<%BANNER%>

Use of Attention Deficit/Hyperactivity Disorder Pharmacological Treatment Following the Public Discussion on Cardiovascu...

Permanent Link: http://ufdc.ufl.edu/UFE0041255/00001

Material Information

Title: Use of Attention Deficit/Hyperactivity Disorder Pharmacological Treatment Following the Public Discussion on Cardiovascular Safety and the introduction of Medication Guides for Central Nervous System Stimulants and Atomoxetine
Physical Description: 1 online resource (149 p.)
Language: english
Creator: Chen, Chih-Ying
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2010

Subjects

Subjects / Keywords: adhd, cns, drug, pharmacoepidemiology
Pharmaceutical Outcomes and Policy -- Dissertations, Academic -- UF
Genre: Pharmaceutical Sciences thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: USE OF ATTENTION DEFICIT/HYPERACTIVITY DISORDER PHARMACOLOGICAL TREATMENT FOLLOWING THE PUBLIC DISCUSSION ON CARDIOVASCULAR SAFETY AND THE INTRODUCTION OF MEDICATION GUIDES FOR CENTRAL NERVOUS SYSTEM STIMULANTS AND ATOMOXETINE Intense public discussion and some regulatory action occurred in 2005- 2007 following a safety signal regarding the cardiovascular (CV) risk of central nervous system stimulants (stimulant) and atomoxetine. Our objective was to describe changes in stimulant utilization and pre-treatment electrocardiography (ECG) screening after this safety signal. A time-series design including 96 months (2001-2008) of Florida Medicaid claims data was used to describe changes in trends of stimulant utilization. Joinpoint regression models were used to detect the number of change points, to estimate the magnitude of change in monthly outcome trends, and to test comparability of trend change(s) among patient subgroups. Both the initial and maintenance daily dose declined by 6 milligram (mg) methylphenidate equivalent dose from a previously steady dose of 26 mg after Canada withdrew one stimulant product (Adderall XR) in 02/05; the trend rebounded to a consistently lower dose (23 mg) after the remarketing of Adderall XR and a debate in the U.S. over issuing a boxed warning on stimulant CV safety in early 2006. The monthly initiation rate increased by 3.9% (95% CI -1.0, 9.1) after the boxed warning debate, but started to decline by 2.4%(-3.6,-1.2) after patient medication guides (MedGuides) released in 02/07. Monthly ECG screening increased by 3.2% (2.3, 4.2) after Canada withdrew Adderall XR and further increased by 13 % (4.2, 23) after the American Heart Association recommended pre-stimulant-treatment ECG. The decline in treatment initiation differed by patient age, presence of mental comorbidities or disability; the reduction in treatment intensity, however, did not show significant difference among subgroups. A more pronounced increase in ECG use was observed among patients with less complicated mental conditions or patients who were diagnosed by psychiatrists. Practitioners reacted to stimulant CV safety signal with immediate reduction in dosing and an increase in preventive screening, affecting however only a marginal proportion of patients. Utilization rates started to drop only after release of MedGuides and affected only initiation but not chronic use of stimulants. Change in stimulant use was not significant among vulnerable subgroups, such as young patients, patients with mental comorbidities or disability. Clinical consequences of these changes are uncertain.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Chih-Ying Chen.
Thesis: Thesis (Ph.D.)--University of Florida, 2010.
Local: Adviser: Winterstein, Almut G.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2012-08-31

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2010
System ID: UFE0041255:00001

Permanent Link: http://ufdc.ufl.edu/UFE0041255/00001

Material Information

Title: Use of Attention Deficit/Hyperactivity Disorder Pharmacological Treatment Following the Public Discussion on Cardiovascular Safety and the introduction of Medication Guides for Central Nervous System Stimulants and Atomoxetine
Physical Description: 1 online resource (149 p.)
Language: english
Creator: Chen, Chih-Ying
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2010

Subjects

Subjects / Keywords: adhd, cns, drug, pharmacoepidemiology
Pharmaceutical Outcomes and Policy -- Dissertations, Academic -- UF
Genre: Pharmaceutical Sciences thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: USE OF ATTENTION DEFICIT/HYPERACTIVITY DISORDER PHARMACOLOGICAL TREATMENT FOLLOWING THE PUBLIC DISCUSSION ON CARDIOVASCULAR SAFETY AND THE INTRODUCTION OF MEDICATION GUIDES FOR CENTRAL NERVOUS SYSTEM STIMULANTS AND ATOMOXETINE Intense public discussion and some regulatory action occurred in 2005- 2007 following a safety signal regarding the cardiovascular (CV) risk of central nervous system stimulants (stimulant) and atomoxetine. Our objective was to describe changes in stimulant utilization and pre-treatment electrocardiography (ECG) screening after this safety signal. A time-series design including 96 months (2001-2008) of Florida Medicaid claims data was used to describe changes in trends of stimulant utilization. Joinpoint regression models were used to detect the number of change points, to estimate the magnitude of change in monthly outcome trends, and to test comparability of trend change(s) among patient subgroups. Both the initial and maintenance daily dose declined by 6 milligram (mg) methylphenidate equivalent dose from a previously steady dose of 26 mg after Canada withdrew one stimulant product (Adderall XR) in 02/05; the trend rebounded to a consistently lower dose (23 mg) after the remarketing of Adderall XR and a debate in the U.S. over issuing a boxed warning on stimulant CV safety in early 2006. The monthly initiation rate increased by 3.9% (95% CI -1.0, 9.1) after the boxed warning debate, but started to decline by 2.4%(-3.6,-1.2) after patient medication guides (MedGuides) released in 02/07. Monthly ECG screening increased by 3.2% (2.3, 4.2) after Canada withdrew Adderall XR and further increased by 13 % (4.2, 23) after the American Heart Association recommended pre-stimulant-treatment ECG. The decline in treatment initiation differed by patient age, presence of mental comorbidities or disability; the reduction in treatment intensity, however, did not show significant difference among subgroups. A more pronounced increase in ECG use was observed among patients with less complicated mental conditions or patients who were diagnosed by psychiatrists. Practitioners reacted to stimulant CV safety signal with immediate reduction in dosing and an increase in preventive screening, affecting however only a marginal proportion of patients. Utilization rates started to drop only after release of MedGuides and affected only initiation but not chronic use of stimulants. Change in stimulant use was not significant among vulnerable subgroups, such as young patients, patients with mental comorbidities or disability. Clinical consequences of these changes are uncertain.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Chih-Ying Chen.
Thesis: Thesis (Ph.D.)--University of Florida, 2010.
Local: Adviser: Winterstein, Almut G.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2012-08-31

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2010
System ID: UFE0041255:00001


This item has the following downloads:


Full Text

PAGE 1

1 USE OF ATTENTION DEF ICIT/HYPERACTIVITY DISORDER PHARMACOLOGI CAL TREATMENT FOLLOWING THE PUBLIC DISCUSSIO N ON CARDIOVASCULAR SAFETY AND THE INTRO DUCTION OF MEDICATIO N GUIDES FOR CENTRAL NERVOUS SYSTEM STIMULANTS AND ATOMOXETINE By CHIHYING CHEN 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 2010

PAGE 2

2 2010 Chih Ying Chen

PAGE 3

3 To my parents

PAGE 4

4 ACKNOWLEDGMENTS I want to thank my advisor, Almut Winterste in, for her continuous support and guidance I especially appreciate that she held a high standard and never s topped challenging me. I would also like to thank my supervisory committee members Regina Bussing, Abraham Hartzema, Richard Segal, and Jon Shuster for their expertise, advice and encouragement. I extend my appreciation to all faculty members and staff in the department of Pharmaceutical Outcomes and Policy for their support and for everything they taught me during my study at the University of Florida. I express gratitude to Floridas Agency for Health Care Administration (AHCA) for the provision of Medicaid data. I appreciate Paul Duncan and Heather Steingraber from University o f FloridaFlorida Center for Medicaid & the Uninsured, and, Chris Mallison from Florida AHCA for their help on facilitating data access. I want to thank my family grandparents, parents, aunties, uncles, brother and cousins for their infinite support throughout my life I am grateful to Szu Ping Lee for his company and emotional support throughout my graduate study. Finally, I thank my fellow graduate students and my friends for their friendship.

