 57.9 <0.0001 <0.0001 <0.0001 Clinic Type (%) General Medicine Surgery Pediatrics Others 45 8.2 3.9 43 36.7 6.2 14.3 42.8 <0.0001 Region (%) Nottheast Midwest South West 40.3 24.1 24.4 11.2 32.4 24.3 31.2 12.1 0.03 *Percent adds up to more than 100%, because all possible threee recorded di agnosis were examined Table 5-11. Categories of Off-Label Indications (NHAMCS) Estimate (million) Proportion (%) (95% CI) Neurology 2.7 27.8 (24.2-31.3) Psychiatry 3.7 38.2 (32.2-44.2) Pain 3.3 34.1 (29.4-38.7) Table 5-12. Off-Label Diagnosis Groups by Drug Types (NHAMCS) Neurology (95% CI) Psychology (95% CI) Pain (95% CI) CBZ (%) 22 (16-28) 51 (43-60) 27 (20-35) DS (%) 17 (12_23) 66 (59-73) 17 (11-22) PHT (%) 34 (24-44) 21 (13-30) 45 (31-58) O-OLD AED (%) 34 (24-45) 27 (14-39) 39 (28-50) GBP (%) 34 (29-39) 24 (17-30) 44 (36-48) O-NEW AED (%) 21 (13-28) 51 (39-64) 28 (18-38)
119 Table 5-13. Association between a 10% Increase in Drug Detailing Contacts and a Percent Increase in all Medication Off-La bel AED Visits between 1994 and 2005 Office (95% CI) Hospital (95% CI) All New AEDs Off-Label Visits 0.008(-0.009-0.024) 0.026(-0.003-0.057) Gabapentin Off-Label Visits 0.018(-0.008-0.044) -0.048(-0.166-0.070) Other New AEDs Off-Label Vis its 0.006 (-0.0001-0.018) 0.023 (0.011-0.036)* Significant at 0.05 level Table 5-14. Association between a 10% Increase in Free Dtug Sampling and Percent Increase in all Medication Off-Label AED Visits between 1994 and 2005 Office (95% CI) All New AEDs Off-Label Visits 0.0003 (0.00007-0.0005)* Gabapentin Off-Label Visits 0.0003 (0.0001-0.0005)* Other New AEDs Off-Label Visits 0.0002 (0.00003-0.0003)* Significant at 0.05 level Table 5-15. Association between a 10% Increase in Off-Label Use Publications and Percent Increase in Medication Off-Labe l AED Visits between 1994 and 2005 Office(95% CI) Hospital (95% CI) All New AEDs Off-Label Visits Controlled Trials Case Reports All Publications -0.001(-0.066-0.065) 0,042( 0.005-0.080.)* 0.012(-0.019-0.044) 0.013(-0.038-0.063) 0.051( 0.022-0.080)* 0.019(-0.007-0.045) Gabapentin Off-Label Visits Controlled Trials Case Reports All Publications -0.034(-0.139-0.070) 0.039(-0.012-0.090) 0.007(-0.033-0.047) -0.007(-0.122-0.108) 0.068( 0.006-0.130)* 0.025( 0.028-0.079) Other New AEDs Off-Label Visits Controlled Trials Case Reports All Publications 0.020(-0.034-0.074) 0.050(0.013-0.088)* 0.019(-0.008-0.046) 0.026(-0.017-0.069) 0.051( 0.025-0.077)* 0.021( 0.001-0.041)* Significant at 0.05 level Table 5-16. Characteristics of NAMCS ALL AEDs Visits between 2001 and 2005 % (95% CI) Age (SD) 48 (0.6) Sex (female) 60 (58-63) Number of Medication (SD) 4.1 (0.9) Race White Blacks Others 90 (87-92) 8 (6-10) 2 (0.8-3) Old AEDs 40 (37-43) GBP Visits 34 (31-36) Other New AEDs 26 (24-29) Source of Payment
120 Private Insurance Medicare Medicaid Others 44 (41-47) 25 (22-27) 15 (13_18) 16 (10-22) Type of Condition Acute Problem Chronic Problem, Routine Chronic Problem, Flare-up Others 17 (15-19) 61 (57-66) 15 (13-17) 7 (05-10) Primary Reason for visit Neurological Psychological Others 17 (16-21) 20 (17-23) 60 (57-63) Diagnosis Neurological only Psychological only Pain only Neurological + Psychological Neurological + Pain Psychological + Pain Neurological + Psyc hological + Pain Others 25 (22-28) 31 (27-34) 21 (18-24) 4 (3-7) 12 (10-13) 5 (4-6) 1 (0-2) 0.2 (0-0.4) Physician Specialty Neurologist Psychiatrist Family Practice Internal Medicine Others 19 (17-22) 13 (27-34) 17 (14-20) 14 (11-17) 18 (16-22) Primary Care Physician (Yes) 37 (33-40) Established Patient (Yes) 93 (92-94) Solo Practice (Yes) 43 (39-47) Metropolitan Statistical Area (Yes) 86 (80-91) Region Northeast Midwest South West 21 (17-25) 20 (16-24) 39 (33-44) 20 (16-25) Visit Year 2001 2002 2003 2004 2005 15 (11-18) 16 (11-20) 23 (19-26) 24 (20-27) 23 (20-28)
121 Table 5-17. Results of Logistic Regression Anal ysis to Determine Predictors of Off-Label Prescribing Odds Ratio 95% Confidence Interval Female 1.0 0.81 -1.30 Age* 1.01 1.01-1.02 Doctor Specialty = Neurologist 0.53 0.23-1.09 Doctor Specialty =Psychiatrist 0.69 0.34-1.40 Doctor Specialty =Internal Medicine 0.90 0.47-1.70 Doctor Specialty =Others* 2.34 1.11-4.92 Primary Care Provider 0.76 0.45-1.28 Primary Visit Reason =Psychological* 1.73 1.13 -2.64 Primary Visit Reason =Others* 2.48 1.72 -3.56 Diagnosis =Psychological 1.35 0.86-2.12 Diagnosis =Pain* 0.52 0.35-0.79 Diagnosis =Neurological + Psychological 1.08 0.51-2.29 Diagnosis =Neurological + Pain* 0.49 0.29-0.82 Diagnosis =Psychological + Pain* 0.25 0.13-0.47 Diagnosis =Neurological +Psyc hological +Pain* 0.14 0.05-0.39 Drug Type=GBP* 43.81 27.55 -69.67 Drug Type=Other New AEDs* 2.27 1.64 -3.16 Number of Medications* 1.23 1.15-1.32 Drug Coverage 1.03 0.71-1.50 Past Visit (>5) 0.99 0.74-1.32 Visit Year=2002* 0.62 0.40-0.96 Visit Year=2003* 0.50 0.32-0.78 Visit Year=2004* 0.55 0.36-0.84 Visit Year=2005* 0.42 0.27-0.65 Significant at 0.05 level.
122 Figure 5-1. Trends in the Pr oportions of All Medication Vis its that are AEDs Visits between1993 and 2005 (NAMCS) Figure 5-2. Trends in the Propor tions of All AED Visits that are Old or New AEDs Visits between 1993 and 2005 (NAMCS)
123 O-OLD AED=Other Old AEDs, O-NEW AED = Other New AEDs Figure 5-3. Proportions of All AE Ds Visits by Drug Type (NAMCS) Figure 5-4. Trends in the Propor tions of All Medication Visits that are AEDs Visits between 1993 and 2005 (NHAMCS)
124 Figure 5-5. Trends in the Propor tions of all AED Visits that are Old or New AEDs Visits between 1993 and 2005 (NHAMCS) O-OLD AED=Other Ol d AEDs, O-NEW AED = Other New AEDs Figure 5-6. Proportions of All AE Ds Visits by Drug Type (NHAMCS)
125 Figure 5-7. Trends in th e Proportions of All Medication and A ll AEDs Visits that are Off-Label AEDs Visits between 1993 and 2005 (NAMCS) Figure 5-8. Trends in th e Proportions of All Off-Label AED Vi sits Attributable to Old or New AEDs between 1993 and 2005 (NAMCS)
126 Figure 5-9. Trends in the Propor tions of All Old and All New AE Ds Visits that are Off-Label Visits between 1993 and 2005 (NAMCS) O-OLD AED=Other Ol d AEDs, O-NEW AED = Other New AEDs Figure 5-10. Proportions of All Off-Label Visits by Drug Type (NAMCS)
127 OLD AED=Other Old AEDs, ONEW AED = Other New AEDs Figure 5-11. Frequencies for On-label a nd Off-Label Visits by Drug Types (NAMCS) Figure 5-12. Trends in the Proportions of All Medication and All AEDs Visits that are OffLabel AEDs Visits between 1993 and 2005 (NHAMCS)
128 Figure 5-13. Trends in the Propor tions of All Off-Label AED Visits Attributable to Old or New AEDs between 1993 and 2005 (NHAMCS) Figure 5-14. Trends in the Propor tions of All Off-Label AED Visits Attributable to Old or New AEDs between 1993 and 2005 (NHAMCS)
129 O-OLD AED=Other OlD AEds, ONEW AED = Other New AEDs Figure 5-15. Proportions of Off-La bel Visits by Drug Type (NHAMCS) OLD AED=Other Old AEDs, ONEW AED = Other New AEDs Figure 5-16. Frequencies for On-label a nd Off-Label Visits by Drug Types (NHAMCS
130 Figure 5-17. Trends in AEDs Detailing Contac ts in Both Physician offices and Hospitals between 1994 and 2005 OLD AED=Other Old AEDs, ONEW AED = Other New AEDs Figure 5-18. Proportions of Detailing Contacts by Drug Types
131 O-OLD AED=Ot her Old AEDs Figure 5-19. Trends in Old AED Detailing Contacts between 1994 and 2005 Figure 5-20. Trends in GBP De tailing Contacts between 1994 and 2005
132 O-NEW AED = Other New AEDs Figure 5-21. Trends in Other New AEDs Detailing Contacts between 1994 and 2005 OFF_CON= OFFICE CONTACTS, HOSP_CON = HOSPITAL CONTACTS Figure 5-22. Trends in All AED Detai ling Contacts by Settings between 1994 and 2005
133 Figure 5-23. Trends in AEDs Free Drug Sampling by Drug Category between 1994 and 2005 CT= CONTROLLED TRIALS, CR=CA SE REPORTS, A=NAMCS, B=NHAMCS Figure 5-24. Comparison of Trends in the Public ations of RCTs and Case Reports of Off-Label Use of Gabapentin and Trends in Proportion of All Medication Visits that are Off-Label Gabapentin Visits in both NAMCS and NHAMCS
134 CT= CONTROLLED TRIALS, CR=CA SE REPORTS, A=NAMCS, B=NHAMCS Figure 5-25. Comparison of Trends in the Public ations of RCTs and Case Reports of Off-Label Use of all Other New AEDs and Trends in Proportion of All Medication Visits that are OffLabel Other New AEDs Visits in both NAMCS and NHAMCS
135 CHAPTER 6 DISCUSSION Overall Off-Label Drug Use The results from this study indicates that from the year 1993 through 2005, 62% (95% CI: 59-64) of all anticonvulsant drug vi sits to both physician office a nd hospitals OPDs, were visits at which at least one anticonvul sant drug was prescribed for an unapproved indication. This offlabel rate is higher than what has been reported fo r other off-label studies such as 32% of all drug treatments for dermatological diseases57, 33% for all cancer drug treatments3, 40% for all drug treatments for AIDS patients4, and 52% of all -blocker use56. However, the off-label rate reported in this study is lower than what was report ed in three earlier studi es of off-label use of anticonvulsant drugs; two in Medicaid populations7, 66 and one in a large psychiatric hospital inpatient population69. The two Medicaid study repor ted off-label rates of 71%7 and 80%66, while that conducted in a psychiat ric hospital reported a rate of 92%. The differences in the population studied (i.e. only Medicaid patients and patients admitted to a psychiatry hospital in three previous studies), the exclusion of benzodi azepines in the pool of anticonvulsant drugs and the narrower definition of off-label drug use mi ght explain the differences between the findings of this study and that of prior off-label anticonvu lsant use studies. One of the Medicaid studies reported a reduction (from 71% to 67%) in the prevalence of off-label use on exclusion of patients who received benzodiazepine. Therefore, the exclusion of benzodiazepines in the pool of included anticonvulsants might have contributed to the lower off-label rates reported in this study. The finding of high off-label use of anticonvuls ant drugs in this study and prior off-label anticonvulsant use studies implies that clinical practice use of anticonvulsant drugs is very different from the directives for approved use given in the product label. The off-label
136 prescribing rates found in this study and in other studies that have eval uated off-label use of drugs questions the policy of the FDA to requi re efficacy certification for unapproved drugs, but not for new uses for approved drugs. In other words, there seems to be a consistency issue surrounding the issue of off-label prescribing of pr escription drugs. In addition, the steep growth of the unregulated off-label pres cribing sector proposes that efficacy requirement in the US drug laws might be expandable. New anticonvulsants were prescr ibed off-label more often th an old anticonvulsants and gabapentin was a key driver of off-label anticonv ulsant use. Twenty-sev en percent and 24% of all anticonvulsant visits were gabapentin visits in office and hospital settings respectively; however, gabapentin off-label vi sits constituted 40% (physicia n office) and 39% (hospitals OPDs) of all off-label visits observed during th e study period. The finding that gabapentin is more frequently used off-label than any of the other anticonvulsant drugs is consistent with what has been reported in the literature6, 7, 66. In their study that examined off-label prescribing among office base physicians, Radley and colleagues re port that of more than 100 most commonly prescribed drugs, gabapentin had the highest proportion of off-label prescriptions (83%)6. Recently, the FDA reviewed 199 placebo controll ed studies of 11 anticonvulsant drugs and found a doubling of the risk of su icidal thoughts and behaviors in patients taking any one of the 11 anticonvulsants when compared to those taking placebo171. The higher risk of suicidal thoughts and behaviors was observe d at one week after starting any of the 11 anticonvulsant drugs and up to at least 24 week s after drug therapy. The results were reported to be generally consistent among the entire drugs studied and we re seen in all demographic subgroups and no pattern of risk modification was observed acro ss the age groups. About 80% of the drugs reviewed were new anticonvulsant drugs. In light of this recent report, the high rate of off-label
137 anticonvulsant use reported in this study becomes even more concerning and requires that clinicians critically weigh the ev idence supporting risks and benef its associated with off-label use of anticonvulsants before pres cribing to their patients. In cases where off-label use of a particular drug becomes very prevalent and may present a potential safety concern, these agencies should be more proactive and initiate studies to quantify the safety and effectiveness of off-label drug use and implement strate gies to ensure patient safety. It has been proposed by some aut hors that the high prevalence of off-label drug use such as that reported in this study may be a source of confusion to patients and the public16. When drugs are prescribed for off-label indications, patien ts may fail to find the indication in common drug information sources such as the drug package inse rt since, such sources predominantly contain information about uses that has been approved by the FDA. Therefore, patients may misinterpret the intention of the prescribing physicians. No law mandates physicians to disclose their intention to use a drug for off-label purposes to patients; therefor e, patients might not be aware that their drug was prescribed for an unapproved indi cation. In a survey of off-label use of mood stabilizers in a larg e psychiatric hospital, only 30% of all patients who received mood stabilizers for off-label indications and their caregivers were informed that the drugs were been prescribed for an off-label indication69. Even in situations where clin icians make it known that a drug is prescribed for off-label indication patients mi ght have problems understanding the information regarding off-label use. If offlabel prescribing intent ion disclosure is uncl ear to patients or inadequate, patients might also mistake off-label us e as illegitimate use rather than as an issue related to with drug regulation. The potential e ffect of this problem on drug adherence is worth evaluating.