PAGE 5

5 TABLE OF CONTENTS page ACKNOWLEDGMENTS .................................................................................................. 4 LIST OF TABLES ............................................................................................................ 8 LIST OF FIGURES ........................................................................................................ 10 1 CHAPTER INTRODUCTION .................................................................................................... 15 Background ............................................................................................................. 15 Need for Study ................................................................................................. 16 Purpose of Study .............................................................................................. 16 Study Significance ............................................................................................ 17 Research Questions and Hypotheses ..................................................................... 17 Part I: Testing of the Change(s) in Stimulant Utilization ................................... 18 Part II: Comparability Test among Subgroups .................................................. 19 2 LITERATURE REVIEW .......................................................................................... 29 Part I ....................................................................................................................... 29 The Role of Stimulants in ADHD Treatment ..................................................... 29 Stimulant Utilization among ADHD Patients ..................................................... 30 Factors associated with stimulant utilization .............................................. 30 Factors associated with stimulant persistence ........................................... 31 Stimulants and Cardiovascular Risk ................................................................. 32 Part II ...................................................................................................................... 33 New Drug Safety Information and Patients or Health Care Providers Behavior ........................................................................................................ 33 Dear healthcare professional letters .......................................................... 33 Public regulatory warnings and/or media reports of new safety information .............................................................................................. 34 Factors mediate the impact of new drug safety information on drug utilization ....................................................................................................... 35 3 METHODOLOGY ................................................................................................... 42 Study Design .......................................................................................................... 42 Data Source ............................................................................................................ 42 Study Population ..................................................................................................... 43 Cohort of New Episode ADHD Patients ........................................................... 43 Cohort of New Stimulant Users ........................................................................ 44 Study Measures ...................................................................................................... 44 Stimulant Initiation ............................................................................................ 44

PAGE 6

6 Stimulant Discontinuation ................................................................................. 45 ADHD Treatment Switching .............................................................................. 45 Stimulant Initial Daily Dose ............................................................................... 46 Stimulant Maintenance Daily Dose ................................................................... 46 Pre treatment Screening (for Cardiac Risk) ...................................................... 47 Data A nalysis .......................................................................................................... 47 Part I: Joinpoint Analysis .................................................................................. 47 Model fitting s pecification ........................................................................... 48 Hypothesis t esting ...................................................................................... 48 Adjustment for seasonality ......................................................................... 49 Part II Sub group Analysis ................................................................................ 50 Patient demographic characteristic ............................................................ 50 Patient clinical characteristics .................................................................... 51 Provider characteristics .............................................................................. 52 Model fitting Specification ................................................................................. 52 Hypothesis Testing ........................................................................................... 53 4 RESULTS ............................................................................................................... 56 Stimulant Initiation Trend ........................................................................................ 56 Stimulant Discontinuation Trend ............................................................................. 56 Stimulant Initial Daily Dose Trend ........................................................................... 59 Stimulant Maintenance Daily Dose Trend ............................................................... 61 Pre treatment ECG Use T rend ............................................................................... 62 5 DISCUSSIONS ..................................................................................................... 122 Stimulant Utilization after Health Canadas Withdrawal of Adderall ...................... 122 Stimulant Utilization after the Debate on Boxed Warnings .................................... 122 Stimulant Utilization after the Distribution of MedGuides ...................................... 123 ECG Utilization after Stimulant CV Safety Wa rnings ............................................ 123 Change in Trends among Subgroups ................................................................... 126 Other Important Findings ...................................................................................... 126 The Application of Joinpoint Analysis ................................................................... 128 Study Limitations .................................................................................................. 129 Summary and Future Research ............................................................................ 129 Conclusions .......................................................................................................... 130 A APPENDIX NATIONAL DRUG CODE (NDC): STIMULANTS ............................................. 134 B STIMULANT DEFINED DAILY DOSE (DDD) ................................................... 135 C NATIONAL DRUG CODE (NDC): SECOND LINE ADHD TRE ATMENT (BUPROPION) .................................................................................................. 136

PAGE 7

7 D NATIONAL DRUG CODE (NDC) : SECOND LINE ADHD T REATMENT (OTHERS) ........................................................................................................ 137 E CPT (CURRRENT PROCEDURAL TERMINOLOGY) CO DES: E LECTRO C ARDIO G RAPHY (ECG) ................................................................ 139 LIST OF REFERENCES ............................................................................................. 140 BIOGRAPHICAL SKETCH .......................................................................................... 149

PAGE 8

8 LIST OF TABLES Table page 1 1 R egulatory actions/discussions on stimulant cardiovascular safety .................... 27 1 2 E vidence regarding safety concerns in recent drug utilization studies ................ 28 2 1 Lists of stimulants approved for ADHD ............................................................... 3 9 2 2 Summary of FDA communication methods of drug safety information ............... 40 2 3 Major factors affecting health services utiliz ation ................................................ 41 3 1 Stimulant refill patterns ....................................................................................... 54 3 2 Secondline ADHD treatment refill patterns ........................................................ 54 4 1 Summary of analysis time frame, eligible population, and event count .............. 66 4 2 Patient and provider characteristics in stimulant initiation trend analysis ........... 67 4 3 Testing results for stimulant initiation trend ........................................................ 69 4 4 Specification for stimulant initiation trend ........................................................... 69 4 5 Testing results for oral stimulant+ initiation trend ................................................ 69 4 6 Specification for oral stimulant+ initiation trend ................................................... 69 4 7 Comparability test results for stratified stimulant initiation trend ......................... 70 4 8 Patient and provider characteristics in stimulant discontinuation trend analysis .............................................................................................................. 71 4 9 Testing results for stimulant discontinuation trend .............................................. 73 4 10 Specification for stimulant discontinuation trend ................................................ 73 4 11 Comparability test results for stratified stimulant discontinuation trend .............. 74 4 12 Patient and provider characteristics in stimulant initial daily dose trend analysis .............................................................................................................. 75 4 13 Testing resu lts for stimulant initial daily dose trend ............................................ 77 4 14 Specificat ion for stimulant initial daily dose trend ............................................... 77 4 15 Comparability test results for stratified stimulant initial daily dose trend ............. 78

PAGE 9

9 4 16 Patient and provider characteristics in stimulant maintenance daily dose trend analysis ..................................................................................................... 79 4 17 Testing results for stimulant maintenance daily dose trend ................................ 81 4 18 Specification for stimulant maintenance daily dose trend ................................... 81 4 19 Comparability test results for stratified stimulant maintenance daily dose trend ................................................................................................................... 82 4 20 Patient and provider characteristics in pretreatment electrocardiograph (ECG) use trend analysis ................................................................................... 83 4 21 Testing results for pre treatment electrocardiograph use trend .......................... 85 4 22 Specification for pre treat ment electrocardiograph use trend ............................. 85 4 23 Comparability test results for stratified pre treatment electrocardiograph use trend ................................................................................................................... 86 5 1 Age distribution of Florida Medicaid feefor service population during 20012008. ................................................................................................................ 132 5 2 Age distribution of newly diagnosed ADHD patient* in Florida Medicaid feefor service population during 20012008. ......................................................... 133

PAGE 10

10 LIST OF FIGURES Figure page 3 1 Original and deseaonalized trends. A) stimulant initiation. B) stimulant discontinuation. ................................................................................................... 55 4 1 Stimulant initiation trend ..................................................................................... 87 4 2 Stimulant initiation trend (oral products only) ...................................................... 89 4 3 Stimulant and atomoxetine initiation trend .......................................................... 88 4 4 Stimulant initiation trend by age (< 5 years versus 513 years) .......................... 90 4 5 Stimulant initiation trend by age (1420 y ears old versus 513 years old) .......... 90 4 6 Stimulant initiation trend by gender .................................................................... 91 4 7 Stimulant initiation trend by race (Blacks versus Whites) ................................... 92 4 8 Stimulant initiation trend by race (Hispanics versus Whites) .............................. 92 4 9 Stimulant initiation trend by eligibility (Supplemental security income [SSI] versus Temporary assistance for needy family [TANF]) ..................................... 93 4 10 Stimulant initiation trend by eligibility (Foster care vs Temporary assistance for needy family [TANF]) ..................................................................................... 93 4 11 Stimulant initiation trend by comorbidity status ................................................... 94 4 12 Stimulant initiation trend by provider type ........................................................... 95 4 13 Stimulant discontinuation trend ........................................................................... 96 4 14 Stimulant discontinuation trend by age group (< 5 years versus 513 years) ..... 97 4 15 Stimulant discontinuation trend by age group (513 years versus 1420 years) ................................................................................................................. 97 4 16 Stimulant discontinuation trend by gender .......................................................... 98 4 17 Stimulant discontinuation trend by race (Blacks versus Whites) ......................... 99 4 18 Stimulant discontinuation trend by race (Hispanics versus Whites) .................... 99 4 19 Stimulant discontinuation trend by eligibility status (Supplemental security income [SSI] versus Temporary assistance for needy family [TANF]) .............. 100

PAGE 11

11 4 20 Stimulant discontinuation trend by eligibility status (Foster care versus Temporary assistance for needy family [TANF]) ............................................... 100 4 21 Stimulant discontinuation trend by comorbidity status ...................................... 101 4 22 Stimulant initial daily dose trend ....................................................................... 102 4 23 Stimulant initial daily dose trend by age group (< 5 years versus 5 13 years) .. 103 4 24 Stimulant i nitial daily dose trend by age group (1420 years versus 513 years) ............................................................................................................... 103 4 25 Stimulant initial daily dose trend by gender ...................................................... 104 4 26 Stimulant initial daily dose trend by race (Blacks versus Whites) ..................... 105 4 27 Stimulant initial daily dose trend by race (Hispanics versus Whites) ................ 105 4 28 Stimulant initial daily dose trend by eligibility status (Supplemental security income [SSI] versus Temporary assistance for needy family [TANF]) .............. 106 4 29 Stimulant initial daily dose trend by eligibility status (Foster care versus Temporary assistance for needy family [TANF]) ............................................... 106 4 30 Stimulant initial daily dose trend by comorbidity status ..................................... 107 4 31 Stimulant initial daily dose trend by provider type ............................................. 108 4 32 Stimulant maintenance daily dose tr end ........................................................... 109 4 33 Stimulant maintenance daily dose trend by age group (< 5 years versus 513 years) ............................................................................................................... 110 4 34 Stimulant maintenance daily dose trend by age group (1420 years versus 513 years) .......................................................................................................... 110 4 35 Stimulant maintenance daily dose trend by gender .......................................... 111 4 36 Stimulant m aintenance daily dose trend by race (Blacks versus Whites) ......... 112 4 37 Stimulant maintenance daily dose trend by race (Hispanics versus Whites) .... 112 4 38 Stimulant maintenance daily dose trend by eligibility status (Supplemental security income [SSI] versus Temporary assistance for needy family [TANF]) 113 4 39 Stimulant maintenance daily dose trend by eligibility status (Foster care versus Temporary assistance for needy family [TANF]) ................................... 113 4 40 Stimulant maintenance daily dose trend by comorbidity status ........................ 114

PAGE 12

12 4 41 Pre treatment ECG use trend ........................................................................... 115 4 42 Pre treatment electrocardiography use trend by age group (< 5 years versus 5 13 years) ....................................................................................................... 116 4 43 Pre treatment electrocardiogr aphy use trend by gender .................................. 117 4 44 Pre treatment electrocardiography use trend by race (Black versus Whites) ... 118 4 45 Pre treatment electrocardiography use trend by race (Hispanics versus Whites) ............................................................................................................. 118 4 46 Pre treatment electrocardiography use trend by eligibility status (Supplemental security income [SSI] versus Temporary assistance for needy family [TANF]) ................................................................................................... 119 4 47 Pre treatment electrocardiography use trend by eligibility status (Foster care versus Temporary assistance for needy family [TANF]) ................................... 119 4 48 Pre treatment electrocardiography use trend by comorbidity status ................. 120 4 49 Pre treatment electrocardiography use trend by provider type ......................... 121

PAGE 13

13 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy USE OF ATTENTION DEF ICIT/HYPERACTIVITY DISORDER PHARMACOLOGI CAL TREATMENT FOLLOWING THE PUBLIC DISCUSSION ON CARDIO VASCULAR SAFETY AND THE INTRO DUCTION OF MEDICATIO N GUIDES FOR CENTRAL NERVOUS SYSTEM STIMULANTS AND ATOMOXETINE By Chih Ying Chen August 2010 Chair: Almut Winterstein Major: Pharmaceutical Sciences Abstract Intense public discussion and some regulatory action occurred in 20052007 following a safety signal regarding the cardiovascular (CV) risk of central nervous system stimulants (stimulant) and atomoxetine. Our objective was t o describe changes in stimulant uti lization and pretreatment electrocardiography (ECG) screening after this safety signal A time serie s design including 96 months ( 20 012008) of Florida Medicaid claims data was used to describe changes in trends of stimulant utilization. Joinpoint regres sion models were used to detect the number of change points, to estimate the magnitude of change in monthly outcome trends, and to test comparability of trend change(s) among patient su bgroups Both the initial and maintenance daily dose declined by 6 milligram (mg) methylphenidate equivalent dose from a previously steady dose of 26 mg a fter Canada withdrew one stimulant product (Adderall XR) in 02/05; the trend rebounded to a consistently lower dose ( 23 mg) after the remar keting of Adder all XR and a debate in

PAGE 14

14 the U.S. over issuing a boxed warning on stimulant CV safety in early 2006. The monthly initiation rate increased by 3.9% (95% CI 1.0, 9.1) after the boxed warning debate, but started to decline by 2.4%( 3.6, 1.2) after patient medi cation guides (MedGuides) released in 02/07. Monthly ECG screening increased by 3.2% (2.3, 4.2) after Canada withdrew Adderall XR and further increased by 13 % (4.2, 23) after the American Heart Association recommended prestimulant treatment ECG The decline in treatment initiation differed by patient age, presence of mental comorbidities or disability; the reduction in treatment intensity, however, did not show significant difference among subgroups A more pronounced increase in ECG use was observed among patients with less complicated mental conditions or patients who were diagnosed by psychiatrists. Practitioners reacted to stimulant CV safety signal with immediate reduc tion in dosing and an increase in preventive screening, affecting however only a m arginal proportion of patients. Utilization rates started to drop only after release of MedGuides and affected only initiation but not chronic use of stimulants. Change in stimulant use was not significant among vulnerable subgroups, such as young patients patients with mental comorbidities or disability Clinical consequence s of these changes are uncertain.