138 Off-Label Prescribing Trends The trends in anticonvulsant drug visits and o ff-label anticonvulsant drug visits observed from both settings (physician offi ce and hospitals outpatient depart ment visits) were remarkably similar. Data from the two sources indicate th at during the period under re view off-label use of new anticonvulsant drugs grew, while the off-la bel use of old anticonvulsant drugs remained stable over time, therefore, the overall increase in the off-label use of anticonvulsant drugs observed in this study was solely attributable to the off-label use of new anticonvulsant drugs. However, it seems that there was a more sustained increase in off-label use in the hospital setting than physician office setting, since the decrease in off-label prescribing no ticed in office setting towards the end of the study period was not observed in hospital setting. Though, there was no change observed in the proportions of all old anticonvulsants prescription that were off-label over time, it is remarkable that thr oughout the study period, about half of all prescriptions for old anticonvulsants were pres cribed for off-label indications. This is a considerably high proportion of off-label use. Similarly, the proportion of all new anticonvulsants prescription that was off-label remained consistently above 65% throughout the study period in both study populations. It seems that new anticonvulsants were extensively used off-label almost immediately after product approval. In other words, there did not seem to be a lag period after product launch, be fore substantial off-label use started. The literature on the epidemiology of epilepsy in the United States in dicates that there has not been a significant increase in the prevalence of epileptic conditions in the past 30 years74, 172. Since no change was observed in the prescription of ol d anticonvulsant drugs over time, it is logical that the increasing prescribing rates of new anticonvulsant drugs we re for conditions other than epilepsy or convulsion related conditions. The growth in o ff-label use of new anticonvulsant drugs over time is not totally unexpected, since the clinic al use of any drug cha nges over time and the
139 direction of change depends on a number of factors such as mark eting and product features such as safety profile, and ease of use. Usually as physicians become more comfortable with a drug, they become more likely to prescribe it for o ff-label purposes and espe cially when no serious adverse reaction has been attributed to the drug. However, the immediate start of off-label use observed with new AEDs proposes that initi al marketing targeted already unapproved indications. The proportions of all off-label visits that were gabapentin visits or other new anticonvulsants visits increased ov er time, (with the proportion for gabapentin higher than that for the other new anticonvulsant drugs thr oughout the period under review), while the proportions of all off-label visits that were old anticonvulsant drugs visits decreased. The controversies and lawsuits (settled late in 2004) over off-label use of gabapentin did not appear to result in a substantial decrease in off-la bel use of gabapentin or all the other new anticonvulsants especially in the hospital setting. Off-Label Drug Use and Drug Promotions The findings from the examination of the effects of drug promotions on off-label prescribing behaviors of physicians suggests that overall, physician deta iling visits for new anticonvulsant drugs might not have had an effect on the observed increase in off-label visits observed in both office and hospital settings. Ho wever, trends in free drug sampling were found to be consistently associated w ith trends in off-label drug prescr ibing of all new anticonvulsants, though the magnitude of the association was small. Although the results from this study do not sh ow a distinct link or strong association between off-label anticonvulsants prescribing and drug promotiona l activities, there are reports in the literature that alludes to a positive relationship betwee n marketing activities of the pharmaceutical industry and off-label prescribing activities of physicians. In their study on the
140 characteristics and impact of drug detailing fo r gabapentin, Steinman and colleagues conducted a qualitative review of market research forms co mpleted by physicians after receiving detailing visits for gabapentin173. They report that detail contacts fo r gabapentin often involved messages about unapproved use, and frequently resulted in physicians intention to increase their future prescribing of gabapentin. Th erefore, detailing efforts by the industry might have an positive influence on off-label prescribi ng behavior, though this relations hip might not be of the type commonly observed between two variables that are associated with each other (e.g. more of variable X = more of variable Y). Rather the re lationship might be such th at regular and constant levels of promotional activities have multiplica tive effects on off-label prescribing activities. Therefore, even with little or no increase in detailing visits over time can result in an overall increase in off-label prescribi ng activities. On the other hand, th e results of this study can also be a reflection of the fact that there might be ot her factors more pertinent in explaining off-label prescribing rates than promotiona l activities of the industry. F actors such as personal off-label drug use communication between physicians, positive feedback on utility of prescription drugs for off-label purposes from patients, lack of approved drugs in trea ting certain conditions, inquiries and pressure from patients with knowle dge of certain off-label use of drugs. Though the effects of these factors on off-label prescrib ing were not studied in this inquiry, there are reports in the literature that indicate that these factors ca n influence physician prescribing behavior129, 160, 161, 166. Finally, it seems that the trend in detailing c ontacts for gabapentin was artificially affected by the outcome of lawsuit over off-label promoti on of gabapentin, which was finally settled in the later part of 2004 (number of detailing contacts for gabapen tin went from about 90,000 in the last quarter of 2004 to less than 10,000 in the first quarter of 2005). It might be difficult to
141 observe the true effects of deta iling contacts on off-label visits in general, since promotional efforts for gabapentin accounted for about a quarter of all new anticonvulsant promotional activities. In summary, the present study does not show any consistent effect of promotional activities on off-label prescribing. More studies are requi red to quantify the effect of promotional activities and other potentially influential factors on off-label prescribing activities. Off-Label Publications Trends in case reports/case series publication of off-label use of all anticonvulsant drugs, gabapentin and all o ther new anticonvulsant dr ugs were found to be positively associated with the variation in the trends in off-label prescr ibing of new anticonvulsant drugs in both settings, while the trends in publication of controlled trials did not appear to influence off-label prescribing behavior of physicians The fact that case reports of off-label use of drugs and not controlled trials of offlabel use of drugs appears to influen ce off-label prescribing behavior of physicians is a cause of concern. Case reports are publications that describe and interpret individual cases, and case series publications usually document retrospective studies of case records, usually collected in one institution or in an individual prac tice. Both kinds of publications are not adequate to explain clinical events or to assign causality. They generally depict the lowest form of clinical evidence, a nd are only valuable in detecting areas for further investigation and research. Consequently, they should not be the basis for treatment decisions. The findings of a consistent association be tween case reports and physicians off-label prescribing behavior raises quest ions about the quality of the ev idence that physicians appear to use in making clinical decisions. Other studies have also re ported an association between case reports and off-label prescribing behaviors. In a survey to ex amine the perception and practice of off-label
142 prescribing in dermatology, dermat ologists were asked to indicate their sources for information on off-label drug indications55. Of the 3 options listed (p ersonal experience, anecdotal experience of colleagues, and case reports in the literature) case re ports in the literature was the most frequently cited information source for poten tial off-label drug use, with about 40% of respondents indicating that they obtained information on offlabel drug indications from published case reports. Although th ere are no laws or guidelines that presently mandate clinical trial evidence for off-label prescribing, it is gene rally expected that prescribing decisions should be based on valid supportiv e clinical data. The results from single case reports or even from a series of case reports that are generally acknowledged to be non-c ontrolled and merely represent individual practice experience are not adequate. There are reports in the literature that suggest that the publications of open label studies on off-label use of gabapentin served as key el ement in the marketing strategy for the drug21. The goal of which was to use publication of research results not as a means to gain FDA approval for new indications but to disseminate the off-labe l information as widely as possible through the worlds medical literature, thereby generating exc itement in the market and stimulating off-label prescribing. This strategy was reported to be focused primarily on expanding gabapentin use in neuropathic pain and bipolar disorder21. Furthermore, a review of the evidence for the off-label use of gabapentin carried out to summarize medical information pertaining to its off-label use reports that of 10 common offlabel indications for gabape ntin only one was supported by evidence from blinded randomized controlled tria ls (neuropathic pain), while the evidence for off-label use in other disease conditions were ma inly based on open label studies, case series reports and case reports. In summary, it seems th at physicians willingne ss to make clinical decisions on off-label drug use based on studies other than blinded randomized controlled trials
143 was exploited by the industry to encourage off-la bel prescribing of gabapentin. Though no such reports have been reported fo r other anticonvulsant drugs, the in creasing off-label use of these other new anticonvulsant drugs and the increasing publications of case reports and case series on successive use of these drugs for off-labe l indications, might suggest that the same relationship exist for the other new anticonvu lsant drugs, though there is no evidence that suggests that these publications ar e sponsored by th e manufacturers. Predictors of Off-Label Prescribing Patient Characteristics Previous studies on off-label drug use have f ound that certain patient characteristics can influence off-label prescribing behavior of physicians. Patient age was the only patient characteristic that was found in this study to be associated with off-label prescribing of anticonvulsant drugs. Increasing age increased the probability of receiving an off-label prescription of anticonvulsant drugs, with a 1% increase in the risk of off-label prescribing of anticonvulsant drugs for every one year increase in age. This is consistent with the findings from the study on off-label use of anticon vulsants in Georgia Medicaid patients66. However, unlike the findings from the present study, the Georgi a Medicaid study also indicates that White patients and female patients were more likely to receive off-label prescrib ing of anticonvulsants. No such association was found in this study as patients gender and race did not appear to significantly predict off-label pr escribing. Though there are some expected differences between the population of patients studied in this inquiry and Georgia Medicaid in terms of racial and gender make up, which might explain the differe nces observed in the findings, it is also important to point that due to the issues a ssociated with the non-ra ndom selection survey methodology used in collecting NAMCS data, the so ftware used in the analysis is programmed to produce standard errors that are very conserva tive. Therefore, the an alysis using such data
144 and software are more likely to be prone to t ype II errors. The asso ciation between off-label prescribing and about 40% of the covariates evaluated were notable, though statistical significance was not observed. Drug coverage status of a patient as de fined in this study was another patient characteristics found not to influence off-label prescribing. Patient drug coverage status was constructed from insurance type information fr om the NAMCS dataset. As of the end of 2005, which was the last NAMCS year studied in this inquiry, the Medicare Part D benefits for Medicare eligible patients had not yet kick ed in. Though, it was and is still possible for Medicare beneficiaries to have supplemental insurance that could cover drug cost, it was impossible to determine whether patients identifie d as having Medicare insurance in the NAMCS datasets also had drug coverage through supplemental insurance plan s or as a result of retirement benefits. Furthermore, a large percentage of the study population had private insurance, and there was no way of identifying pa tients drug coverage status based on the information from the NAMCS dataset. Therefore, only Medicaid pa tients and patients with workers compensation were categorized as having drug co verage (since these were the onl y groups that could be said with some form of certainty to have drug covera ge). Hence, the lack of association observed between off-label drug prescribing and patient dr ug coverage status mi ght be due to bias introduced as a result of misclassificati on of patients drug coverage status. The duration or time span of a disease conditi on was also found not to be associated with off-label prescribing. This finding might not be totally unexpecte d, since most on-label indications and most of the o ff-label indications of anticonvu lsants are generally chronic conditions. Consequently, the time span of the condition for which an anticonvulsant was prescribed for might not be a good predictor of physician off-label prescribing behavior.
145 Finally, compared to patients who only had diag noses of neurological or igin, patients with diagnoses related only to pa in, or neurology and pain, or psychology and pain or neurology, psychology and pain all had a lower/risk of offlabel prescribing. The previously referenced study in Medicaid patients had examined the relationship between off-label prescribing and different psychological conditions and found that patients with certain psychological conditions had a higher risk of off-label prescribing (e.g. schi zophrenic disorders) while patients some other types of psychological conditions had a lower risk of label prescribing (e.g. bipolar affective disorders)66. It is therefore difficult to say how th e finding from this study compares to the findings from the study done in Medicaid patien ts, since all psychological conditions were examined as a group in this study. Physician Characteristics Off-label prescribing rate by general/fam ily pr actice practitioners was comparable to those by neurologist, psychiatrist, and internal me dicine practitioners, however, compared to general/family practitioners, provid ers who were categorized as other specialist groups (excludes neurologist, psychiatrist, family/g eneral and internal medicine practitioners) were more likely to prescribe anticonvulsants for off-label indications Findings from the Me dicaid studies indicate that neurologist in one study7 and psychiatrists in the other study66 were more likely to prescribe anticonvulsants for off-label indications than nonneurologist and non-psychiatrists respectively. Though in this study, physician specialty was subdivided into more categories than just two (neurologist/non-neurologist, or psychiatrist/non-psychiatrist), th e observed association between physician specialty and off-label prescribing did not change on categorizing physician specialty as neurologist/non-neurol ogist or psychiatrist /non-psychiatrist .Therefore the difference in physician specialty classification methods does not a ppear to explain differe nces in the observed results. Other studies have also examined th e relationship between off-label prescribing and
146 physician specialty in other therapeutic classe s. The study on off-label prescribing of -blockers found that cardiologist were more likely to prescribe -blockers for off-label purposes than noncardiologist, while the study in dermatology foun d that dermatologist were less likely to prescribe drugs for off-label dermatology conditions However, both studies report unadjusted risk. From the ongoing discussion, it can be said that the direction of th e effect of physician specialization on off-labe l prescribing behavior is debatabl e and might depend on the type of drug under evaluation. Visits made to physicians in solo practices or to physicians practic ing in urban areas did not result in higher likelihood of off-label prescription. No othe r study has evaluated the effects of the type of physician practice on off-label pr escribing behavior; howev er, the first Medicaid study also reports no association be tween location of physician (major city vs. rural location) and off-label prescribing behavior. Visit Characteristics Patients who had drug visits in 2002, 2003, 2004 or 2005 had a lower likelihood of getting an anticonvulsant drug for off-label purposes when com pared to patients who had such visits in 2001. It seems off-label proportion peaked in 200 1 and gradually decreased over time. Since gabapentin was a major contributor to the observed volume in off-label prescribing of anticonvulsants, it is reasonable to expect that the reduced risk in off-label prescribing observed with each year after 2001, might be associated with the lawsuit over off-label promotion for gabapentin. Though the lawsuit was filed before 2001 (in 1996), and finall y settled in after 8 years (in 2004), the publicity generated by the lawsuit might have resulted in more cautious attitude of physicians towards off-label prescribing of gabapen tin, thereby resulting in less offlabel gabapentin prescriptions over time.
147 An increasing number of medications prescr ibed at visits were found to be positively associated with higher odds of off-label prescr iption of anticonvulsants. Visits in which the primary reason were for psychological or other r easons that did not incl ude neurological and psychological related reasons were more likely to result in off-label prescribing when compared to visits at which the primary reason for visit wa s neurological in origin. This finding seems contrary to the findings related to diagnosis discussed earlier (vis its with neurological diagnosis having a higher probability of off-label prescrib ing than other diagnostic categories), since it is logical to think that reasons for physician visit and physician diagnos is should have some commonality. The primary reason for visit va riable in the NAMCS is coded based on information from the patients, while the diagnosis va riable as used in this inquiry consists of a combination of both primary and any additional re ported physician diagnosis recorded for that visit (up to 2 additional diagnoses), therefor e both variables might be measuring two very different constructs. Finally, compared to old anticonvulsants, gaba pentin and all the ot her new anticonvulsants were more likely prescribed for off-label indica tions. Gabapentin visits were 28 to 70 times more likely to be off-label visits, while other new anticonvulsants visits had double the risk of off-label prescribing. This is consiste nt with reports from previous studies7, 60. Differences in presumed safety profiles, effectiveness for o ff-label indications and marketing might explain some of the differences in off-label prescribing rates between new and the old anticonvulsants. Limitations This study exam ined the trends in off-labe l prescribing of anticonvulsant drugs and also examined possible predictors of off-label an ticonvulsant prescribing. The most important limitation of this study is that the two ambulatory datasets used in this inquiry do not have drugdiagnosis linkage, therefore it impossible to be certain about the inte nt of the physician in
148 prescribing the study drugs. In additi on, the NAMCS and NHAMCS data set coded 3 diagnosis and 6 medications in the earlier studies (1993 to 2002) and 8 medications thereafter (20032005). Theoretically, it is possible that some off-l abel visits that were identified might actually have been misclassified as off-la bel visits due to the restriction on 3 or less diagnosis recorded for each visit. Upon examination of the included observations, about 38% (95% CI 36-40) of all included AED visits and 41% (95% CI 38-44) had up to 3 valid diagnoses as per study protocol. Therefore it is likely that there was an overestimation of off-label visits due to this limitation of the data. Furthermore, the NAMCS and NHAMCS datase ts do not capture all relevant patient physician and visit characteristics that might in fluence physicians prescribing behavior. The NCHS uses a survey instrument to collect data for the both the NAMCS and NHAMCS. Due to time constraints on participating phy sicians, the data reported seems to lack sufficient detail. Another limitation that is related to the type of information obtaina ble from the data is that this study could only measure indication related off-label; therefore it is possible that the proportion of off-label visits would have been higher if prescribing relate d to dosage, duration of time and route of administration were considered. Another limitation of this study is the potential error that is associated in analyzing national sampling datasets such as the NAMCS and NHAMCS. Due the way these data are collected, statistical programs appropriate to analyze such no n-random data are programmed to produce conservative standard errors, therefore such analysis are more prone to type II errors. Additionally according to the NCHS documenta tions for the NAMCS and NHAMCS, statistics that are based on <30 raw data visits or havi ng a relative SED >30% are deemed to have low reliability. To increase the re liability of results in this st udy, some variable categories (e.g.