PAGE 15

15 CHAPTER 1 INTRODUCTION Background Intense public discussion and regulatory action have taken place during 2005 and 2007 regarding the cardiovascular (CV) safety of central nervous system (CNS) stimula nts and atomoxetine ( Table 11). In 2005, Health Canada, the national regulatory agency in Canada suspended the extended release (ER) form of mixed amphetamine salts (Adderall XR, Shire, Canada) for 6 months (February to August) following 20 reports of cardiac sudden deaths.13 A review by the US Food and Drug Administration (FDA) revealed reports of sudden death in patients with underly ing heart conditions, and reports of stroke and heart attack in adults with cardiac problems while taking stimulants.4 A heated debate followed in February / March 2006 because two FDA expert advisory panels proposed conflicting actions on this issue.5, 6 The Drug Safety and Risk Management Advisory Committee voted to add a black box warning about the CV risk. This recommendation was based on the proven potential for sympathomimetic agents to raise heart rate and blood pressure, serious adverse effects described for other members of this drug class,7 and the rapid increase in stimulant use.5 The Pediatric Advisory Committee suggested that a black box warning may not be warranted, given the strong evidence on treatment effectiveness and the weak evidence on risk.6 The FDA decided against the boxed warning but required a product labeling change to reflect concerns about adverse CV ev ents. Furthermore, beginning in February 2007, all ADHD ( Attention Deficit/hyperactivity Disorder ) drugs (including stimulants and atomoxetine) have been required to be dispensed with a medication guide (MedGuide) to notify patients about CV adverse events .8

PAGE 16

16 This study aims to evaluate the possible impact of the public discussions of CV safety concerns and the FDA requi rement for a MedGuide for stimulants on trends in stimulants utilization, and explore patient and physician characteristics that might mediate such effects. Need for Study Stimulants and atomoxetine are the only drug tr eatments approved for ADHD by the FDA ADHD affects 3 7% of children and 4.4% of adults in the United States.9 An estimated 2.5 mi llion youths and 1.5 million adults take medication for ADHD.9The uncertainty about the magnitude and generalizability of CV safety concerns further makes this study an i ntriguing case, since existing drug utilization research has been focused on warnings that were triggered by information indicating a strong association or revealing a causal relationship between the drug(s) and the adverse event(s) (Table 12 ). We will ta ke advantage of the opportunity to understand how people react while there are conflicting opinions in the interpretation of existing evidence. No study has examined whether t he discussions on CV safety conce rns have resulted in changes in pharmaceutical therapy and what patient groups were more affected by such change. Purpose of Study The study intends to examine the effect of CV safety warnings on stimulant treatment utilizati on among beneficiaries of a southern state Medicaid program. We first assess if trends in stimulant utilization and pretreatment screening for CV risk changed between 2001 to 2008 and inspect the impact of three critical events: (1) Health Canadas announcement of suspending the approval for Adderall XR in February 2005; (2) the FDA Drug Safety and Risk Management (DSRM) Advisory Committee

PAGE 17

17 recommendation of a black box warning for stimulants in February 2006; and (3) the FDA mandate for the distribution of MedGuides in February 2007. Specifically, this study evaluates the changes relevant to stimulant CV safety warnings from the following perspectives : Whether the proportion of ADHD patients who were exposed to stimulants and the intensity of stimulant treatment (i.e., the prescribed dosage ) have changed Whether the proportion of ADHD patients who switched to secondline treatment options (i.e., bupropion, tricyclic antidepressant, atypical antipsychotics) changed and what alternative pharmacological treatments were chosen if stimulants were discontinued; Whether the proportion of patients who underwent pretreatment screening for CV risk (i.e., electrocardiogram [ECG]) has changed. The second part of this study investigates if the change in stimulant use differed by patients socioeconomic status or clinical characteristics, or by physician specialty. Study Significance To our knowledge, this is the first study to investigate how ADHD pharmacological treatment utilization changed since the discussion about stimulant safety issues arose. Determining how physicians and patients responded to safety warnings related to stimulants will bring insights in the effectiveness of risk communication in ADHD management. Research Questions and Hypotheses The rese arch questions in this dissertation are grouped into two sections. 0 refers to a null hypothesis and Ha to an alternative hypothesis.

PAGE 18

18 Part I : Testing of the Change(s) i n Stimulant U tilization R esearch question1a: What is the trend in stimulant initiation from 2001 to 2008? General form of hypothesis1 a : H0: there are kmin change point(s) in the trend of stimulant initiation Ha: there are kmaxR esearch question1 b: What is the trend in stimulant discontinuation from 2001 to 2008? change point(s) in the trend of stimulant initiation General form of hypothesis1 b : H0: there are kmin change point(s) in the trend of stimulant discontinuation. Ha: there are kmaxR esearch question1c: What is the trend in stimulant initial daily dose from 2001 to 2008? change point(s) in the trend of stimulant discontinuation. General Form of Hypothesis1 c : H0: there are kmin change point(s) in the trend of stimulant initial daily dose Ha: there are kmaxR esearch question1d: What is the trend in stimulant maintenance daily dose from 2001 to 2008? change point(s) in the t rend of stimulant initial daily dose General form of hypothesis1 d H0: there are kmin change point(s) in the trend of stimulant maintenance daily dose Ha: there are kmaxR esearch question1e: What is the trend in treatment switching to 2 change point(s) in the trend of stimulant maintenance daily dose nd line ADHD drug from 2001 to 2008?

PAGE 19

19 General form of hypothesis1 e : H0: there are kmin change point(s) in the trend of treatment switching to 2nd line ADHD drug Ha: there are kmax change point(s) in the trend of treatment switching to 2ndR esearch question1f: What is the trend in pretreatment ECG use from 2001 to 2008? line ADHD drug General form of hypothesis1 f : H0: there are kmin change point(s) in the trend of pretreatment ECG use Ha: there are kmaxThe sequential testing started from k change point(s) in the trend of pre treatment ECG use min =0 and kmax =3. If H0 wa s rejected, the testing moved on to kmin =1 and kmax =3 ; if Ha wa s rejected the testing moved on to kmin =0 and kmax =2 ; testing continue d until the model with the lowest number of change point was determined. If kmax joinpoint model was selected, testing proceeded to kmax vs kmax+1Part II : Comparability Test among S ubgroups R esearch question2a1: Was the trend in stimulant initiation comparable between young children and older children from 2001 to 2008? R esearch question2a2 : Was the trend in stimulant initiation comparable between older children and adolescents from 2001 to 2008? R esearch question2a3 : Was the trend in stimulant initiation comparable between female and male patients from 2001 to 2008? R esearch question2a4 : Was the trend in stimulant initiation comparable between Blacks and Whites from 2001 to 2008? R esearch question2a5 : Was the trend in stimulant initiation comparable between Hispanics and Whites from 2001 to 2008?

PAGE 20

20 R esearch question2a6 : Was the trend in stimulant initiation comparable between children eligible due to SSI status and children eligible under other plans from 2001 to 2008? R esearch question2a7 : Was the trend in stimulant initiation comparable between foster care children and children eligible under other plans from 2001 to 2008? R esearch question2a8 : Was the trend in stimulant initiation comparable between patients with mental comorbidity and patients without mental comorbidity from 2001 to 2008? R esearch q uestion2a9 : Was the trend in stimulant initiation comparable between patients diagnosed by primary care physicians and patients diagnosed by psychiatrists from 2001 to 2008? General form of hypothses 2a1a to 2a9a (for testing parallelism): H0: The stimulant initiation trends are parallel between the two groups (i.e., share the same change point[s] and the same slope[s] in each segment) HaGeneral form of hypotheses 2a1b to 2a9b (for testing coincidence) : H : The stimulant initiation trends are not parallel between the two groups 0: The stimulant initiation trends are identical between the two groups (i.e., share the same change point[s] and the same slope and interception in each segment) HaR esearch question2b1: Was the trend in stimulant discontinuation comparable between young children and older children from 2001 to 2008? : The stimulant initiation trends are not identical betw een the two groups R esearch question2b2 : Was the trend in stimulant discontinuation comparable between older chi ldren and adolescents from 2001 to 2008?