149 physician specialty) were collapsed to obtain suffi cient raw frequencies. Therefore there is a possibility that the way some of the study vari ables were defined and measured might have introduced imprecision in the study. Finally, various assumptions were made in the conduct of this st udy that might have affected the observed results. Such assumpti ons include the application of a quarterly depreciation rate of 30% for physician detailing visits in the construction of a cumulative measure of detailing visits to physicians. Other assumptions incl ude treating Medicaid patients as the only group of patients with drug coverage and assuming that only pub lications that report positive results for off-label drug use can influence off-label prescribing activities. Policy Implications Off-label p ractice represents a peculiar state where there is a departure from the regulatory institutions directives on how drugs or medical devices should be used; however, such departure is not considered illegitimate or prohibited. In most cases, the off-label drug practices usually have some form of rationale for its use, though in most cases, there is insufficient evidence to allay safety, efficacy, and cost-effectiveness concerns. Off-label drug use raises a multitude of ethical and policy questions and concerns for which to date there has not been suitable resolutions. In a likewise manner there are a numb er of questions that ar e left unanswered after examining the results of this study. These include but are not limited to ques tions like: Is the rate of off-label drug prescribing reported unacceptabl y high? What is an acceptable off-label prescribing rate? Should the acceptable off-label ra tes/standards be the same for different drugs? Who or what agencies should decide on what an acceptable off-label rate should be? What are the clinical and economic outcomes of off-label use? By far the most important question that will impact the policy on off-label drug practice is the safety and efficacy questions surrounding off-la bel drug practices. While a few studies such
150 as the present study have examined the extent of off-label use of certain drugs, studies on the outcomes as a result of off-label practices are very few. The present si tuation is unacceptable, since the main concern against off-label drug use is the lack of adequate safety and efficacy data. Therefore there is an urgent n eed to study the clinical outcomes of off-label practices. Such studies should ideally be prosp ective in nature and should be centered on drugs that have been reported to have high off-label use and drugs that have been associated with serious adverse effects. The FDA recognizes the fact that the best interests of the pa tient requires that providers use legally available drugs, biologics and devices according to their best knowledge and judgment174. In other words, physicians should be fr ee to make a judgment about the appropriate use of any FDA-approved drug by examining the pack age insert information, available scientific and clinical information, and thei r own experience (and that of th eir colleagues) in prescribing the drug. While the FDA policy of not regulating the practice of medicine is considered satisfactory by most stakeholders and therefore, has not been the in the forefront of the debate regarding of-label practice, this has not been th e case with off-label promotional practice. This study found some association between some promo tional activities of the industry, case report publications and off-label prescribing rates. Ther efore, there is reason to believe that there are certain activities of the industry and other bodies that might have the potential to increase offlabel prescribing rates beyond what would have been expected without those activities. Potential Solutions to the Various Prob lems Posed by Off-Label Prescribing a) Physicians should m ake it their responsibility to be armed with adequate information on off-label use of any drug they are considering fo r off-label prescribing. In addition to this is also to inform patients on the off-label stat us of their drug therapy (informed consent). In this case, informed consent is considered as a process of discussing the drug therapy
151 with the patient and not the traditional an inform ed consent as practiced in clinical trials. However, physicians are encouraged to document such conversations. b) Professional bodies should consider doing more to offer guidance on off-label drug use. As more information about a particular o ff-label drug use becomes available, this information should be passed down to me mbers through newsletters, continuing education programs and presentations at meetin gs. This however, has to be independent of the drug manufacturers, unless it becomes a vehicle for manufacturers to promote offlabel drug use to physicians. There is also potential for government agencies such as AHRQ to participate in the offering guidan ce on off-label prescribing. Just recently AHRQ commissioned a study that examined the evidence for certain off-label use of antipsychotics175. c) Medical journals should solic it and publish early accounts of off-label drug uses. Such articles should ideally be from independent researchers with no tie s to the pharmaceutical industry. Journal should al so publish updated and correcting accounts as new cases and more information becomes available. d) Patients should be more proactive in their care, especially when a doctor initiates the conversation about off-label drug treatment. They should at minimum, ask the doctor for the information on the risk associated with the use of the drug in general and in particular for the off-label use in the patients particular indication. It will also benefit the patient to ask about his or her plan coverage po licy concerning off-label drug use. e) Patients might also consider enrolling in tria ls that are evaluating an off-label use of drugs. In 2005 the CMS released new draft gui delines for what it called Coverage with Evidence Development. This guideline required patients to take part in data collection
152 on off-label use of certain drugs in other to r eceive coverage for these drugs. The goal of this guideline was to collect better evidence to improve health outcomes. While there are many people that do not think it is unfair to force people to enroll in a study just so that their drugs will be covered, it is an effective way of obtaining data on the effectiveness and safety of off-label use of drugs. f) The FDA should make more effort to impr ove post-marketing scrutiny of approved drug products. Improved post-marketing surveillanc e system will give health professionals and the FDA better information about the sa fety and effectivene ss of both on and offlabel drug uses. In cases where off-label use of a particular drug becomes so widespread and becomes a potential safety concern, the FDA should be more proactive and on their own initiate studies to quantif y such safety issues. In other words, the FDA should depend less on industry sponsored studies and rely more on in-house studies. g) There FDA should consider creating less de manding approval process for supplemental new drug approval. There are suggestions for the FDA to use combined data from offlabel drug use clinical trials and data collected from off-label use of drugs in clinical practice in accessing the safety and effectiveness of dr ugs under consideration for expanding their approved uses. Conclusion This study found that of f-label use rate of an ticonvulsants was higher than what has been previously reported for other medications and th at there has been a general increase in the proportion of anticonvulsant drugs prescribed for o ff-label purposes. Though safety of off-label use was not evaluated in this study, the observed trend is of great con cern considering safety issues regarding the use of drugs for off-labe l indications. The study also provided additional evidence for the effects of certain patients, prov iders and environment factors on the rates in off-
153 label prescribing. In light of the present political interest in the safe and cost-effective use of prescriptions, a there is a need for the creation of a national forum of stakeholders to enable the formulation of policies that will offer providers the autonomy to tr eat patients according to their best knowledge and judgment and at the same time ensure safe and cost-effective use of prescription drugs. In addition ther e is an urgent need for studies that will examine the legal, economic and clinical impact of off-label drug use.
154 APPENDIX A DESCRIPTION OF NAMCS AND NHAM CS DATA COL LECTION PROCEDURE A. INTRODUCTION : The NAMCS is a national probability sample survey conducted by the Division of Health Care Statistics, National Center for Health Statistics (NCHS), Centers for Disease Control and Prevention (CDC). A nationa l sample of office-based physicians provides data on patients office visits. These data are weighted to produce na tional estimates that describe the utilization of ambulatory medi cal care services in the United States. B. SCOPE OF THE SURVEY : The basic sampling unit for the NAMCS is the physicianpatient encounter or visit. Only visits to the offices of nonfederally employed physicians classified by the American Medical Association (AMA) or the American Osteopathic Association (AOA) as "office-base d, patient care" are included in the NAMCS. Physicians in the specialties of anesthesiology, pathology, a nd radiology were excluded from the physician universe. Types of contacts not included in the NAMCS were those made by telephone, those made outside the physician's office (for example, house calls), visits made in hospital settings (unless the physician has a private office in a hospital and that office meets the NAMCS definition of office), visits made in instituti onal settings by patients for whom the institution has primary responsibility over time (for example, nursing homes), and visits to doctors offices that are made for administrative purposes only (for example, to leave a specimen, pay a bill, or pick up insurance forms). C. SAMPLING FRAME AND SIZE OF SAMPLE : The sampling frame for the NAMCS is composed of all physicians contained in the ma ster files maintained by the AMA and AOA, at a point roughly 6 months prior to the start of the survey year, who met the following criteria: -Office-based, as defined by the AMA and AOA; -Prin cipally engaged in patie nt care activities; -Nonfederally employed; -No t in specialties of anesthes iology, pathology, and radiology. D SAMPLE DESIGN : The NAMCS utilizes a multistage probability design that involved probability samples of primary sampling units (PSUs), physician practices within PSUs, and patient visits within practices. The first-stag e sample included 112 PSUs. A PSU consists of a county, a group of counties, county equivalents (suc h as parishes and independent cities), towns, townships, minor civil divisions (for some PSUs in New England), or a metropolitan statistical area (MSA). MSAs were defined by the U.S. Of fice of Management and Budget on the basis of the 1980 Census. The first-stage sample consisted of 112 PSUs that comprised a probability subsample of the PSUs used in the 1985-94 Na tional Health Interview Survey (NHIS). The NAMCS PSU sample included with certainty the 26 NHIS PSUs with the largest populations. In addition, the NAMCS sample included one-half of the next 26 largest PSUs, and one PSU from each of the 73 PSU strata formed from the re maining PSUs for the NHIS sample. The NHIS PSU sample was selected from approximately 1,900 ge ographically defined PSUs that covered the 50 States and the District of Columbia. The 1,900 PSUs were stratified by socioeconomic and demographic variables and then selected with a probability proportional to their size. Stratification was done within four geogra phical regions by MSA or non-MSA status. The second stage consisted of a probability sample of practici ng physicians selected from the master files maintained by the American Medica l Association (AMA) and American Osteopathic Association (AOA). Within each PSU, all eligible physicians were stratified into fifteen specialty groups: general and family practi ce, osteopathy, internal medicine pediatrics, general surgery, obstetrics and gynecology, orthopedic surgery, cardiovascular diseases, dermatology, urology, psychiatry, neurology, ophthalm ology, otolaryngology, and "all othe r" specialties. The final
155 stage was the selection of patient visits within the annual practices of sample physicians. This involved two steps. First, the to tal physician sample was divided into 52 random subsamples of approximately equal size, and each subsample was randomly assigned to 1 of the 52 weeks in the survey year. Second, a systematic random sample of visits was selected by the physician during the assigned week. The sampling rate varied for this final step from a 100-percent sample for very small practices to a 20-percent sample for ve ry large practices as determined in a presurvey interview. The method by which the sampling rate was determined is available on the NCHS website. H. CONFIDENTIALITY In April 2003, the Privacy Rule of the Health Insura nce Portability and Accountability Act (HIPAA) was implemented to establish minimum Federal standards for safeguarding the privacy of individually iden tifiable health information. No personally identifying information, such as patients name or address or Social Security number, is collected in the NAMCS. Data collection is authorized by Section 306 of the Public Health Service Act (Title 42, U.S. Code, 242k). All information coll ected is held in th e strictest confidence according to law [Section 308(d) of the Public H ealth Service Act (42, U.S. Code, 242m(d))] and the Confidential Information Protection and Statistical Efficiency Act (Title 5 of PL 107-347). The NAMCS protocol was approved by the NCHS Research Ethics Review Board in February 2003. Waivers of the requirements to obtain informed consent of patients and patient authorization for release of patient medical record data by health care providers were granted. In the spring of 2003, the NAMCS implemented additio nal data collection procedures to help providers assure patient confid entiality. Census Bureau Field Representatives were trained on how the Privacy Rule allows physicians to make disclosures of protected health information without patient authorization for public health purposes and for research that has been approved by a Research Ethics Review Board. Physicians were encouraged to accept a data use agreement between themselves and NCHS/CDC, since the Pr ivacy Rule allows physicians to disclose limited data sets (i.e., data sets with no direct patient identifiers) for re search and public health purposes if such an agreement exists. Assurance of confidentiality was provided to all physicians according to Section 308 (d) of the Public Health Service Act (42 USC 242m). Strict procedures were utilized to prevent disclosure of NAMCS data. All information which could identify the physician was confidential and was seen only by persons engaged in the NAMCS, and was not disclosed or released to others for any other pur pose. Names or other identifying information for individual patients were not rem oved from the physicians office. K. ESTIMATION PROCEDURES : Statistics produced from th e National Ambulatory Medical Care Survey were derived by a multistage estimation procedure. The procedure produces essentially unbiased national es timates and has basically four components: 1) inflation by reciprocals of the probabilities of selection, 2) adjustment for non response, 3) a ratio adjustment to fixed totals, and 4) weight smoothing. Each of these components is described below. 1. Inflation of Reciprocals by Sampli ng Probabilities Since the survey utilized a three-stage sample design, there were three probabilities: a) the prob ability of selecting the PSU; b) the probability of selecting a physician w ithin the PSU; and c) the probability of selecting a patient visit within the physician's practice. The last probability was defined to be the exac t number of office visits during the physician's specified reporting week divided by the number of Patient Record forms completed. All weekly estimates were inflated by a factor of 52 to derive annual estimates. 2. Adjustment for Nonresponse Estimates from NAMCS data were adjusted to account for in-scope physicians who did not provide PRFs (non-PRF physicians) either because they saw no patients during their
156 sample week or failed to provide PRFs for visi ts by patients they did see during their sample week. Beginning with 2004 data, changes were ma de to the nonresponse ad justment factor to account for the seasonality of the reporting period. Extra weights for nonresponding physicians were shifted to responding physicians in reportin g periods within the same quarter of the year. The shift in nonresponse adjustment did not signi ficantly affect any of the overall annual estimates. Beginning with 2003 data, the adjustme nt for non-PRF physicians differs from the adjustment used in prior years. Previous ly the adjustment accounted for non-response by physician specialty, geographic re gion, and metropolitan statistical area status. The revised nonresponse adjustment also accounts for non-res ponse from physicians by practice size, as measured by number of weekly visits, and for vari ability in number of weeks that participating physicians saw patients during the y ear. Previously, these characteri stics were assumed to be the same for physicians providing patient encounter information and those not providing such information. However, research done for the fi rst time with 2003 data showed that these two assumptions are not always true. In general, the weekly visit volume for non-PRF physicians was larger than for PRF physicians. Also, physicians who saw no patients during their sample week tended to see patients fewer weeks annually than di d physicians who saw patients during their week. To minimize understatement (and in some cases, overstatement) of visits, the nonresponse adjustment factor was revised to include information on the number of weeks physicians actually practiced duri ng a typical year and the number of visits physicians reported during a week. Both data items were collected for responding and nonresponding physicians during the induction interview st arting with the 2001 survey. Th e 2003 weight with the revised non-response adjustment increased the overall visit estimate by 12 pe rcent over the same estimate obtained using the original weight. For this reason, 2003 and s ubsequent year visit estimates are not entirely comparable to visit estimates computed using the previous weighting strategy. If researchers are presenting data w ith estimates or rates across these years, we recommend including a footnote such as the following: 3. Ratio Adjustment A postratio adjustment was made within each of the fifteen physician specialty groups. The ratio adjustment is a multiplication factor which had as its numerat or the number of physicians in the universe in each physician specialty group and as its denominator the estimated number of physicians in that particular specialty group. The numerator was based on figures obtained from the AMA and AOA master files, and the denominator was base d on data from the sample. 4. Weight smoothing each year there are a few sample physicians whose final visit weights are large relative to those for the rest of the sample. There is a concern th at those few may adversely affect the ability of the resulting statistics to reflect the universe, especially if the sampled patient visits to some of those few physicians should be unusual relative to the universe. Extremes in final weights also increase the resulting variances. Extreme wei ghts can be truncated, but this leads to an understatement of the total vis it count. The technique of weight smoothing is used instead, because it preserves the total estimated visit count within each specialty by shifting the "excess" from visits with the largest weight s to visits with smaller weights. Excessively large visit weights were truncated, and a ratio adjustment was performed. The ratio adjustment is a multiplication factor that uses as its numerator the total visit count in each physician specialty group before the largest weights ar e truncated, and, as its denominator, the total visit count in th e same specialty group af ter the largest weights are truncated. The ratio adjustment was made within each of the fifteen p hysician specialty groups and yields the same estimated total visit count as the unsmoothed weights.