PAGE 21

21 R esearch question2b3 : Was the trend in stimulant discontinuation comparable between female and male patients from 2001 to 2008? R esearch question2b4 : Was the trend in stimulant discontinuation comparable between Blacks and Whites from 2001 to 2008? R esearch question2b5 : Was the trend in stimulant discontinuation comparable between Hispanics and Whites from 2001 to 2008? R esearch question2b6 : Was the trend in stimulant discontinuation comparable between childr en eligible due to SSI status and children eligible under other plans from 2001 to 2008? R esearch question2b7 : Was the trend in stimulant discontinuation comparable between foster care children and children eligible under other plans from 2001 to 2008? R esearch question2b8 : Was the trend in stimulant discontinuation comparable between patients with mental comorbidity and patients without mental comorbidity from 2001 to 2008? General form of hypothses 2b1a to 2b8a (for testing parallelism): H0: The sti mulant discontinuation trends are parallel between the two groups (i.e., share the same change point[s] and the same slope[s] in each segment) HaGeneral form of hypotheses 2b 1b to 2b8b (for testing coincidence) : H : The stimulant discontinuation trends are not parallel between the two groups 0: The stimulant discontinuation trends are identical between the two groups (i.e., share the same change point[s] and the same slope and interception in each segment) Ha: The stimulant discontinuation trends are not identical between the two groups

PAGE 22

22 Research question2c1: Was the trend in stimulant initial daily dose comparable between young children and older children from 2001 to 2008? Research question2c2: Was the trend in stimulant initial daily dose comparable between older children and adolescents from 2001 to 2008? Research question2c3: Was the trend in stimulant initial daily dose comparable between female and male from 2001 to 2008? Research question 2c4: Was the trend in stimulant initial daily dose comparable between Blacks and Whites from 2001 to 2008? Research question2c5: Was the trend in stimulant initial daily dose comparable between Hispanics and Whites from 2001 to 2008? Research question2c6: Was the trend in stimulant initial daily dose comparable between children eligible due to SSI status and children eligible under other plans from 2001 to 2008? Research question2c7: Was the trend in stimulant initial daily dose comparable between foster care children and children eligible under other plans from 2001 to 2008? Research question2c8: Was the trend in stimulant initial daily dose comparable between patients with mental comorbidity and patients without mental comorbidity from 2001 to 2008? Research question2c9: Was the trend in stimulant initial daily dose comparable between patients diagnosed by primary care physicians and patients diagnosed by psychiatrists from 2001 to 2008? General form of hypothses 2c1a to 2c9a (for testing parallelism): H0: The stimulant initial daily dose trends are parallel between the two groups (i.e., share the

PAGE 23

23 same change point[s] and the same slope[s] in each segment) HaGeneral form of hypotheses 2c1b to 2c9b (for testing coincidence) : H : The stimulant initial dail y dose trends are not parallel between the two groups 0: The stimulant initial daily dose trends are identical between the two groups (i.e., share the same change point[s] and the same slope and interception in each segment) HaResearch question2d1: Was the trend in stimulant maintenance daily dose comparable between young children and older children from 2001 to 2008? : The stimulant initial daily dose trends are not identical between the two groups Research question2d2: Was the trend in stimulant maintenance daily dose comparable between older children and adolescents from 2001 to 2008? Research question2d3: Was the trend in stimulant maintenance daily dose comparable between f emale and male from 2001 to 2008? Research question2d4: Was the trend in stimulant maintenance daily dose comparable between Blacks and Whites from 2001 to 2008? Research question2d5: Was the trend in stimulant maintenance daily dose comparable between Hispanics and Whites from 2001 to 2008? Research question2d6: Was the trend in stimulant maintenance daily dose comparable between children eligible due to SSI status and children eligible under other plans from 2001 to 2008? Research question2d7: Was the trend in stimulant maintenance daily dose comparable between foster care children and children eligible under other plans from 2001 to 2008?

PAGE 24

24 Research question2d8: Was the trend in stimulant maintenance daily dose comparable between patients with mental comorbidity and patients without mental comorbidity from 2001 to 2008? General form of hypothses 2d1a to 2d8a (for testing parallelism): H0: The stimulant maintenance daily dose trends are parallel between the two groups (i.e., share the same change point[s] and the same slope[s] in each segment) HaGeneral form of hypotheses 2d1b to 2d8b (for testing coincidence) : H : The stimulant maintenance daily dose trends are not parallel between the two groups. 0: The stimulant maintenance daily dose trends are identical between the two groups (i.e., share the same change point[s] and the same slope and interception in each segment) HaResearch question2e1: Was the trend in stimulant treatment switching comparable between young children and older children from 2001 to 2008? : The stimulant maintenance daily dose trends are not identical between the two groups Research question2e2: Was the trend in stimulant treatment switching comparable between older children and adolescents from 2001 to 2008? Research question2e3: Was the trend in stimulant treatment switching comparable between female and male from 2001 to 2008? Research question2e4: Was the trend in stimulant treatment switching comparable between Blacks and Whites from 2001 to 2008? Research question2e5: Was the trend in stimulant treatment switching comparable between Hispanics and Whites from 2001 to 2008?

PAGE 25

25 Research question2e6: Was the trend in stimulant treatment switching comparable between children eligible due to SSI status and children eligible under other plans from 2001 to 2008? Research question2e7: Was the trend in stimulant treatment switching comparable between fost er care children and children eligible under other plans from 2001 to 2008? Research question2e8: Was the trend in stimulant treatment switching comparable between patients with mental comorbidity and patients without mental comorbidity from 2001 to 2008? General form of hypothses 2e1a to 2e8a (for testing parallelism): H0: The stimulant treatment switching trends are parallel between the two groups (i.e., share the same change point[s] and the same slope[s] in each segment) HaGeneral form of hypotheses 2e1b to 2e8b (for testing coincidence) : H : The stimulant treatment switching trends are not parallel between the t wo groups 0: The stimulant treatment switching trends are identical between the two groups (i.e., share the same change point[s] and the same slope and interception in each segment) HaResearch question2f 1: Was the trend in pretreatment ECG use comparable between young children and older children from 2001 to 2008? : The stimulant treatment switching trends are not identical between the two groups Research question2f 2: Was the trend in pretreatment ECG use comparable between older children and adolescents from 2001 to 2008? Research question2f 3: Was the trend in pretreatment ECG use comparable between female and male from 2001 to 2008?

PAGE 26

26 Research question2f 4: Was the trend in pretreatment ECG use comparable between Blacks and Whites from 2001 to 2008? Research question2f 5: Was the trend in pretreatment ECG use comparable between Hispanics and Whites from 2001 to 2008? Research question2f 6: Was the trend in pretreatment ECG use comparable between children eligible due to SSI status and children eligible under other plans from 2001 to 2008? Research question2f 7: Was the trend in pretreatment ECG use comparable between foster care children and children eligible under other plans from 2001 to 2008? Research question2f 8: Was the trend in pretreatment ECG use comparable between p atients with mental comorbidity and patients without mental comorbidity from 2001 to 2008? Research question2f 9: Was the trend in pretreatment ECG use comparable between patients diagnosed by primary care physicians and patients diagnosed by psychiatris ts from 2001 to 2008? General form of hypothses 2f 1a to 2f 9a (for testing parallelism): H0: The pretreatment ECG use trends are parallel between the two groups (i.e., share the same change point[s] and the same slope[s] in each segment) HaGeneral form of hypotheses 2f 1b to 2f 9b (for testing coincidence) : H : The pretrea tment ECG use trends are not parallel between the two groups 0: The pretreatment ECG use trends are identical between the two groups (i.e., share the same change point[s] and the same slope and interception in each segment) Ha: The pretreatment ECG use trends are not identical between the two groups

PAGE 27

27 Tabl e 11. R egulatory actions/discussions on stimulant cardiovascular safety Year Month Event 2005 February Health Canada suspends Adderall XR sale Aug ust Adderall XR reenters Canadian market 2006 February FDA Drug Safety & Risk Management Advisory Board proposes boxed warning Mar ch FDA Pediatric Advisory Board votes against proposal for boxed warning May FDA advises labeling change for all stimulants 2007 February FDA requires stimulant medication guide (MedGuide)

PAGE 28

28 Tabl e 12 E vidence regarding safety concerns in recent drug utilization studies Drug/drug class Adverse event Source of evidence Cisapride (P ropulsid) Serious drug drug interactions (i.e., cardiac arrhythmias and death ) with macrolide antibiotics or imidazole antifungals 341 case reports, including 80 deaths. Estrogen+progesteron (Hormone replacement therapy) Breast cancer and CV event (i.e., heart attacks, strokes, R andomized control trial venous thromboembolism) SSRI antidepressants Suicidality and suicidal death Meta analysis of randomized control trial Rosiglitazone (Avendia) Myocardial ischemic event (i.e., angina, myocardial infarction) Meta analysis of randomized control trial