157 The NHAMCS: Scope and Sample Design The National Hospital Ambulatory Medical Care Survey (NHAMCS) was initiated to learn more about the ambulatory care rendered in hospital emergency and outpatient departments in the United States. Ambulatory medical care is th e predominant method of providing health care services in the United States (1). Since 1973, data on ambulatory patient visits to physicians' offices have been collected through the Nationa l Ambulatory Medical Care Survey (NAMCS). However, visits to hospital emergency and outpati ent departments, which represent a significant portion of total ambulatory medical care, are not included in the NAMCS (2). Furthermore, hospital ambulatory patients are known to diffe r from office patients in their demographic characteristics and medical aspects (3). Therefor e, the omission of hospital ambulatory care from the ambulatory medical care database leaves a signi ficant gap in coverage and limits the utility of the current NAMCS data. The NHAMCS fills th is data gap. The NHAMCS is endorsed by the Emergency Nurses Association, the Society for Emergency Academic Medicine, the American College of Emergency Physicians, and the American College of Osteopathic Emergency Physicians. The NHAMCS is conducted by the Am bulatory Care Statistics Branch of the National Center for Health Statistics, Cent ers for Disease Control and Prevention. The NHAMCS provides data from samples of patie nt records selected from the emergency departments (EDs) and outpatient departments (O PDs) of a national sample of hospitals. The national estimates produced from these studies de scribe the utilization of hospital ambulatory medical care services in the United States. SAMPLE DESIGN The NHAMCS was a national probability sample of visits to the emergency and outpatient departments of noninstitutional general and shortstay hospitals, exclusive of Federal, military, and Veterans Administration hospita ls, located in the 50 states and the District of Columbia. The NHAMCS was designed to provide estimates based on the following priority of survey objectives: United States; region; emergency and outpatient departme nts; and type of ownership. The NHAMCS used a four-stage probability design with samples of primary sampling units (PSUs), hospitals within PSUs, clinics/emergency service areas within outpatient/emergency departments, and patient visits within clinics/em ergency service areas. Each stage of sampling is described below. 1. Primary Sampling Units (PSUs): A PSU consists of a county, a group of counties, county equivalents (such as parishes and independent ci ties), towns, townships, minor civil divisions (for some PSUs in New England), or a metropolit an statistical area (MSA). MSAs were defined by the U.S. Office of Management and Budget on the basis of the 1980 Census. The first-stage sample consisted of 112 PSUs that comprised a probability subsample of the PSUs used in the 1985-94 National Health Interview Survey (NHI S). The NHAMCS PSU sample included with certainty the 26 NHIS PSUs with the largest populations. In additi on, the NHAMCS sample included one-half of the next 26 largest PSUs, and one PSU from each of the 73 PSU strata formed from the remaining PSUs for the NHIS sample. The NHIS PSU sample was selected from approximately 1,900 geographically define d PSUs that covered the 50 States and the District of Columbia. The 1,900 PSUs were stratified by socioeconomic and demographic variables and then select ed with a probability proportional to their size. Stratification was done within four geographical regi ons by MSA or non-MSA status. 2. Hospitals The sampling frame for the NHAMCS was c onstructed from products of Verispan L.L.C., specifically Healthcare Market Index and Hospital Market Pr ofiling Solution. These products were formerly known as the SMG Hospital Database. The original sample frame was
158 compiled as follows. Hospitals with an average lengt h of stay for all patients of less than 30 days (short-stay) or hospitals whose specialty was genera l (medical or surgical) or children's general were eligible for the NHAMCS. Excluded were Fe deral hospitals, hospital units of institutions, and hospitals with less than six beds staffed for patient use. In 1991, the SMG Hospital Database contained 6,249 hospitals that met these eligibility criteria. Of the eligible hospitals, 5,582 (89 percent) had emergency departments (EDs) and 5,654 (90 percent) had outpatient departments (OPDs). Hospitals were defined to have an ED if the hospital file indicat ed the presence of such a unit or if the file indicated a non-zero number of vi sits to such a unit. A similar rule was used to define the presence of an OPD. Hospitals were classified into four groups: those with only an ED; those with an ED and an OPD; those with only an OPD; and those with neither an ED nor an OPD. Hospitals in the last class were considered as a separate stratum and a small sample (50 hospitals) was selected from this stratum to allow for estimation to the total universe of eligible hospitals and the opening and clos ing of EDs and OPDs in the sample hospitals. All hospitals in non-certainty PSUs with five or fewer hospitals were selected with cer tainty. There were 149 hospitals in 55 PSUs in this ca tegory. In non-certainty PSUs with more than five hospitals, hospitals were stratified by hos pital class; type of ownershi p (not-for-profit, non-Federal government, and for-profit); and hospital size. Hospital size was measured by the combined volume of ED and OPD visits. From the stratified hospital list, fi ve hospitals were selected in each PSU with probability proportional to the numb er of ED and OPD patient visits. A total of 161 hospitals was selected from this group. In th e certainty PSUs, hospita ls were stratified by region, hospital class, ownership, and size. From the stratified hospital list, 240 hospitals were selected based on probability proportional to size A sample of 50 hospitals was selected from the 427 hospitals that had neither an ED nor an OPD. The hospital selections were made so that e ach hospital would be chosen only once to avoid multiple inclusions of very large hospitals. A fi xed panel of 600 hospitals was selected for the NHAMCS sample; 550 hospitals had an ED and/or an OPD and 50 hospitals had neither an ED nor an OPD. To preclude hospitals participa ting during the same time period each year, the sample of 600 hospitals was random ly divided into 16 subsets of approximately equal size. Each subset was assigned to 1 of the 16 4-week reporting periods, beginning December 2, 1991, which continues to rotate across each su rvey year. Therefore, the entire sample does not participate in a given year, and each hospital is inducted approximately once every 15 months. 3. Outpatient Clinics and Emergency Service Areas Within each hospital, either all outpatient clinics and emergency service areas (ESAs) or a sa mple of such units were selected. Clinics were in scope if ambulatory medical care was provided under the supe rvision of a physician and under the auspices of the hospital. Clin ics were required to be "organi zed" in the sense that services were offered at established locations and schedul es. Clinics where only an cillary services were provided or other settings in which physician se rvices were not typically provided were out of scope. In addition, freestanding clinics were out of scope since they are included in the NAMCS, and ambulatory surgery centers, whether in hospi tals or freestanding, were out of scope since they were included in the Nati onal Survey of Ambulatory Surgery which was conducted between 1994-96. The OPD clinic definition ex cluded the "hospital as landlor d" arrangement in which the hospital only rented space to a physician group and was not otherwis e involved in the delivery of services. These physicians are considered office-based and are currently included in the NAMCS. Emergency services provided under the hospital as landlord" arrangem ent, however, were eligible for the study. An emergency depa rtment was in scope if it was staffed 24 hours a day. If an in-scope emergency department had an emergency service area that was open less than
159 24 hours a day, then it was included under the emergency department. If a hospital had an emergency department that was staffed less th an 24 hours a day, then it was considered an outpatient clinic. Hospitals may define the term "separate clinic" differently, for example, by physical location within the hospital, by staff provi ding the services, by speci alty or subspecialty, by schedules, or by patients' source of payment. B ecause of these differences, "separate clinics" in the NHAMCS were defined as the smallest ad ministrative units for which the hospital kept patient volume statistics. During the visit by a field representative to induct a hospital into the survey, a list of all emergency service areas and outpatient clinics was obtained from the sample hospital. Each outpatient department clinic's f unction, specialty, and expected number of visits during the assigned reporti ng period were also collect ed. If there were five or fewer clinics, then all were included in the sample. If an outpatien t department had more than five clinics, the clinics were assigned into one of six specialty groups: general medicine, surgery, pediatrics, obstetrics/gynecology, substance a buse, and other. With in these specialty groups, clinics were grouped into clinic sampling units (SUs). A clinic sampling unit was generally one clinic, except when a clinic expected fewer than 30 visits. In that case, it was grouped with one or more other clinics to form a clinic SU. If the grouped SU wa s selected, all clinics included in that SU were included in the sample. Prior to 2001, a sample of generally five clinic SUs was selected per hospital based on probability propor tional to the total expected numb er of patient visits to the clinic during the assigned 4-week reporting peri od. Starting in 2001, clinic sampling within each hospital was stratified. If an OPD had more than five clinics, two clinic sample units were selected from each of the six specialty groups with a probability proportional to the total expected number of visits to th e clinic. The change was to ensure that at least two SUs were sampled from each of the specialty group strata. The emergency department was treated as a sepa rate stratum, and all emergency service areas were selected with certainty. In the rare inst ance that a sample hospital had more than five emergency service areas, a sample of five emer gency service areas was se lected with probability proportional to the expected number of visits to each emergency servi ce area during the assigned 4-week reporting period 4. Visits The basic sampling unit for the NHAMCS is th e patient visit or en counter. Only visits made in the United States by patients to EDs a nd OPDs of non-Federal, short-stay, or general hospitals were included in the 2005 NHAMCS. With in emergency service areas or outpatient department clinics, patient visits were systema tically selected over a ra ndomly assigned 4-week reporting period. A visit was defined as a direct personal exchange be tween a patient and a physician, or a staff member acting under a physicia n's direction, for the purpose of seeking care and rendering health services. Visits solely fo r administrative purposes, such as payment of a bill, and visits in which no medical care was provid ed, such as visits to deliver a specimen, were out of scope. The target numbers of Patient Record forms to be completed for EDs and OPDs in each hospital were 100 and 200, respectively. In clin ics with volumes higher than these desired figures, visits were sampled by a systematic procedure which selected every nth visit after a random start. Visit sampling rates were determined from the expected number of patients to be seen during the reporting period and the desired number of completed Patient Record forms. E. CONFIDENTIALITY In April 2003, the Privacy Rule of the Health Insurance Portability and Accountability Act (HIPAA) wa s implemented to establish minimum Federal standards for safeguarding the privacy of individually iden tifiable health information. No personally identifying information, such as patients name or address or Social Security number, is collected in the NHAMCS. Data collection is authorized by Section 306 of the Public Health Service Act
160 (Title 42, U.S. Code, 242k). All information coll ected is held in th e strictest confidence according to law [Section 308(d) of the Public Health Service Act (42, U.S. Code, 242m (d))] and the Confidential Information Protection and St atistical Efficiency Act (Title 5 of PL 107347). The NHAMCS protocol was approved by the NCHS Research Ethics Review Board in February 2003. Waivers of the requirements to obt ain informed consent of patients and patient authorization for release of patient medical record data by health care providers were granted. In the Spring of 2003, the NHAMCS implemented additional data collection procedures to help providers assure patient confid entiality. Census Bureau Field Representatives were trained on how the Privacy Rule allows hospitals to make disclosures of protected health information without patient authorization for public health purposes and for research that has been approved by a Research Ethics Review Board. Hospitals were encouraged to accept a data use agreement between themselves and NCHS/CDC, since the Privacy Rule allows hospitals to disclose limited data sets (i.e., data sets with no direct patient identifiers) for re search and public health purposes if such an agreement exists. Assurance of conf identiality was provided to all hospitals according to Section 308 (d) of the Public Health Serv ice Act (42 USC 242m). Strict procedures were utilized to prevent disclosure of NHAMCS data. All inform ation which could identify the hospital or its facilitie s was confidential and was seen onl y by persons engaged in the NHAMCS, and was not disclosed or released to others for any other purpo se. Names or other identifying information for individual patients were not removed from the hosp itals or individual facilities. Data users are advised that for some hospitals, se lected characteristics may have been masked to minimize the potential for disclosure. F ESTIMATION PROCEDURES Statistics from the NHAMCS were derived by a multistage estimation procedure that produces essentially un biased estimates. The estimation procedure has three basic components: 1) inflation by reciprocals of the sampling selection probabilities; 2) adjustment for nonresponse; a nd 3) a population weighting ratio adjustment. Beginning with 1997 data, the population weighting ratio adjustme nt for OPD estimates was replaced by an adjustment which controls for effects of rotati ng hospital sample panels into and out of the sample each year. (The full NHAMCS hospital sample is partitioned into 16 panels which are rotated into the sample over 16 periods of four w eeks each so that only 13 panels are used in each year.) 1. Inflation by reciprocals of selection probabilities There is one probability for each sampling stage: a) the probability of sel ecting the PSU; b) the probability of selecting the hospital; c) the probability of selecting the emergency service area (ESA) or OPD clinic from within the hospital; and d) the probability of selecting the visit within the ESA or clinic. The last probability is calculated to be the sample size from the ESA or clinic divided by the total number of visits occurring in that unit during that units data collection period. The overall probability of selection is the product of the probabilities at ea ch stage. The inverse of the overall selection probability is the basic inflation weight. Be ginning with the 1997 data the overall selection probabilities of some OPDs were permanently trimmed to prevent individual OPDs from contributing too much of their regions tota l to OPD visit estimates. 2. Adjustment for nonresponse NHAMCS data are adjusted to account for two types of nonresponse. The first type of nonresponse occurred when a sample hospital re fused to provide information about its ED(s) and/or OPD(s) which were publicly known to exist. In this case, the weights of visits to hospitals similar to the nonrespondent hospitals were infl ated to account for visits represented by the nonrespondent hospitals where hospitals were judged to be sim ilar if they were in the same region, ownership control group (government, nonFederal; voluntary non-profit; or
161 proprietary), and had the same metropolitan statistical area (M SA) status (that is, whether they were located in an MSA or not in an MSA). This adjustment was made separately by department type. The second type of nonresponse occurred wh en a sample ESA or OPD clinic within a respondent hospital failed to prov ide completed Patient Record form s for a sample of its patient visits. The weights of visits to ESAs/OPD clin ics similar to the nonrespondent ESAs/OPD clinics were inflated to account for visits represente d by the nonrespondent ESAs/OPD clinics where ESAs/OPD clinics were judged to be similar if they were in th e same region, ownership control group, MSA status group and ESA/OPD clinic group. For this purpose, there were six OPD clinic groups: general medicine, pediatrics, su rgery, OB/GYN, alcohol a nd/or substance abuse, and other. Beginning with 2004 data, changes were made to the nonresponse adjustment factor to account for the seasonality of the reporti ng period. Extra weights for nonresponding hospital outpatient departments and emergency department s were shifted to responding outpatient and emergency departments in reporting periods within the same quarter of the year. The shift in nonresponse adjustment did not significantly affect any of the overall annual estimates. 3. Ratio adjustments Adjustments were made within hos pital strata defined by region and by hospital ownership control groups. Within the Northeast, the Midwest and the South, the adjustment strata were further defined by MSA status. Th ese adjustments were made separately for emergency and outpatient departments. For EDs, the adjustment was a multiplicative factor that had as its numerator the sum of annual visit volumes reported to EDs in sampling frame hospitals in the stratum and as its denominator the estimat ed number of those visi ts for that stratum. Through the 1996 NHAMCS, the adjustment for vi sits to OPDs was a multiplicative factor which had as its numerator the number of OPDs reported in sampling frame hospitals in the stratum and as its denominator th e estimated number of those OPDs for that stratum. The data for the numerator and denominator of both adjust ments were based on figures recorded for the data year in the Verispan Hospital Database Beginning with the 1997 NHAMCS, the adjustment for OPD estimates was replaced by a ratio which had as its numerator the weighted OPD visit volumes of hospitals in the full NHAMCS sample (16 hospital panels) and as its denominator the weighted OPD visit volumes of hosp itals in the 13 hospital panels in cluded in that years sample. This adjustment used visit volumes that were based on the most recent survey data collected from hospitals that had participated in the NHA MCS for at least one year. For hospitals which had never participated, visit volumes were obt ained by phone, from Verispan data, or by using the average of visit volumes for refusal hospitals which had convert ed to respondent status in the 2002 survey.
162 APPENDIX B SEARCH STRATEGY FOR NEW ANTICONVULSANTSOFF-LABEL USE PUBLIC ATIONS (felbamate OR fosphenytoin OR gabapen tin OR lamotrigine OR levetiracetam OR oxcarbazepine OR pregabalin OR tiagabine OR topiramate OR zonisamide) AND ( ADHD OR bipolar OR "drug dependence" OR "peripheral ne uropathy" OR fibromyalgia OR "hot sweats" OR migraine OR neuropathy OR nystagmus OR "o rthostatic tremor" OR "r estless leg syndrome" OR "social phobia" OR "tardive dyskinesia" OR "essential tremor" OR "multiple sclerosis" OR "trigeminal neuralgia" OR "alc ohol dependence" OR "bulimia ne rvosa" OR "essential tremor" OR obesity OR "west syndrome" OR "smoking cessation" OR "parkinson disease" OR agitation OR dementia OR "mental retard ation" OR chorea OR depression OR "neurogenic pain" OR pain OR "psychotic disorder" OR hyperbilirubinemia OR "cardiac arrhythmia" OR "decubitus ulcer" OR eclampsia OR tremor OR myelodysplastic OR "neuropathic pain" OR "herpes zoster" OR arthritis OR gout OR arthropathy OR rheumatoid OR osteoarthrosis OR "joint disorder" OR dorsopathies OR "spina bifida" OR "renal colic" OR sprains OR strains OR "attention deficit hperactivity disorder" OR anxiety OR menopaus e OR "panic disorder" OR "post traumatic stress disorder" OR "borderline personality disorder" OR headach e OR "impulsive aggressive disorder" OR "binge eating disorder" OR psoriasi s OR "alzheimer's disease" OR "hiccups" OR "schizoaffective disorder") AND (("2007/01/01"[PDat]:"2007/12/31"[PDat]) AND (Humans[Mesh]) AND (English[lang]) AND (Clinical Trial[ptyp] OR Meta -Analysis[ptyp] OR Randomized Controlled Trial[ptyp] OR Review[ptyp] OR Case Reports[ptyp]))
163 APPENDIX C ICD-9-CODES FOR ON-LABEL INDICATIONS OF AE DS Table B-1. ICD-9 Codes for On-Label Indications from 1993 to 1999 1993 1994 1995 1996 1997 1998 1999 CBZ 345.xx, 780.3780.39, 350.1, 352.1 345.xx, 780.3780.39, 350.1, 352.1 345.xx, 780.3780.39, 350.1, 352.1 345.xx, 780.3780.39, 350.1, 352.1 345.xx, 780.3780.39, 350.1, 352.1 345.xx, 780.3780.39, 350.1, 352.1 345.xx, 780.3780.39, 350.1, 352.1 DS 345.xx, 780.3780.39 345.xx, 780.3780.39 345.xx, 780.3780.39, 296.00-296.06, 296.10-296.16, 296.40-296.46, 296.50-295.56, 296.60-296.66, 297.7, 296.80, 296.81, 296.82, 296.89 345.xx, 780.3780.39, 296.00296.06, 296.10296.16, 296.40296.46, 296.50295.56, 296.60296.66, 297.7, 296.80, 298.81, 298.82, 298.89, 346-346.91 345.xx, 780.3780.39, 296.00296.06, 296.10296.16, 296.40296.46, 296.50295.56, 296.60296.66, 297.7, 296.80, 298.81, 298.82, 298.89, 346-346.91 345.xx, 780.3780.39, 296.00296.06, 296.10296.16, 296.40296.46, 296.50295.56, 296.60296.66, 297.7, 296.80, 298.81, 298.82, 298.89, 346-346.91 345.xx, 780.3780.39, 296.00296.06, 296.10296.16, 296.40296.46, 296.50295.56, 296.60296.66, 297.7, 296.80, 298.81, 298.82, 298.89, 346-346.91 ESM 345.xx, 780.3780.39 345.xx, 780.3780.39 345.xx, 780.3780.39 345.xx, 780.3780.39 345.xx, 780.3780.39 345.xx, 780.3780.39 345.xx, 780.3780.39 MSM 345.xx, 780.3-780.39 345.xx, 780.3780.39 345.xx, 780.3780.39 345.xx, 780.3780.39 345.xx, 780.3780.39 345.xx, 780.3780.39 345.xx, 780.3780.39 PB 345.xx, 780.3780.39, 300.00300.09, 307.41, 307.42, 780.51, 780.52, 345.xx, 780.3780.39, 300.00300.09, 307.41, 307.42, 780.51, 780.52, 345.xx, 780.3780.39, 300.00300.09, 307.41, 307.42, 780.51, 780.52, 345.xx, 780.3780.39, 300.00300.09, 307.41, 307.42, 780.51, 780.52, 345.xx, 780.3780.39, 300.00300.09, 307.41, 307.42, 780.51, 780.52, 345.xx, 780.3780.39, 300.00300.09, 307.41, 307.42, 780.51, 780.52, 345.xx, 780.3780.39, 300.00300.09, 307.41, 307.42, 780.51, 780.52, PHT 345.xx, 780.3780.39 345.xx, 780.3780.39 345.xx, 780.3780.39 345.xx, 780.3780.39 345.xx, 780.3780.39 345.xx, 780.3780.39 345.xx, 780.3780.39
164 Table B-1 CONTINUED 1993 1994 1995 1996 1997 1998 1999 PRM 345.xx, 780.3780.39 345.xx, 780.3780.39 345.xx, 780.3780.39 345.xx, 780.3780.39 345.xx, 780.3780.39 345.xx, 780.3780.39 345.xx, 780.3780.39 VPA 345.xx, 780.3780.39 345.xx, 780.3780.39 345.xx, 780.3780.39, 296.00296.06, 296.10296.16, 296.40296.46, 296.50295.56, 296.60296.66, 297.7, 296.80, 296.81, 296.82, 296.89 345.xx, 780.3780.39, 296.00296.06, 296.10296.16, 296.40296.46, 296.50295.56, 296.60296.66, 297.7, 296.80, 298.81, 298.82, 298.89, 346-346.91 345.xx, 780.3780.39, 296.00296.06, 296.10296.16, 296.40296.46, 296.50295.56, 296.60296.66, 297.7, 296.80, 298.81, 298.82, 298.89, 346-346.91 345.xx, 780.3780.39, 296.00296.06, 296.10296.16, 296.40296.46, 296.50295.56, 296.60296.66, 297.7, 296.80, 298.81, 298.82, 298.89, 346-346.91 345.xx, 780.3780.39, 296.00296.06, 296.10296.16, 296.40296.46, 296.50295.56, 296.60296.66, 297.7, 296.80, 298.81, 298.82, 298.89, 346-346.91 FBM NA 345.xx, 780.3780.39 345.xx, 780.3780.39 345.xx, 780.3780.39 345.xx, 780.3780.39 345.xx, 780.3780.39 345.xx, 780.3780.39 FPHT NA NA NA NA 345.xx, 780.3780.39 345.xx, 780.3780.39 345.xx, 780.3780.39 GBP NA NA 345.xx, 780.3780.39 345.xx, 780.3780.39 345.xx, 780.3780.39 345.xx, 780.3780.39 345.xx, 780.3780.39 LTG NA NA NA 345.xx, 780.3780.39 345.xx, 780.3780.39 345.xx, 780.3780.39 345.xx, 780.3780.39 LEV NA NA NA NA NA NA NA OXC NA NA NA NA NA NA NA PGB NA NA NA NA NA NA NA TGB NA NA NA NA NA 345.xx, 780.3780.39 345.xx, 780.3780.39 TPM NA NA NA NA NA 345.xx, 780.3780.39 345.xx, 780.3780.39 ZNS NA NA NA NA NA NA NA
165 Table B-2. ICD-9 Codes for On-Label Indications from 2000 to 2005 2000 2001 2002 2003 2004 2005 CBZ 345.xx, 780.3780.39, 350.1, 352.1 345.xx, 780.3780.39, 350.1, 352.1 345.xx, 780.3780.39, 350.1, 352.1 345.xx, 780.3780.39, 350.1, 352.1 345.xx, 780.3-780.39, 350.1, 352.1 345.xx, 780.3780.39, 350.1, 352.1 DS 345.xx, 780.3780.39, 296.00296.06, 296.10296.16, 296.40296.46, 296.50295.56, 296.60296.66, 297.7, 296.80, 298.81, 298.82, 298.89, 346346.91 345.xx, 780.3780.39, 296.00296.06, 296.10296.16, 296.40296.46, 296.50295.56, 296.60296.66, 297.7, 296.80, 298.81, 298.82, 298.89, 346-346.91 345.xx, 780.3780.39, 296.00296.06, 296.10296.16, 296.40296.46, 296.50295.56, 296.60296.66, 297.7, 296.80, 298.81, 298.82, 298.89, 346-346.91 345.xx, 780.3780.39, 296.00296.06, 296.10296.16, 296.40296.46, 296.50295.56, 296.60296.66, 297.7, 296.80, 298.81, 298.82, 298.89, 346346.91 345.xx, 780.3-780.39, 296.00-296.06, 296.10-296.16, 296.40-296.46, 296.50-295.56, 296.60-296.66, 297.7, 296.80, 298.81, 298.82, 298.89, 346346.91 345.xx, 780.3780.39, 296.00296.06, 296.10296.16, 296.40296.46, 296.50295.56, 296.60296.66, 297.7, 296.80, 298.81, 298.82, 298.89, 346-346.91 ESM 345.xx, 780.3780.39 345.xx, 780.3780.39 345.xx, 780.3780.39 345.xx, 780.3780.39 345.xx, 780.3-780.39 345.xx, 780.3780.39 MSM 345.xx, 780.3780.39 345.xx, 780.3780.39 345.xx, 780.3780.39 345.xx, 780.3780.39 345.xx, 780.3-780.39 345.xx, 780.3780.39 PB 345.xx, 780.3780.39, 300.00300.09, 307.41, 307.42, 780.51, 780.52, 345.xx, 780.3780.39, 300.00300.09, 307.41, 307.42, 780.51, 780.52, 345.xx, 780.3780.39, 300.00300.09, 307.41, 307.42, 780.51, 780.52, 345.xx, 780.3780.39, 300.00300.09, 307.41, 307.42, 780.51, 780.52, 345.xx, 780.3-780.39, 300.00-300.09, 307.41, 307.42, 780.51, 780.52, 345.xx, 780.3780.39, 300.00300.09, 307.41, 307.42, 780.51, 780.52, PHT 345.xx, 780.3780.39 345.xx, 780.3780.39 345.xx, 780.3780.39 345.xx, 780.3780.39 345.xx, 780.3-780.39 345.xx, 780.3780.39 PRM 345.xx, 780.3780.39 345.xx, 780.3780.39 345.xx, 780.3780.39 345.xx, 780.3780.39 345.xx, 780.3-780.39 345.xx, 780.3780.39
166 Table B-2 CONTINUED VPA 345.xx, 780.3780.39, 296.00296.06, 296.10296.16, 296.40296.46, 296.50295.56, 296.60296.66, 297.7, 296.80, 298.81, 298.82, 298.89, 346346.91 345.xx, 780.3780.39, 296.00296.06, 296.10296.16, 296.40296.46, 296.50295.56, 296.60296.66, 297.7, 296.80, 298.81, 298.82, 298.89, 346346.91 345.xx, 780.3780.39, 296.00296.06, 296.10296.16, 296.40296.46, 296.50295.56, 296.60296.66, 297.7, 296.80, 298.81, 298.82, 298.89, 346-346.91 345.xx, 780.3780.39, 296.00296.06, 296.10296.16, 296.40296.46, 296.50295.56, 296.60296.66, 297.7, 296.80, 298.81, 298.82, 298.89, 346346.91 345.xx, 780.3-780.39, 296.00-296.06, 296.10296.16, 296.40-296.46, 296.50-295.56, 296.60296.66, 297.7, 296.80, 298.81, 298.82, 298.89, 346-346.91 345.xx, 780.3-780.39, 296.00-296.06, 296.10-296.16, 296.40-296.46, 296.50-295.56, 296.60-296.66, 297.7, 296.80, 298.81, 298.82, 298.89, 346346.91 FBM 345.xx, 780.3780.39 345.xx, 780.3780.39 345.xx, 780.3780.39 345.xx, 780.3-780.39 345.xx, 780.3-780.39 345.xx, 780.3-780.39 FPHT 345.xx, 780.3780.39 345.xx, 780.3780.39 345.xx, 780.3780.39 345.xx, 780.3-780.39 345.xx, 780.3-780.39 345.xx, 780.3-780.39 GBP 345.xx, 780.3780.39 345.xx, 780.3780.39 345.xx, 780.3780.39, 053.12053.14, 053.19 345.xx, 780.3780.39, 053.12053.14, 053.19 345.xx, 780.3-780.39, 053.12-053.14, 053.19 345.xx, 780.3-780.39, 053.12-053.14, 053.19 LTG 345.xx, 780.3780.39 345.xx, 780.3780.39 345.xx, 780.3780.39 345.xx, 780.3780.39, 296.00296.06, 296.10296.16, 296.40296.86, 345.xx, 780.3-780.39, 296.00-296.06, 296.10296.16, 296.40-296.86, 345.xx, 780.3-780.39, 296.00-296.06, 296.10-296.16, 296.40-296.86, LEV 345.xx, 780.3780.39 345.xx, 780.3780.39 345.xx, 780.3780.39 345.xx, 780.3-780.39 345.xx, 780.3-780.39 345.xx, 780.3-780.39 OXC 345.xx, 780.3780.39 345.xx, 780.3780.39 345.xx, 780.3780.39 345.xx, 780.3-780.39 345.xx, 780.3-780.39 345.xx, 780.3-780.39 PGB NA NA NA NA 345.xx, 780.3-780.39, 250.60, 357.20, 053.12, 053.13, 729.10 345.xx, 780.3-780.39, 250.60, 357.20, 053.12, 053.13, 729.10 TGB 345.xx, 780.3780.39 345.xx, 780.3780.39 345.xx, 780.3780.39 345.xx, 780.3-780.39 345.xx, 780.3-780.39 345.xx, 780.3-780.39 TPM 345.xx, 780.3780.39 345.xx, 780.3780.39 345.xx, 780.3780.39 345.xx, 780.3-780.39 345.xx, 780.3-780.39, 346-346.91 345.xx, 780.3-780.39, 346-346.91 ZNS 345.xx, 780.3780.39 345.xx, 780.3780.39 345.xx, 780.3780.39 345.xx, 780.3-780.39 345.xx, 780.3-780.39 345.xx, 780.3-780.39