PAGE 29

29 CHAPTER 2 LITERATURE REVIEW This ch apter is divided into two parts, first, a n overview of the role of stimulants in ADHD treatment, determinants of stimulant utilization, and stimulant cardiovascular risk ; second, A review of existing research on the influences of new drug safety information on drug utilization pattern Part I The R ole of Stimulants in ADHD Treatment ADHD is a common child psychiatric disorder characterized by a persistent pattern of impulsiveness and inattention, with or without hyperactivity.16 It is a chronic di sorder with 30 to 50% of individuals diagnosed in childhood continuing to have symptoms into early adulthood.17 ADHD is associated with social, educational, occupational, and interpersonal difficultie s as well as a higher risk for accidents,18 and psychiatric comorbidities. Pharmacotherapy, sp ecifically with central nervous system stimulants or atomoxetine, and/or behavioral interventions are the most recommended treatment options for ADHD.19 The efficacy of stimulants in treating ADHD has been documented in randomized controlled trials. A metaanalysis found that the use of stimulants improved teachers' and parents' ratings of disruptive behavior; however it did not improve academic achievement.20 Stimulants neither increased nor decreased rates of delinquency or substance abuse at 3 years of treatment.20 No significant differences betw een the various drugs in terms of efficacy or side effects have been found.21, 22 Stimulants have been found to be safe in clinical trials,23 though, no study has assessed the adverse effect of treatment beyond two years and assessments have usually occurred in small samples.20

PAGE 30

30 Stimulant Utilization among ADHD Patients Amphetamines and amphetaminelike stimulants were introduced in the 1940s (Table 21). Stimulants are now use d chronically in more than five percent of American children and in a rapidly growing number of adults.9, 24 Longitudinal comparisons suggest that both the diagnosis and treatment of childhood ADHD have continued to increase over the last decade.25, 26Factors associated with stimulant utilization According to the latest CDC (center of disease control) survey, only 56% of ADHD patients were reported to be taking medication for this disorder.9 Patient characteristics Initial decision regarding utilization of stimulant treatment has been found to be based on patients clinical conditions, and also be influenced by patients sociodemographic characteristics and provider characteristics. The probability of stimulant initiation was found to decrease if the patient has a comorbid mental condition. Importantly, the presence of a more complex mental health diagnosis, such as bipolar disorders, schizophrenia, or autism, showed a more significant impact on the likelihood for stimulant use.27 In contract, the pres ence of hyperactivity increased the probability of stimulant initiation.27Stimulant initiation is higher among male, white, and school age children. 27 The disparities by race/ethnicity may relate to differences in cultural constructs regarding deviant behavior, helpseeking, and appropriate treatments 2831. Stimulant use prevalence for y ouths is lowest in the western US and highest in the South.9 Patients living in rural areas we re found to be more likely to start stimulant treatment than those who lived in urban areas.27

PAGE 31

31 Children without insurance show lower levels of stimulants use relative to children with insurance.32 Some studies have reported that stimulant use is higher among Medicaid than among privately insured youths;33 while another study found that stimulant use is more likely among privately insured children than their publicly insured counterparts, once an ADHD diagnosis has been established.32 These studies did not adjust for patients' demographic or clinical characteristics, as well as, administrative differences in the insurance plan (i.e. the proportion of patients under capitated or feefor service arrangements). How insurance type predicts stimulant initiation is still controversial. Among Medicaid beneficiaries, those who were eligible due to foster care had the highest propensity to initiate stimulants, compared to those in TANF or SCHIP programs.27 Provider characteristics Youths who were diagnosed by psychiatri sts were less likely to receive ADHD drugs than those diagnosed by primary care physicians.27Factors associated with stimulant persistence Early stimulant discontinuat ion is common in the community care of ADHD.34, 35 One study revealed that 54% of children who were prescribed stimulants received only 1 prescription, 19% received 2 prescriptions, 16% received 3 or 4 prescriptions, and only 11% of the children received 5 or more prescriptions during a 1year period.34 The persistence of stimulant treatment varies greatly among ADHD youths with only half of new users receiving stimulant after the first year of treatment, yet, another 17% continuing for 5 years or more.26Decision on stimulant continuation has been associated mostly with patients socio demographic or clinical characteristics or their drug regimens. Younger age,

PAGE 32

32 Whites, foster care status or receiving Medicaid due to disabilities predicts greater treatment persistence.3639 The racial/ethnic difference is believed to reflect cultural variation in parental acceptability regarding stimulant treatment.28 Fewer AD HD symptoms, absence of hyperactivity, as well as, presence of ADHD family history or certain mental comorbidities (i.e., bipolar disorder, tic disorder, oppositional defiant disorder) determine lower treatment persistence.36, 38, 40 High starting dose, use of extendedrelease formulations or other psychotropic co medications increases stimulant persistence.38, 41, 42Stimulants and Cardiovascular Risk Stimulants are closely related to sympathomimetic biological amines (i.e., e pinephrine, norepinephrine). The cardiovascular effects (i.e., increase heart rate and stroke volume, constrict arterioles) of sympathomimetic amines have been thoroughly described in the medical literature. As for manifestation of CV disease, case reports of sudden death, stroke, myocardial infarction and cardiomyopathy has been associated with stimulant use.4346 Stimulants demonstrate increases in blood pressure and heart rates in clinical trials, 4752 but are t ypically described as mild, of short duration, and responsive to dosing or timing adjustments.22, 53 A large scale observational study, consisting of ~55,000 youths (~125,000 personyear observation) among 3 to 20 years olds who were newly diagnosed with AD HD, reported that stimulant use was associated with an increase in emergency department and physician office visits for cardiac symptoms, such as syncope, tachycardia, or palpitations.54 This study also found the incidence rates of cardiac events requiring hospitalization (i.e., myocardial infarction, hypertensive disease, angina, aortic or thoracic aneurysm and arrhythmia) to be small and similar to national background rates, suggesting that a manifestation of severe

PAGE 33

33 heart disease secondary to cardiac symptoms is likely rare within the age group and followup time period studied. Nevertheless, it is still not clear whether stimulants can cause manifestation of severe heart disease in adult populations or chronic drug users. Part II New Drug Safety Information and Patients or Health Care Providers B ehavior Product label changes and Dear healthcare professional letters to disseminate information regarding drug labeling changes have long been the most common way for the FDA (Table 22) to communicate new drug safety issues to healthcare providers in the U.S. In recent years, new information on drug safety has spread through methods that are more approachable to the general public, 55 such as regulatory public health advisories or media ( refers to principle means of communicating to the general public, such as newspapers, radio, television, magazines, journals the internet, etc.) Dear healthcare professional letters reports. Presently, only a handful of studies in the literature inform on the impact of communicating new drug safety information on drugtaking or prescribing behavior. Some case studies have evaluated the effectiveness of Dear heal thcare professional letters on drug utilization patterns; the others have examined changes in drug utilization pattern after widely circulated safety warnings. The literature generally favors the conclusion that Dear healthcare professional letters are not particularly effective in changing prescribing behavior among physicians. Two U.S. studies quantified the impact of cisapride (Propulsid, Ortho McNeil Janssen) on prescribing pattern following labeling changes and Dear healthcare professional letters indicating serious drugdrug interactions with macrolide antibiotics or imidazole antifungals.40, 56 The background contraindicated use rate was found to be 45% in