167 APPENDIX D SAS CODE FOR UNIVERSAL OFFLABEL I NDICATIONS OF AEDS PROC SQL ; UPDATE NAM4.ALLNAM3 SET DIAG1_TYPE= 1 WHERE DIAG1R BETWEEN 132700 AND 132780 /*ORGANIC SLEEP DISORDERS*/ OR DIAG1R BETWEEN 133200 AND 133210 /*PARKINSON'S DISEASE*/ OR DIAG1R BETWEEN 133300 AND 133399 /*OTHER EXTRAPYRAMIDAL DISEASE*/ OR DIAG1R BETWEEN 133400 AND 133499 /*SPINOCEREBELLAR DISEASE*/ OR DIAG1R BETWEEN 133500 AND 133599 /*ANTERIOR HORN DISEASE*/ OR DIAG1R BETWEEN 133600 AND 133699 /*OTHER DISEASES OF THE SPINAL CORD*/ OR DIAG1R BETWEEN 133700 AND 133799 /*DISORDERS OF THE AUTONOMIC NERVIOS SYSTEM*/ OR DIAG1R BETWEEN 134000 AND 134009 /*MUITIPLE SCLEROSIS*/ OR DIAG1R BETWEEN 134100 AND 134190 /*OTHER DEMYLINATING DISEASES OF CENTRAL NERVOUS SYSTEM*/ OR DIAG1R BETWEEN 134200 AND 134290 /*HEMIPLEGIA AND HEMIPARESIS*/ OR DIAG1R BETWEEN 134300 AND 124390 /*INFATILE CEREBRAL PALSY*/ OR DIAG1R BETWEEN 134400 AND 134490 /*OTHER PARALYTIC SYNDROMES*/ OR DIAG1R BETWEEN 134500 AND 134599 /*EPILEPSY AND RE CURRENT SEIZURES*/ OR DIAG1R BETWEEN 178030 AND 178039 /*CONVULSIONS*/ OR DIAG1R BETWEEN 134600 AND 134690 /*MIGRAINE*/ OR DIAG1R BETWEEN 134700 AND 134790 /*CATAPLEXY AND NACROPLEXY*/ OR DIAG1R BETWEEN 134800 AND 134880 /*OTHER CONDITIONS OF THE BRAIN*/ OR DIAG1R BETWEEN 134900 AND 134990 /*OTHER AND UNSPECIFIED CONDITIONS OF THE BRAIN*/ OR DIAG1R BETWEEN 135000 AND 135090 /*TRIGEMINAL NERVE DISORDRS*/ OR DIAG1R BETWEEN 135100 AND 135190 /*FACIAL NERVE DISORDERS*/ OR DIAG1R BETWEEN 135200 AND 135290 /*DISORDERS OF OTHER CRANIAL NERVES*/ OR DIAG1R BETWEEN 135300 AND 135390 /*NERVE ROOT AND PLEXUS DISORDERS*/ OR DIAG1R BETWEEN 135400 AND 135490 /*MONONEURITIS OF UPPER LIMB AND MONONEURITIS MULTIPLEX*/ OR DIAG1R BETWEEN 135500 AND 135590 /*MONONEURITIS OF LOWER LIMB AND UNSPECIFIED SITES*/ OR DIAG1R BETWEEN 135600 AND 135690 /*HEREDIATARY AND IDOPATHIC PERIPHERAL NEUROPATHY*/ OR DIAG1R BETWEEN 135700 AND 135790 /*INFLAMMATORY AND TOXIC NEUROPATHY*/ OR DIAG1R BETWEEN 135800 AND 135890 /*MYONEURAL DISORDERS*/ OR DIAG1R BETWEEN 137950 AND 137956 /*NYSTAGMUS*/ OR DIAG1R BETWEEN 139200 AND 139299 /*RHEUMATIC CHOREA*/ OR DIAG1R BETWEEN 141100 AND 141189 /*ANTISCHEMIC EFFECTS*/ OR DIAG1R BETWEEN 142700 AND 142790 /*ARRHYTHMIAS*/ OR DIAG1R BETWEEN 164240 AND 164260 /* PRE/ECLAMPSIA*/ OR DIAG1R BETWEEN 198000 AND 198098 /*TOXIC EFFECTS OF ALCOHOL*/ OR DIAG1R BETWEEN 105300 AND 105390 /*POSTHERPTIC NEURALGIA AND POLYNEUROPATHY*/ OR DIAG1R BETWEEN 125060 AND 125069 /*DIABETES WITH NEUROLOGICAL MANIFESTATIONS*/ OR DIAG1R BETWEEN 132000 AND 138900 OR DIAG1R BETWEEN 178000 AND 178469 OR DIAG1R BETWEEN 179300 AND 179419 OR DIAG1R BETWEEN 201240 AND 201249 OR DIAG1R BETWEEN 204840 AND 204850 OR DIAG1R BETWEEN 178030 AND 178039 OR DIAG1R = 204520 OR DIAG1R = 204930 OR DIAG1R = 205302
168 OR DIAG1R = 205309 OR DIAG1R = 172920 /*NEURALGIA, NEURITIS AND RADICULITIS, UNSPECIFIED*/ OR DIAG1R = 195880 /*OTHER EARLY COMPLICATIONS OF TRAUMA*/ OR DIAG1R = 190860 /*LATE EFFECTS OF UNSPECIFIED INJURY*/ OR DIAG1R = 172885 /*SPASM OF MUSCLE*/ QUIT ; PROC SQL ; UPDATE NAM4.ALLNAM3 SET DIAG1_TYPE= 2 WHERE DIAG1R BETWEEN 129000 AND 129090 /*DEMENTIAS*/ OR DIAG1R BETWEEN 129100 AND 129190 /*ALCOHOL INDUCED MENTAL DISORDER*/ OR DIAG1R BETWEEN 129200 AND 129290 /*DRUG INDUCED MENTAL DISORDERS*/ OR DIAG1R BETWEEN 129300 AND 129390 /*TRANSIENT MENTAL DISORDERS DURE TO CONDITION CLASSIFED ELSEWHERE*/ OR DIAG1R BETWEEN 129400 AND 129490 /*PERSISTENT MENTAL DISORDERS DUR TO CONDITIONS CLASSIFIED ELSEWHERE*/ OR DIAG1R BETWEEN 129500 AND 129590 /*SCHIZOPHRENIC DISORDERS*/ OR DIAG1R BETWEEN 129600 AND 129690 /*EPISODIC MOOD DISORDERS*/ OR DIAG1R BETWEEN 129700 AND 129790 /*DELUSIONAL DISORDERS*/ OR DIAG1R BETWEEN 129800 AND 129890 /*OTHER NONORGANIC PSYCHOSES*/ OR DIAG1R BETWEEN 129900 AND 129990 /*PERVASIVE DEVELOPMENTAL DISORDERS*/ OR DIAG1R BETWEEN 130000 AND 130019 /*ANXIETY STATES, DISSOIATIVE, CONVERSION AND FACTITIOUS DISORDERS*/ OR DIAG1R BETWEEN 130020 AND 130029 /*PHOBIC DISORDERS*/ OR DIAG1R BETWEEN 130030 AND 130090 /*OCD, DYSTHYMIC, NEURASTHENIA, DEPERSONALIZATION, HYPOCHONDRIAIS AND SOMATOFORM DISORDER*/ OR DIAG1R BETWEEN 130100 AND 130189 /*PERSONALITY DISORDERS*/ OR DIAG1R BETWEEN 130200 AND 130290 /*PSYCHOSEXUAL DYSFUNCTION*/ OR DIAG1R BETWEEN 130300 AND 130390 /*ALCOHOL DEPENDENCE SYNDROME*/ OR DIAG1R BETWEEN 130400 AND 130490 /*DRUG DEPENDENCE*/ OR DIAG1R BETWEEN 130500 AND 130590 /*NONDEPENDENT ABUSE OF DRUGS*/ OR DIAG1R BETWEEN 130600 AND 130690 /*PHYSIOLOGICAL MALFUNCTION ARISING FROM MENTAL FACTORS*/ OR DIAG1R BETWEEN 130700 AND 130790 /*SPECIAL SYSMPTOMS OR SYNDROMES, NOT ELSEWHERE CLASSIFIED*/ OR DIAG1R BETWEEN 130800 AND 130890 /*ACUTE REACTION TO STRESS*/ OR DIAG1R BETWEEN 130900 AND 130990 /*ADJUSTMENT DISORDER*/ OR DIAG1R BETWEEN 131000 AND 131090 /*SPECIFIC NONPSYCHOTIC MENTAL DISORDERS DUE TO BRAIN DAMAGE*/ OR DIAG1R BETWEEN 131100 AND 131199 /*DEPRESSIVE DISORDER, NOT CLASSIFIED ELSEWHERE*/ OR DIAG1R BETWEEN 131200 AND 131290 /*DISTURBANCE OF CONDUCT, NOT ELSEWHERE CLASSIFIED*/ OR DIAG1R BETWEEN 131300 AND 131390 /*DISTURBANCE OF EMOTION SPECIFIC TO CHILDHOOD AND ADOLESCENCE*/ OR DIAG1R BETWEEN 131400 AND 131490 /*HYPERKINETIC SYNDROME OF CHILDHOOD*/ OR DIAG1R BETWEEN 131500 AND 131599 /*SPECIFIC DELAYS IN DEVELOPMENT*/ OR DIAG1R BETWEEN 131600 AND 131690 /*PSYCHIC FACTORS ASSOCIATED WITH DISEASE CLASSIFEID ELSEWHERE*/ OR DIAG1R BETWEEN 131700 AND 131799 /* MILD MENTAL RETARDATION*/ OR DIAG1R BETWEEN 131800 AND 131899 /*OTHER SPECIFIED MENTAL RETARDATION*/ OR DIAG1R BETWEEN 131900 AND 131999 /* OTHER MENTAL RETARDATION*/ OR DIAG1R BETWEEN 201100 AND 201199 /*PERSONAL HISTORY OF MENTAL DISORDERS*/ OR DIAG1R BETWEEN 201540 AND 201549 /*PERSONAL HISTORY PSYCHOLOGICAL TRAUMA*/ OR DIAG1R BETWEEN 129000 AND 131900
169 OR DIAG1R BETWEEN 164830 AND 164835 OR DIAG1R BETWEEN 165550 AND 165553 OR DIAG1R BETWEEN 176072 AND 176075 OR DIAG1R BETWEEN 196500 AND 196509 OR DIAG1R BETWEEN 201100 AND 201190 OR DIAG1R BETWEEN 201540 AND 201549 OR DIAG1R BETWEEN 207010 AND 207990 OR DIAG1R BETWEEN 157100 AND 157130 OR DIAG1R BETWEEN 164840 AND 164844 OR DIAG1R BETWEEN 204020 AND 204030 OR DIAG1R = 177950 OR DIAG1R = 201582 OR DIAG1R = 201582 OR DIAG1R = 206630 OR DIAG1R = 142550 OR DIAG1R = 153530 OR DIAG1R = 176070 OR DIAG1R = 179030 OR DIAG1R = 201130 OR DIAG1R = 207910 OR DIAG1R = 204090 OR DIAG1R = 206730 quit ; PROC SQL ; UPDATE NAM4.ALLNAM3 SET DIAG1_TYPE= 3 WHERE DIAG1R = 137991 /*PAIN IN OR AROUND THE EYE*/ OR DIAG1R = 156942 /*ANAL OR RECTAL PAIN*/ OR DIAG1R = 172410 /*PAIN IN THE THORAIC SPINE*/ OR DIAG1R = 172420 /*LUBBAGO OR LOW BACK PAIN LOW BACK SYNDROME LUMBIAGIA*/ OR DIAG1R = 172950 /*PAIN IN THE LIMB*/ OR DIAG1R = 178410 /*THROAT PAIN*/ OR DIAG1R = 178730 /*FLATULENCE ERUCTATION AND GAS PAIN*/ OR DIAG1R BETWEEN 114000 AND 119599 /*MALIGNANT NEOPLASMS, STATED OR PRESUMED TO BE THE PRIMARY EXCEPT OF LYMPHATIC AND HEMATOPOIETIC TSSUE*/ OR DIAG1R BETWEEN 119600 AND 119899 /*MALIGNANT NEOPLASMS, STATED OR PRESUMED TO BE SECONDARY EXCEPT OF LYMPHATIC AND HEMATOPOIETIC TSSUE*/ OR DIAG1R BETWEEN 119900 AND 119999 /*MALIGNANT NEOPLASM, WITHOUS SITE SPECIFICATIOIN*/ OR DIAG1R BETWEEN 120000 AND 120899 /*MALIGNANT NEOPLASM, STATED OR PRESUMED TO BE DPRIMARY OF LYMPHATIC AND HEMATOPOIETIC TSSUE*/ OR DIAG1R BETWEEN 127400 AND 127490 /*ARTHRITIS GOUT*/ OR DIAG1R BETWEEN 133800 AND 133899 /*PAIN, NOT ELSEWH ERE CLASSIFIED*/ OR DIAG1R BETWEEN 138871 AND 138872 /*OTALGIA PAIN AND REFERED PAIN*/ OR DIAG1R BETWEEN 171000 AND 171099 /*DIFFUSE DX OF THE CONNECTIVE TISSUE*/ OR DIAG1R BETWEEN 171100 AND 171199 /*ARTHOPATY UNSPECIFIED*/ OR DIAG1R BETWEEN 171500 AND 171599 /*OSTEOARTHROSIS AND ALLIED DISORDERS*/ OR DIAG1R BETWEEN 171600 AND 171699 /*TRAUMATIC ATHROPATHY/OTHER INFLAMMATORY POLYARTHROPATHIES*/ OR DIAG1R BETWEEN 171700 AND 171799 /*RHEUMATOID ARTHRITIS AND OTHER INFLAMMATORY POLYARTHROPATHIES*/ OR DIAG1R BETWEEN 171900 AND 171999 /*PAIN IN THE JOINT*/ OR DIAG1R BETWEEN 172000 AND 172499 /*DORSOPATHIES*/ OR DIAG1R BETWEEN 172500 AND 172999 /*RHEUMATISM, EXCLUDING THE BACKE*/ OR DIAG1R BETWEEN 173700 AND 173799 /*CURVATURE OF THE SPINE*/
170 OR DIAG1R BETWEEN 173800 AND 173899 /*OTHER ACQUIRED DEFORMITIES*/ OR DIAG1R BETWEEN 173900 AND 173999 /*NONALLOPATHIC LESIONS NOT ELSEWHERE CLASSIFIED*/ OR DIAG1R BETWEEN 174100 AND 174299 /*SPINAL BIFIDA AND OTHER SPECIFIED ANOMALIES OF SPINAL CORD*/ OR DIAG1R BETWEEN 178650 AND 178659 /*CHEST PAIN*/ OR DIAG1R BETWEEN 178000 AND 178099 /*GENERAL SYMPTOMS*/ OR DIAG1R BETWEEN 178100 AND 178199 /*SYMPTOMS INVOLVING NERVOUS AND MUSCULOSKELETAL SYSTEM*/ OR DIAG1R BETWEEN 178300 AND 178330 /*SYMPTOMS CONCERNING NUTRITION, METABOLISM AND DEVELOPMENT*/ OR DIAG1R BETWEEN 178400 AND 178410 /*HEADACH AND THROAT PAIN*/ OR DIAG1R BETWEEN 178900 AND 178909 /*ABDOMINAL PAIN*/ OR DIAG1R BETWEEN 184000 AND 184899 /*SPRAINS AND STRAINS OF JOINTS AND ADJACENT MUSCLES*/ OR DIAG1R BETWEEN 189500 AND 189799 /*TRAUMATIC AMPUTATION*/ OR DIAG1R BETWEEN 195000 AND 195799 /*INJURY TO NERVES AND SPINAL CORD*/ OR DIAG1R BETWEEN 199760 AND 199769 /*AMPUTATIOIN STUMP COMPLICATIONS*/ OR DIAG1R BETWEEN 201000 AND 201099 /*PERSONAL HISTORY OF MALIGNANT NEOPLASM*/ ; quit; \
171 LIST OF REFERENCES 1. Erickson PG. The law, social control, and drug policy: models, factors, and processes. Int.J Addict. 10 1993;28(12):1155-1176. 2. MacCoun RJ. Drugs and the law: a psyc hological analysis of drug prohibition. Psychol.Bull. 05 1993;113(3):497-512. 3. Office USGA. Off-label Drugs: initial re sults of a national survey 2007/05/10 1991. 4. Brosgart CL, Mitchell T, Charlebois E, et al. Off-label drug use in human immunodeficiency virus disease. J Acquir.Immune.Defic.Syndr.Hum.Retrovirol. 05/01/ 1996;12(1):56-62. 5. Unapproved uses of approved drugs: the phys ician, the package insert, and the Food and Drug Administration: subject review. American Academy of Pediatrics Committee on Drugs. Pediatrics. 07 1996;98(1):143-145. 6. Radley DC, Finkelstein SN Stafford RS. Off-label prescribing among office-based physicians. Arch Intern Med. 05/08/ 2006;166(9):1021-1026. 7. Chen H, Deshpande AD, Jiang R, Martin BC. An epidemiological investigation of offlabel anticonvulsant drug use in the Georgia Medicaid population. Pharmacoepidemiol.Drug Saf. 09 2005;14(9):629-638. 8. Graham DJ, Green L. Further cases of valvular heart disease associated with fenfluramine-phentermine. N.Engl.J Med. 08/28/ 1997;337(9):635. 9. Mark EJ, Patalas ED, Chang HT, Evans RJ, Kessler SC. Fatal pulmonary hypertension associated with short-term use of fenfluramine and phentermine. N.Engl.J Med. 08/28/ 1997;337(9):602-606. 10. Trifiro G, Verhamme KM, Ziere G, Capu ti AP, Ch Stricker BH Sturkenboom MC. Allcause mortality associated with atypical and typical antipsychotics in demented outpatients. Pharmacoepidemiol.Drug Saf. 05 2007;16(5):538-544. 11. Wang PS, Schneeweiss S, Avorn J, et al. Risk of death in elderly users of conventional vs. atypical antipsychotic medications. N.Engl.J Med. 12/01/ 2005;353(22):2335-2341. 12. Choonara I, Conroy S. Unlicensed and off-la bel drug use in children: implications for safety. Drug Saf. 2002;25(1):1-5. 13. Gill AM, Leach HJ, Hughes J, Barker C, N unn AJ, Choonara I. Adverse drug reactions in a paediatric intensive care unit. Acta Paediatr. 04 1995;84(4):438-441. 14. Horen B, Montastruc JL, Lapeyre-Mestre M. Adverse drug reactions and off-label drug use in paediatric outpatients. Br J Clin.Pharmacol. 12 2002;54(6):665-670.
172 15. Turner S, Nunn AJ, Fielding K, Choonara I. Adverse drug reactions to unlicensed and off-label drugs on paediatric wards: a prospective study. Acta Paediatr. 09 1999;88(9):965-968. 16. Beck JM, Azari ED. FDA, off-label use, and informed consent: Debunking myths and misconceptions. Food and Drug Law Journal. 1998;53(1):71-104. 17. Barbui C, Danese A, Guaiana G, et al. Prescribing second-generation antipsychotics and the evolving standard of care in Italy. Pharmacopsychiatry. 11 2002;35(6):239-243. 18. Prescription Drugs: Implications of Drugs Labeling and Off-label Use. Government Reform and Oversight, U.S. House of Representative ; 1996. 19. O'Reilly J, Dalal A. Off-label or out of bounds? Prescriber and marketer liability for unapproved uses of FDA-approved drugs. Ann Health Law. 2003;12(2):295-324, table. 20. Kocs D, Fendrick AM. Effect of off-label use of oncology drugs on pharmaceutical costs: the rituximab experience. Am.J Manag.Care. 05 2003;9(5):393-400. 21. Steinman MA, Bero LA, Chren MM, Landefeld CS. Narrative review : the promotion of gabapentin: an analysis of internal industry documents. Ann Intern Med. 08/15/ 2006;145(4):284-293. 22. Tabarrok A. Assessubg the FDA via the anomaly of off-label drug prescribing. 6/6/2000 2000. 23. Vivian J. Off-label us e of prescription drugs US Pharmacist. Vol 28: Jobson; 2003. 24. DiMasi JA, Kaitin KI, Fernandez-Carol C, Lasagna L. New indications for alreadyapproved drugs: an analysis of regulatory review times. J Clin Pharmacol. Mar 1991;31(3):205-215. 25. Gosse ME, Nelson TF. Approval times for supplemental indications for recombinant proteins. Nat Biotechnol. Feb 1997;15(2):130-134. 26. Weeks EA. Is it worth the trouble? The new policy on diss emination of information on off-label drug use under the Food and Drug Administration Modern ization Act of 1997. Food Drug Law J. 1999;54(4):645-665. 27. Nightingale SL. Off-label use of prescription drugs. Am Fam Physician. Aug 1 2003;68(3):425-427. 28. US F. Draft Guidance for Industry: Good Reprint Practices for the Distribution of Medical Journal Articles and Medical or Scientific Refere nce Publication on Unapproved New Uses of Approved Drugs and Approved or Cleared Medical Devices. In: Services HaH, ed: FDA; 2008.
173 29. Barr JMD, Brenner-Zada GMD, Heiman EM D, et al. Unlicensed and Off-Label Medication Use in a Neonatal Intensive Care Unit: A Prospective Study. American Journal of Perinatology. 2002(2):067-072. 30. Conroy S, McIntyre J, Choonara I. Unli censed and off label drug use in neonates. Arch Dis Child Fetal Neonatal Ed. Mar 1999;80(2):F142-144; discussion F144-145. 31. Dell'Aera M, Gasbarro AR, Padovano M, et al Unlicensed and off-label use of medicines at a neonatology clinic in Italy. Pharm World Sci. Aug 2007;29(4):361-367. 32. O'Donnell CPF, Stone RJ, Morley CJ. Unlicensed and Off-Label Drug Use in an Australian Neonatal Intensive Care Unit Vol 110; 2002:e52-. 33. Carvalho PR, Carvalho CG, Alievi PT, Martin biancho J, Trotta EA. [Prescription of drugs not appropriate for children in a Pediatric Intensive Care Unit]. J Pediatr (Rio J). Sep-Oct 2003;79(5):397-402. 34. Conroy S, Choonara I, Impicciat ore P, et al. Survey of un licensed and off label drug use in paediatric wards in European countries. European Network for Drug Investigation in Children. Bmj. Jan 8 2000;320(7227):79-82. 35. Conroy S, Newman C, Gudka S. Unlicensed and off label drug use in acute lymphoblastic leukaemia and other malignancies in children Vol 14; 2003:42-47. 36. Craig JS, Henderson CR, Magee FA. The extent of unlicensed and off-label drug use in the paediatric ward of a district general hospital in Northern Ireland. Ir Med J. Sep 2001;94(8):237-240. 37. Dick A, Keady S, Mohamed F, et al. Use of unlicensed and off-label medications in paediatric gastroenterology w ith a review of the commonly used formularies in the UK Vol 17; 2003:571-575. 38. Eiland LS, Knight P. Evaluating the o ff-label use of medications in children. Am J Health Syst Pharm. Jun 1 2006;63(11):1062-1065. 39. Gavrilov V, Lifshitz M, Le vy J, Gorodischer R. Unlicensed and off-label medication use in a general pediatrics ambul atory hospital unit in Israel. Isr Med Assoc J. Aug 2000;2(8):595-597. 40. McKinzie JP, Wright SW Wrenn KD. Pediatric drug therapy in the emergency department: does it meet FDA-a pproved prescribing guidelines? Am J Emerg Med. Mar 1997;15(2):118-121. 41. Pandolfini C, Impicciatore P, Provasi D, Rocchi F, Campi R, Bonati M. Off-label use of drugs in Italy: a prospective, observational and multicentre study Vol 91; 2002:339-347.
174 42. Shah SS, Hall M, Goodman DM, et al. O ff-label drug use in hospitalized children. Arch Pediatr Adolesc Med. Mar 2007;161(3):282-290. 43. t Jong GW, Vulto AG, de Hoog M, Schimmel KJ, Tibboel D, van den Anker JN. A survey of the use of off-la bel and unlicensed drugs in a Dutch children's hospital. Pediatrics. Nov 2001;108(5):1089-1093. 44. Turner S, Gill A, Nunn T, Hewitt B, Choonara I. Use of "off-label" and unlicensed drugs in paediatric intensive care unit. Lancet. Feb 24 1996;347(9000):549-550. 45. Turner S, Longworth A, Nunn AJ, Choonara I. Unlicensed and off label drug use in paediatric wards: prospective study. Bmj. Jan 31 1998;316(7128):343-345. 46. Bucheler R, Schwab M, Morike K, et al. Off label prescribin g to children in primary care in Germany: retrospective cohort study. Bmj. Jun 1 2002;324(7349):1311-1312. 47. Chalumeau M, Treluyer JM, Salanave B, et al. Off label and unlicensed drug use among French office based paediatricians. Arch Dis Child. Dec 2000;83(6):502-505. 48. Ekins-Daukes S, Helms PJ, Simpson CR, Ta ylor MW, McLay JS. Off-label prescribing to children in primary care: retrospective observational study. Eur J Clin Pharmacol. Jul 2004;60(5):349-353. 49. Schirm E, Tobi H, de Jong-van den Berg LT. Risk factors for unlicensed and off-label drug use in children out side the hospital. Pediatrics. Feb 2003;111(2):291-295. 50. t Jong GW, Eland IA, Sturkenboom MCJM, van den Anker JN, Stricker BHC. Unlicensed and off label prescription of drugs to children: population based cohort study Vol 324; 2002:1313-1314. 51. t Jong GW, Eland IA, Sturkenboom MCJM, van den Anker JN, Stricker BHC. Unlicensed and off-label prescription of respiratory drugs to children Vol 23; 2004:310313. 52. Ufer M, Rane A, Karlsson A, Kimland E, Bergman U. Widespread off-label prescribing of topical but not systemic drugs for 3 50,000 paediatric outpatients in Stockholm. Eur J Clin Pharmacol. Mar 2003;58(11):779-783. 53. Wilton LV, Pearce G, Mann RD. The use of newly marketed drugs in children and adolescents prescribed in general practice. Pharmacoepidemiol Drug Saf. Apr 1999;8 Suppl 1:S37-45. 54. Yoon EY, Dombkowski KJ, Rocchini A, Lin JJ, Davis MM. Off-la bel utilization of antihypertensive medi cations in children. Ambul Pediatr. Jul-Aug 2007;7(4):299-303.
175 55. Li VW, Jaffe MP, Li WW, Haynes HA. Off-labe l dermatologic therapies. Usage, risks, and mechanisms. Arch Dermatol. Nov 1998;134(11):1449-1454. 56. Lin HW, Phan K, Lin SJ. Trends in off-label beta-blocker use: a secondary data analysis. Clin Ther. Oct 2006;28(10):1736-1746; discussion 1710-1731. 57. Sugarman JH, Fleischer AB, Jr., Feldman SR. Off-label prescribing in the treatment of dermatologic disease. J Am Acad Dermatol. Aug 2002;47(2):217-223. 58. Weiss E, Hummer M, Koller D, Pharmd, Ulme r H, Fleischhacker WW. Off-label use of antipsychotic drugs. J Clin Psychopharmacol. Dec 2000;20(6):695-698. 59. Atkinson CV, Kirkham SR. Unlicensed uses for medication in a palliative care unit Vol 13; 1999:145-152. 60. Chen C, Danekas LH, Ratko TA, Vlasses PH, Matuszewski KA. A multicenter drug use surveillance of intravenous immunoglobulin ut ilization in US academic health centers Vol 34; 2000:295-299. 61. Cras A, Conscience MA, Rajzbaum G, et al Off-label prescribing in a French hospital. Pharm World Sci. Apr 2007;29(2):97-100. 62. Loder EW, Biondi DM. Off-label prescribing of drugs in specialty headache practice. Headache. Jul-Aug 2004;44(7):636-641. 63. Poole SG, Dooley MJ. Off-label prescribing in oncology. Support Care Cancer. May 2004;12(5):302-305. 64. Assion HJ, Jungck C. Off-label prescrib ing in a German psychiatric hospital. Pharmacopsychiatry. Jan 2007;40(1):30-36. 65. Barbui C, Ciuna A, Nose M, et al. Of f-label and non-classi cal prescriptions of antipsychotic agents in ordinary in-patient practice. Acta Psychiatr Scand. Apr 2004;109(4):275-278. 66. Chen H, Reeves JH, Fincham JE, Kennedy WK, Dorfman JH, Martin BC. Off-label use of antidepressant, anticonvul sant, and antipsychotic medications among Georgia medicaid enrollees in 2001. J Clin Psychiatry. Jun 2006;67(6):972-982. 67. Gleason PP, Walters C, Heaton AH, Schafer JA. Telithromycin: the perils of hasty adoption and persistence of off-label prescribing. J Manag Care Pharm. Jun 2007;13(5):420-425. 68. Haw C, Stubbs J. A survey of off-label pres cribing for inpatients with mild intellectual disability and mental illness. J Intellect Disabil Res. Nov 2005;49(Pt 11):858-864.
176 69. Haw C, Stubbs J. A survey of the off-labe l use of mood stabilizers in a large psychiatric hospital. J Psychopharmacol. Jul 2005;19(4):402-407. 70. Martin-Latry K, Ricard C, Verdoux H. A one-d ay survey of characteristics of off-label hospital prescription of psychotropic drugs. Pharmacopsychiatry. May 2007;40(3):116120. 71. Rayburn WF, Turnbull GL. Off-label drug pr escribing on a state university obstetric service. J Reprod Med. Mar 1995;40(3):186-188. 72. Rijcken CA, Boelema GJ, Slooff CJ, Beuger PJ, Tanja TA, de Jong-van den Berg LT. Off-label use of antipsychotics in the community pharmacy: the sex differences. Pharmacopsychiatry. Sep 2003;36(5):187-191. 73. Trifiro G, Spina E, Brignoli O, Sessa E, Caputi AP, Mazzaglia G. Antipsychotic prescribing pattern among Italian general practitioners: a population-based study during the years 1999-2002. Eur J Clin Pharmacol. Mar 2005;61(1):47-53. 74. Hauser WA, Annegers JF, Kurland LT. Prev alence of epilepsy in Rochester, Minnesota: 1940-1980. Epilepsia. Jul-Aug 1991;32(4):429-445. 75. Perucca E, Alexandre Jr V, Tomson T. Ol d versus new antiepileptic drugs: the SANAD study. The Lancet. 370(9584):313. 76. Levy Rene H MRH, Meldrum Brian S, Perucca Emilio, ed. Antiepileptic Drugs. fifth ed. Philadelphia: Lippincott Williams & Wilkins 2002. 77. Perucca E. An introduction to antiepileptic drugs. Epilepsia. 2005;46 Suppl 4:31-37. 78. Rochat P, Hallas J, Gaist D, Friis ML. Antiepileptic drug utilization: a Danish prescription database analysis. Acta Neurol Scand. Jul 2001;104(1):6-11. 79. Schachter SC, Cramer GW, Thompson GD, Chaponis RJ, Mendelson MA, Lawhorne L. An evaluation of antiepileptic dr ug therapy in nurs ing facilities. J Am Geriatr Soc. Sep 1998;46(9):1137-1141. 80. Gatti G, Bonomi I, Jannuzzi G, Perucca E. The new antiepileptic drugs: pharmacological and clinical aspects. Curr Pharm Des. May 2000;6(8):839-860. 81. Perucca E. Marketed new antiepileptic drugs: are they better than old-generation agents? Ther Drug Monit. Feb 2002;24(1):74-80. 82. Rogawski MA, Loscher W. The neurobiology of antiepileptic drugs for the treatment of nonepileptic conditions. Nat Med. Jul 2004;10(7):685-692.
177 83. Gomez-Perez FJ, Choza R, Rios JM, et al. Nortriptyline-fluphenazine vs. carbamazepine in the symptomatic treatment of diabetic neuropathy. Arch Med Res. Winter 1996;27(4):525-529. 84. Tripathi M, Kaushik S. Carbamezapine for pain management in Guillain-Barre syndrome patients in the intensive care unit. Crit Care Med. Mar 2000;28(3):655-658. 85. McCleane G. 200 mg daily of lamotrigine has no analgesic effect in neuropathic pain: a randomised, double-blind, placebo controlled trial. Pain. Oct 1999;83(1):105-107. 86. McCleane GJ. Intravenous infusion of phenytoin relieves neuropathic pain: a randomized, double-blinded, placebo-controlled, crossover study. Anesth Analg. Oct 1999;89(4):985-988. 87. Morello CM, Leckband SG, Stoner CP, Moorhouse DF, Sahagian GA. Randomized double-blind study comparing the efficacy of gabapentin with amitriptyline on diabetic peripheral neuropathy pain. Arch Intern Med. Sep 13 1999;159(16):1931-1937. 88. Spina E, Perugi G. Antiepileptic dr ugs: indications other than epilepsy. Epileptic Disord. Jun 2004;6(2):57-75. 89. Krymchantowski AV, Bigal ME, Moreira PF. New and emerging prophylactic agents for migraine. CNS Drugs. 2002;16(9):611-634. 90. Mathew NT, Rapoport A, Saper J, et al. Effi cacy of gabapentin in migraine prophylaxis. Headache. Feb 2001;41(2):119-128. 91. Adelman JU, Adelman LC, Von Seggern R. Cost-effectiveness of antiepileptic drugs in migraine prophylaxis. Headache. Nov-Dec 2002;42(10):978-983. 92. Mack A. Examination of the evidence for off-label use of gabapentin. J Manag Care Pharm. Nov-Dec 2003;9(6):559-568. 93. D'Andrea G, Granella F, Cadaldini M, Manz oni GC. Effectiveness of lamotrigine in the prophylaxis of migraine with aura: an open pilot study. Cephalalgia. Jan 1999;19(1):6466. 94. Lampl C, Buzath A, Klinger D, Neumann K. Lamotrigine in the prophylactic treatment of migraine aura--a pilot study. Cephalalgia. Jan 1999;19(1):58-63. 95. Steiner TJ, Findley LJ, Yuen AW. Lamotrig ine versus placebo in the prophylaxis of migraine with and without aura. Cephalalgia. Apr 1997;17(2):109-112. 96. Koller WC, Hristova A, Brin M. Pharm acologic treatment of essential tremor. Neurology. 2000;54(11 Suppl 4):S30-38.