PAGE 34

34 prevalent cisapride users or 2660% in new users, depending on selection of study populations and definition of contraindicated use. The warnings only demonstrated a 12% crude reduction, or 325% relative reduction, annually in the rate of this potentially fatal drug use. Another study examined the impact of four Dear healthcare professional letters to provide information about trogli tazone (Rezulin, Pfizer) label changes regarding the recommendation for periodic liver enzyme monitoring during the course of treatment.57Public regulatory warnings and/or media reports of new safety information This study found that the baseline testing (i.e., the testing before drug initiation) among troglitazone users increased from 15% before FDA monitoring recommendations to 44.6% follo wing the four letters, though, less than 5% of users received the adequate follow up liver enzyme tests by the third month of continuous use. Case studies on widespread new drug safety alerts reveal that physicians and patients often respond quickly to the publicity of new safety information. The Women's Health Initiative (WHI) Estrogen Plus Progestin Trial demonstrated that standarddose conjugated estrogens/medroxyprogesterone acetate (Prempro Wyeth Pharmaceuticals Inc ) resulted in a slightly increased risk of cardiac adverse events. Following the publication of WHI trial results in July 2002, hormone therapy prescriptions declined in successive months: relative to January June 2002, prescriptions from January June 2003 declined by 66% for the conjugated estrogens/medroxyprogesterone acetate.58 In 2003, after the FDA issued a public health advisory about selective serotonin reuptake inhibitors (SSRIs) related suicidal risk in the pediatric population, utilization of SSRIs among depressed patients (both pediatric and adult cases) was reduced by 58% in the following 24 months.59, 60 Likewise, a meta analysis reported increased cardiovascular

PAGE 35

35 risk associated with rosiglitazone in May, 2007 and resulted in a black box warning for heart failure. The average number of claims (per day per million people) for rosiglitazone began to decrease immediately, falling to 41.0 in December 2007, for a total decrease of 58.6% from the February 2007 peak (99.1).61Factors M ediate the Impact of New Drug Safety Information on Drug U tilization Factors found to be associated with health services utilization can be categorized as individual, institutional and environmental factors (Table 23).62, 63 Over the past 30 years, there has been considerable interest in integrating those factors into conceptual models to guide the conduct of research for understanding who uses health care services and who does not and why.64 According to Williams and Torrens,64The systems approach is represented to a considerable extent by the Behavioral Model of Health Services Utilization introduced by Ronald Andersen in 1968. those models can be broadly divided into two categories: system and patient decisionmaking models. The system framework represents a macro perspective and usually tries to address whether services are fairly or equitably distributed or how a change in the environment (e.g., regulatory change) impacts the delivery of health care; the patient decisionoriented frameworks represents a micro perspectiv e and focus on illustrating the psychosocial dynamics underlying decisions to seek medical care. 65, 66 It focuses on p redisposing (i.e., demographic, social structure and health beliefs), enabling (i.e., personal/family, community, insurance coverage), and need (i.e., perceived, evaluated) factors as variables explaining the use of health services.65 It views utilization as conditioned on individuals beliefs about medical care, and their need for help, their access to economic and geographic resources and their subjective evaluation of the potential outcomes of their health care use.67 Andersons model

PAGE 36

36 integrated a range of institutional and individual factors to explain decisions to seek care. It has been criticized, however, for not recognizing the importance of social and cultural practices and attitudes in health service utilization.68, 69 More recently, researchers have begun to develop models to incorporate environment al or, sociocultural, influences on health service use.69 The Network Episode Model (NEM), proposed by Bernice Pescosolido in early 1990s, proposes social networks, within the community (the social support system ) or within the health care system (the treatment system ), as the mechanism through which individuals recognize and evaluate health problems, determine when and where to seek professional medical help, and decide whether or not to comply with medical advice;63, 68, 69 the structure of networks (defined as characteristics that describe the form of the network or the geometry of ties, such as size, frequency of contact, multiplicity, density, strength of tie and scope/range) calibrates the amount of social influence and the context of networks ( defined as the characteristics that describe the substance of the network, the things that flow across the ties. Examples of this concept include positive or negative valence, attitudes and belief s held and cultural meanings) determined the direction of the force, i.e., either toward or away from the formal medical system.63The Health Belief Model (HBM) is one of the patient decision making models that have been subject to considerable empirical testing. 64 In the HBM, a variety of diverse demographic, sociopsychological, and structural factors may influence behavior.70, 71 They are, however, believed to work through their effects on the individuals subjective perceptions and motivations (beliefs), rather than functioning as direct causes of the behavior themselves.64 This model proposes that individuals general and specific

PAGE 37

37 health beliefs (e.g. willingness to seek and accept medical direction, beliefs about the severity of symptoms), preferences (e.g. perceived benefits of treatment), experiences (e.g. with probl ems and providers), and knowledge about the health problem in question and types of treatment affect their decision to seek care and their health behavior .72Both system and patient decision models provide complementary views on factors that influence utilization behavior, 64From the macro perspective, Guay et.al. nevertheless, empirical support on using either types of model to identify factors underlying the impact of new drug safety information on drug utilization is scarce. 73 identified effect modification of the influence of WHI publications on hormone replaced therapy (HRT) persistence from patients socioeconomic status and drug regimen. Women who were on social assistance after the release of WHI publication were less lik ely to stop HRT than those in the preWHI publication period; instead, women who used high dose HRT in the post WHI publication period were more likely to cease HRT than those in the preWHI publication period. Several studies implied effect modification f rom the specialty of health care provider. After the warning on SSRI related suicidal risk, the drug utilization rate among depressed youths who were seen by generalists (i.e. pediatricians and primary care physicians) decreased more pronouncedly than those who were seen by psychiatrists.59, 60Gerend et.al. 74 conducted the only study that applied the patient decision making model (the HBM framework) to explore factors associated with HRT use before and after the WHI study. The proposed model demonstrated adequate fit to data. The study supported that, after the release of WHI findings, the relative role of cognitive factors

PAGE 38

38 perceived benefits versus perceived barriers in HRT use decisionmaking was changed; whereas perceived benefits was the stronger predictor of intentions in preWHI periods, benefits dropped away as a predictor of post WHI HRT use and barriers moved to the forefront. Women who were older or who expressed higher perceived barriers to HRT use were more likely to be motivated by the announcement of WHI findings to stop HRT use. In summary, previous research has provided evidence on the influence of some patients sociodemographic characteristics, psychological factors and drug regimen on the effect of new drug safety information on drug utilization. This proposed study will further examine how ot her factors, particularly patients clinical characteristics (i.e. comorbidities and co medications) and physician specialty, mediate such change. Medication Guides Of particular interest in the study at hand is the fact that the increased risk is supported by weaker evidence than any of the previous therapeutic eras that were examined in the above mentioned studies. How decisions are made and to what extent utilization changes when evidence is heavily debated has not been investigated.

PAGE 39

39 Table 21 Lists of stimulants approved for ADHD Source: Drugs@FDA http://www.accessdata.fda.gov/scripts/cder/drugsatfda/index.cfm Last accessed June, 2010. Brand n ame Generic n ame Approved a ge Approved year Manufacturer Adderall Amphetamines old 1960 Duramed Research Inc Adderall XR A mphetamines (extended release) old 2001 Shire Concerta M ethylphenidate (long acting) old 2000 Ortho McNeil Janssen Daytrana M ethylphenidate patch old 2006 Shire Desoxyn M ethamphetamine hydrochloride old 1943 Ovation Pharms Dexedrine Dextroamphetamine old 1976 SmithKline Beecham Dextrostat Dextroamphetamine old 1975 Shire Focalin Dexmethylphenidate old 2001 Novartis Focalin XR Dexmethylphenidate (extended release) old 2005 Novartis Metadate ER M ethylphenidate (extended release) old 1999 UCB Inc Methylin ER M ethylphenidate (extended release) old 2000 Mallinckrodt Methylin M ethylphenidate (oral solution and chewable tablets) old 2002 Mallinckrodt Ritalin Methylphenidate old 1955 Novartis Ritalin SR M ethylphenidate (extended release) old 1982 Novartis Ritalin LA M ethylphenidate (long acting) old 2002 Novartis Strattera Atomoxetine old 2002 Lilly Vyvanse L isdexamfetaine dimesylate old 2007 New River Pharms