178 97. Gorman WP, Cooper R, Pocock P, Campbell MJ. A comparison of primidone, propranolol, and placebo in essential tremor, using quantitative analysis. J Neurol Neurosurg Psychiatry. Jan 1986;49(1):64-68. 98. Gironell A, Kulisevsky J, Barbanoj M, L opez-Villegas D, Hernandez G, Pascual-Sedano B. A randomized placebo-controlled comparativ e trial of gabapentin and propranolol in essential tremor. Arch Neurol. Apr 1999;56(4):475-480. 99. Pahwa R, Lyons K, Hubble JP, et al. Doubl e-blind controlled tria l of gabapentin in essential tremor. Mov Disord. May 1998;13(3):465-467. 100. Connor GS. A double-blind placebo-controlled tria l of topiramate treatment for essential tremor. Neurology. Jul 9 2002;59(1):132-134. 101. Cookson J. Toward a clinical understanding of bipolar disorders: classification and presentation. Epilepsia. 2005;46 Suppl 4:3-7. 102. Cookson J. Lithium: balancing risks and benefits. Br J Psychiatry. Aug 1997;171:120124. 103. Emrich HM, Dose M, von Zerssen D. The use of sodium valproate, carbamazepine and oxcarbazepine in patients with affective disorders. J Affect Disord. May-Jun 1985;8(3):243-250. 104. Altshuler LL, Keck PE, Jr., McElroy SL, et al. Gabapentin in the acute treatment of refractory bipolar disorder. Bipolar Disord. Sep 1999;1(1):61-65. 105. Carta MG, Hardoy MC, Dessi I, Hardoy MJ, Carpiniello B. Adjunctive gabapentin in patients with intellectual disability and bipolar spectrum disorders. J Intellect Disabil Res. Apr 2001;45(Pt 2):139-145. 106. Ghaemi SN, Goodwin FK. Gabapentin treatment of the non-refractory bipolar spectrum: an open case series. J Affect Disord. Jul 2001;65(2):167-171. 107. Sokolski KN, Green C, Maris DE, DeMet EM. Gabapentin as an adjunct to standard mood stabilizers in outpatients w ith mixed bipolar symptomatology. Ann Clin Psychiatry. Dec 1999;11(4):217-222. 108. Young LT, Robb JC, Hasey GM, et al. Gabapent in as an adjunctive treatment in bipolar disorder. J Affect Disord. Sep 1999;55(1):73-77. 109. Pande AC, Crockatt JG, Janney CA, Werth JL Tsaroucha G. Gabapentin in bipolar disorder: a placebo-controlled trial of adjunc tive therapy. Gabapentin Bipolar Disorder Study Group. Bipolar Disord. Sep 2000;2(3 Pt 2):249-255.
179 110. Pande AC, Pollack MH, Crockatt J, et al Placebo-controlled study of gabapentin treatment of pa nic disorder. J Clin Psychopharmacol. Aug 2000;20(4):467-471. 111. Suppes T. Review of the use of topiramate for treatment of bipolar disorders. J Clin Psychopharmacol. Dec 2002;22(6):599-609. 112. Grunze HC, Normann C, Langosch J, et al. Antimanic efficacy of topiramate in 11 patients in an open trial with an on-off-on design. J Clin Psychiatry. Jun 2001;62(6):464468. 113. Normann C, Langosch J, Schaerer LO, Grun ze H, Walden J. Treatment of acute mania with topiramate. Am J Psychiatry. Dec 1999;156(12):2014. 114. Ghaemi SN, Manwani SG, Katzow JJ, Ko JY, Goodwin FK. Topira mate treatment of bipolar spectrum disorders: a retrospective chart review. Ann Clin Psychiatry. Dec 2001;13(4):185-189. 115. McIntyre RS, Mancini DA, McCann S, Srinivasan J, Sagman D, Kennedy SH. Topiramate versus bupropion SR when added to mood stabilizer therapy for the depressive phase of bipolar disord er: a preliminary single-blind study. Bipolar Disord. Jun 2002;4(3):207-213. 116. Grunze H, Erfurth A, Marcuse A, Amann B, Normann C, Walden J. Tiagabine appears not to be efficacious in the treatment of acute mania. J Clin Psychiatry. Nov 1999;60(11):759-762. 117. Schaffer LC, Schaffer CB, Howe J. An open cas e series on the utility of tiagabine as an augmentation in refractory bipolar outpatients. J Affect Disord. Sep 2002;71(1-3):259263. 118. Suppes T, Chisholm KA, Dhavale D, et al. Tiagabine in treatment refractory bipolar disorder: a clinical case series. Bipolar Disord. Oct 2002;4(5):283-289. 119. Kanba S, Yagi G, Kamijima K, et al The first open study of zonisamide, a novel anticonvulsant, shows efficacy in mania. Prog Neuropsychopharmaco l Biol Psychiatry. Jul 1994;18(4):707-715. 120. Zaremba PD, Bialek M, Blaszczyk B, Cioczek P, Czuczwar SJ. Non-epilepsy uses of antiepilepsy drugs. Pharmacol Rep. Jan-Feb 2006;58(1):1-12. 121. Hosak L, Libiger J. Antiepileptic drugs in schizophrenia: a review. Eur Psychiatry. Nov 2002;17(7):371-378. 122. Simhandl C, Meszaros K. The use of carba mazepine in the treatment of schizophrenic and schizoaffective psychoses: a review. J Psychiatry Neurosci. Mar 1992;17(1):1-14.
180 123. Dursun SM, McIntosh D, Milliken H. Clozapine plus lamotrigine in treatment-resistant schizophrenia. Arch Gen Psychiatry. Oct 1999;56(10):950. 124. Erfurth A, Walden J, Grunze H. Lamotrigine in the treatment of schizoaffective disorder. Neuropsychobiology. Oct 1998;38(3):204-205. 125. Ettinger AB, Argoff CE. Use of antiepile ptic drugs for nonepileptic conditions: psychiatric disorders and chronic pain. Neurotherapeutics. Jan 2007;4(1):75-83. 126. Rickels K, Pollack MH, Feltner DE, et al. Pregabalin for treatment of generalized anxiety disorder: a 4-week, multicenter, double-blind, pl acebo-controlled trial of pregabalin and alprazolam. Arch Gen Psychiatry. Sep 2005;62(9):1022-1030. 127. McIntyre J, Conroy S, Avery A, Corns H, Choonara I. Unlicensed and off label prescribing of drugs in general practice. Arch Dis Child. Dec 2000;83(6):498-501. 128. Neubert A, Dormann H, Weiss J, et al. The impact of unlicensed and off-label drug use on adverse drug reactions in paediatric patients. Drug Saf. 2004;27(13):1059-1067. 129. Denig P, Haaijer-Ruskamp FM, Zijsling DH. How physicians choose drugs. Soc Sci Med. 1988;27(12):1381-1386. 130. Rogers EM SK. The diffusion model and outre ach from the national network of libraries of medicine to native americian committees 1997. 131. Eisenberg JM. Sociologic influenc es on decision-making by clinicians. Ann Intern Med. Jun 1979;90(6):957-964. 132. Clark JA, Potter DA, McKinlay JB. Bringing so cial structure back in to clinical decision making. Soc Sci Med. 1991;32(8):853-866. 133. Linn BS, Linn MW. Patient symptoms and physic ian prescribing patterns in the elderly. Soc Sci Med. 1982;16(17):1531-1538. 134. Roe CM, McNamara AM, Motheral BR. Genderand age-related prescription drug use patterns. Ann Pharmacother. Jan 2002;36(1):30-39. 135. Westbrook MT, Mitchell RA. Changes in sex-ro le stereotypes from health to illness. Soc Sci Med. May 1979;13A(3):297-302. 136. Copeland LA, Zeber JE, Valenstein M, Blow FC. Racial disparity in the use of atypical antipsychotic medications among veterans. Am J Psychiatry. Oct 2003;160(10):18171822.
181 137. Davis M, Poisal J, Chulis G, Zaraboz o C, Cooper B. Prescription drug coverage, utilization, and spending am ong Medicare beneficiaries. Health Aff (Millwood). Jan-Feb 1999;18(1):231-243. 138. Johnston SC, Fung LH, Gillum LA, et al. Utilization of intravenous tissue-type plasminogen activator for ischemic stroke at academic medical centers: the influence of ethnicity. Stroke. May 2001;32(5):1061-1068. 139. Kressin NR, Petersen LA. Racial differences in the use of invasive cardiovascular procedures: review of th e literature and prescrip tion for future research. Ann Intern Med. Sep 4 2001;135(5):352-366. 140. Kuno E, Rothbard AB. Racial disparities in antipsychotic prescription patterns for patients with schizophrenia. Am J Psychiatry. Apr 2002;159(4):567-572. 141. Penchansky R, Fox D. Frequency of referral and patient characteristics in group practice. Med Care. Sep-Oct 1970;8(5):368-385. 142. Perkoff GT, Anderson M. Relationship between demographic characteristics, patient's chief complaint, and medical care de stination in an emergency room. Med Care. Jul-Aug 1970;8(4):309-323. 143. Himmelstein DU, Woolhandler S, Harnly M, et al. Patient transfers: medical practice as social triage. Am J Public Health. May 1984;74(5):494-497. 144. Stolley PD, Becker MH, Lasagna L, McEvilla JD, Sloane LM. The relationship between physician characteristics and prescribing appropriateness. Med Care. Jan-Feb 1972;10(1):17-28. 145. Hemminki E. The effect of a doctor's pers onal characteristics and working circumstances on the prescribing of psychotropic drugs. Med Care. Apr 1974;12(4):351-357. 146. Russell ML, Love EJ. Contraceptive prescrip tion: physician beliefs, attitudes and sociodemographic characteristics. Can J Public Health. Jul-Aug 1991;82(4):259-263. 147. Benson PR. Factors associated with an tipsychotic drug prescribing by southern psychiatrists. Med Care. Jun 1983;21(6):639-654. 148. Huang N, Chou YJ, Chang HJ, Ho M, Morl ock L. Antibiotic prescribing by ambulatory care physicians for adults with nasopharyngitis URIs, and acute bronchitis in Taiwan: a multi-level modeling approach. Fam Pract. Apr 2005;22(2):160-167. 149. Rosenblatt RA, Moscovice IS. The physician as gatekeeper. Determinants of physicians' hospitalization rates. Med Care. Feb 1984;22(2):150-159.
182 150. Chin MH, Friedmann PD, Cassel CK, Lang RM. Differences in generalist and specialist physicians' knowledge and use of angiotensin-converting enzyme inhibitors for congestive heart failure. J Gen Intern Med. Sep 1997;12(9):523-530. 151. Chin MH, Zhang JX, Merrell K. Specialty diffe rences in the care of older patients with diabetes. Med Care. Feb 2000;38(2):131-140. 152. Pugh MJ, Anderson J, Pogach LM, Berl owitz DR. Differential adoption of pharmacotherapy recommendations for type 2 diabetes by generalists and specialists. Med Care Res Rev. Jun 2003;60(2):178-200. 153. Greer AL. The state of the art versus the state of the science. The diffusion of new medical technologies into practice. Int J Technol Assess Health Care. 1988;4(1):5-26. 154. Charles D. Hepler RS. Preventing Medication Errors and Improving Drug Therapy Outcomes: A management Systems Approach 1ed: CRC Press; 2003. 155. Bell RA, Wilkes MS, Kravitz RL. Advertis ement-induced prescription drug requests: patients' anticipated reactions to a physician who refuses. J Fam Pract. Jun 1999;48(6):446-452. 156. Lyles A. Direct marketing of pharmaceuticals to consumers. Annu Rev Public Health. 2002;23:73-91. 157. Weissman JS, Blumenthal D, Silk AJ, Zapert K, Newman M, Leitman R. Consumers' reports on the health effects of di rect-to-consumer drug advertising. Health Aff (Millwood). Jan-Jun 2003;Suppl Web Exclusives:W3-82-95. 158. Donohue JM, Berndt ER, Rosenthal M, Epstein AM, Frank RG. Effects of pharmaceutical promotion on adherence to the treatment guidelines for depression. Med Care. Dec 2004;42(12):1176-1185. 159. Hansen RA, Shaheen NJ, Schommer JC. Factors influencing the shift of patients from one proton pump inhibitor to another: the effect of direct-to-consumer advertising. Clin Ther. Sep 2005;27(9):1478-1487. 160. Zachry WM, 3rd, Shepherd MD, Hinich MJ, Wilson JP, Brown CM, Lawson KA. Relationship between direct-to-consumer advertising and physician diagnosing and prescribing. Am J Health Syst Pharm. Jan 1 2002;59(1):42-49. 161. Williamson PM. The adoption of new drugs by doctors practising in group and solo practice. Soc Sci Med. Apr-May 1975;9(4-5):233-236. 162. Renaud M, Beauchemin J, Lalonde C, Poirier H, Berthiaume S. Practice settings and prescribing profiles: the simulation of tension headaches to general practitioners working
183 in different practice setti ngs in the Montreal area. Am J Public Health. Oct 1980;70(10):1068-1073. 163. Avorn J, Chen M, Hartley R. Scientific versus commercial sources of influence on the prescribing behavior of physicians. Am J Med. Jul 1982;73(1):4-8. 164. McCue JD, Hansen CJ, Gal P. Physicians' opinions of the accurac y, accessibility, and frequency of use of ten sources of new drug information. South Med J. Apr 1986;79(4):441-443. 165. McKinney WP, Schiedermayer DL, Lurie N, Simpson DE, Goodman JL, Rich EC. Attitudes of internal medicine faculty and residents toward professional interaction with pharmaceutical sales representatives. Jama. Oct 3 1990;264(13):1693-1697. 166. Peay MY, Peay ER. Patterns of preference fo r information sources in the adoption of new drugs by specialists. Soc Sci Med. 1990;31(4):467-476. 167. Finley PR, Crismon ML, Rush AJ. Evaluating the impact of pharmacists in mental health: a systematic review. Pharmacotherapy. Dec 2003;23(12):1634-1644. 168. Rollason V, Vogt N. Reduction of polypharmacy in the elderly: a systematic review of the role of the pharmacist. Drugs Aging. 2003;20(11):817-832. 169. US F. Drugs@FDA. http://www.accessdata.fda.gov/scripts/cder/drugsatfda/. Accessed 11/2007, 2007. 170. Ling DC BE, Kyle Mk. Deregulating direct-t o-consumer marketing of prescription drugs: effects on prescription and over-the-counter product sales. Journal of Law and Economics. 2002;45:691-723. 171. FDA U. Information for Healthcare Professionals Suicidality and Antiepileptic Drugs. In: Services DoHaH, ed; 2008. 172. Annegers JF, Dubinsky S, Coan SP, Newmar k ME, Roht L. The incidence of epilepsy and unprovoked seizures in multiethnic, urban health maintenance organizations. Epilepsia. Apr 1999;40(4):502-506. 173. Steinman MA, Harper GM, Chren MM, La ndefeld CS, Bero LA. Characteristics and impact of drug detailing for gabapentin. PLoS Med. Apr 2007;4(4):e134. 174. Russell ME, Friedman MI, Mascioli SR, Stolz LE. Off-label use: an industry perspective on expanding use beyond approved indications. J Interv Cardiol. Oct 2006;19(5):432438. 175. Quality AfHRa. Outcomes/Effectiveness Res earch: Evidence is lacking to support many off-label uses of atypica l antipsychotic drugs; 2007
184 BIOGRAPHICAL SKETCH Ahunna Onyenwenyi received a Bachelor of Pharm acy degree from the University of Lagos, Nigeria in 2000. She worked with with a non-governmental organization that cared for persons living with HIV/AIDS in Nigeria before coming to the University of Florida. She is a recipient of the University of Florida alumni fellowship.