PAGE 40

40 Table 22 Summary of FDA communication methods of drug safety information Source: FDA Guidance on Communication of Drug Safety Information, http://www.fda.gov/cder/guidance/7477fnl.pdf last accessed June, 2010. Type of communication Content Target audience Labeling Summary of essential information needed for safe and effective use of the drug. Healthcare providers Dear healthcare professional letter I nformation regarding a significant hazard to health, an important change in product labeling, or a correction to drug advertising or labeling. Healthcare providers Patient directed labeling (i.e., patient package insert or Medication Guides) Summary of essential information needed for safe and effective use of the drug. (in nontechnical language) Patients Public health advisory Information and advice regarding an emerging drug safety issue or other important public health information General Public Patient information sheets Concise summary in plain language of the most important information about a drug that is a new molecular entity or a marketed drug with new safety information. Patients and/or lay caregivers, and interested members of the general public Healthcare professional sheets Concise summary of an important, and often emerging, drug safety issues, including background information about the detection of the issue and points to consider for clinical decision making Healthcare professionals Alerts on patient information and healthcare professional sheets Summary of an important, and often emerging, drug safety issue, including a statement that reflects the stage of the analysis with respect to regulatory decision making or other potential limitations on the interpretation of the safety information. Healthcare professionals, patients and/or lay caregivers, and interested members of the general public

PAGE 41

41 Table 23 Major factors affecting health services utilization Source: McKinlay JB. Some approaches and problems in the study and use of services -and overview. J Health Soc Behav 1972;13:115152; Williams JW, Torrens PR: Introduction to health services New York: Wiley; 1988; a nd Pescosolido BA: Of pride and prejudice: the role of sociology and social networks in integrating the health sciences. J Health Soc Behav Sep 2006;47(3):189208. Attributes Variables Individual factor Patient characteristics Sociodemographic: g ender, age, marital status, family size, residence, social class, ethnicity, education, occupation E conomic : Income, insurance coverage P sychological factors : Health beliefs, values, attitudes C linical characteristics H istory of illness, severity Provider characteristics Age/recency of professional training, gender, specialty Institutional factor Treatment system characteristics A ccessibility of care, financing of care Environment factor Treatment network structure Size, density, strength of tie (e.g., relationship between physician and patient, interaction between physician and medical professional and health care system) Treatment context P rovider & staff attitudes, and culture toward health, clients community and treatment organization Social support system structure Size, density, strength of tie Social support system context B eliefs and attitudes toward health or professional medical care ( i.e. perceived efficacy)

PAGE 42

42 CHAPTER 3 METHODOLOGY Study Design A time series design was used to describe changes in trends of several measures of stimulant utilization between 2001 and 2008. The study included 8 years of claims data for Florida Medicaid beneficiaries, segmented into 96 months. Joinpoint regression models were use to detect the number of change points, to estimate the magnitude of change in monthly outcome trends, and to test comparability of trend change(s) among subgroups, defined by several patient and provider characteristics. Data Source The dataset was assembled from the Florida Medicaid feefor service program and include d monthly subject specific information on eligibility and demographic information, as well as all healthcare claims submitted for The validity of Medicaid claims data with regard to measuring the incidence of ADHD diagnosis or stimulant use has not been investigated. However, it has been recognized that the validity of Medicaid pharmacy claims data in capturing psychotropic exposure is good, with reported percent agreement, positive predictive value an d negative predictive value over 85%. reimbursement. The dataset covered t he years 2001 2008, thus including a roughly equal time frame (4 years) before and after th e first publicized concern related to stimulant CV safety (i.e. Canadas withdrawal of Adderall XR in February, 2005) 75 According to this study the validi ty in capturing dosage for psychotropics was also 0.81.75ADHD is diagnosed predominant ly among children and adolescents. The pediatric population is often under represented in national surveys that allow investigating

PAGE 43

43 medication utilization (e.g., the National Ambulatory Medical Care Survey [NAMCS], or the Medical Expenditure Panel Survey [MEPS]). As Florida Medicaid Fee for service recipients represent 18% of the youth under 18 years residing in the state, the size of the represe nted population is the greatest strength of our dataset. Study Population Two cohorts w ere extracted from the dataset. A cohort of patients with a new ADHD episode was used to assess trend in stimulants initiation; the second cohort of new stimulant treatm ent episode was used to investigate stimulant discontinuation, treatment switching, and treatment intensity (i.e., initial dose, average daily dose) and pretreatment screening for cardiac risk (i.e., electrocardiogram [ECG]) Assembly of these cohorts is described below. Cohort of New Episode ADHD Patients All beneficiaries aged 064 years with new ADHD diagnosis between 01/2001 and 12/2008 were included. Diagnosis of ADHD was determined based on the presence of at least two inpatient or outpatient claims with International Classification of Disease Clinical Modification Version 9 (ICD9 CM) code 314.xx (hyperkinetic syndrome of childhood) during the study period. Their first listed ADHD diagnosis date was identified as the index date. A minimum of six mont hs continuous Medicaid eligibility before the ind ex date was required to allow identification of a new ADHD episode. These six months could not include ADHD diagnosis or any stimulant claims. Patients were also required to have more than 30 days continuous insurance coverage after the index date to provide sufficient follow up time to determine stimulant initiation.

PAGE 44

44 The requirement of having at least two inor out patient services with ADHD diagnosis in order to be categorized as ADHD cases attempted to reduce misclassification in the cohort. Cohort of New Stimulant Users ADHD patients who received stimulant treatment between 01/2001 and 12/2008 were identified from the datasets. New episode users are those who had a stimulant prescription (defined based on National Drug Code [NDC], see Appendix A ) pr e ceded by at least six months stimulant free period. Patients contributed only the earliest episode to the analysis. The first listed prescription filling date was identified as the index date. A minimum of six months continuous insurance eli gibility before the index date was required to allow identification of prior drug exposure. The analyses excluded patients who received atomoxetine (N=2,020) since atomoxetine had a separate warning on suicidal risk in 200576 Study Measures a nd is commonly classified as nonstimulant. Thus, utilization patterns were expected to be different from the rest of ADHD drugs. However, atomoxetine utilization was plotted separately to explore whether changes in utilization might have affect stimulant use. Patients who had missing or zero value in the days supply fi el d (N=1,572) were also excluded since it prohibited the calculation of several study measures Stimulant Initiation Stimulant initiation was defined as having at least one pharmacy claim for central nervous stimulants (i.e. excluding atomoxetine) within 30 days of the first listed ADHD diagnosis date.

PAGE 45

45 Stimulant Discontinuation Treatment di scontinuation occurred if a patients proportion of days covered (PDC)2 for stimulants was lower than 40%3For the analysis of treatment discontinuation, s timulant users were required to have at least six months continuous insurance coverage after treatment initiati on for calculating the metric. Patients only contributed the first treatment episode in the analysis as the probability of discontinuation differ s among treatment episodes. PDC is calculated by dividing the number of days of medication supplied (numerator) by the number of days (denominator) in a given time interval. For this study, the numerator was the sum of days of medication supplied on each prescription identified from drug initiation through six months after the first prescription date. The denominator was 180. For prescriptions written near the end of the observa tion window that had more days supply than were left in the observati on window, the days supplied were counted as only the number of days between that prescription date and the end of the observation window. ADHD Treatment Switching Treatm ent switching was defined as a drug claim that was followed by a 45day interval without a refill or a claim for a new stimulant, but a new claim for a secondline ADHD treatment ( i.e., bupropion, desipramine, imipramine or nortriptyline, see Appendix B and C ) The length of the observa tion window was selected empirically by the distribution of the difference between refill interval (i.e. days between two consecutive stimulant claim s for the same patient) minus the days of medication supplied on the earlier stimulant claim (Table 31); if stimulant treatment was continued, it was refilled within 45 days after the end of medication supply, 90% of the time To be sure that the initiatio n of a secondline drug was truly treatment switching, the claim was further

PAGE 46

46 required to be followed by another secondline medication claim within 60 days of the end of medication supply. The 60 days interval was chosen using the same logic as above ( Tabl e 3 2) For the analysis of treatment switching stimulant users were required to have at least four months continuous insurance coverage after treatment initiation for identifying switching events Patients only contributed the first treatment episode i n the analysis as the probability of switching differs among treatment episodes. S timulant Initial Daily D ose The initial dose was identified from a patients first stimulant claim. The total quantity supplied was converted to methylphenidate (MPH) equival ents based on the World Health Organization (WHO) Anatomical Therapeutic Chemical (ATC)/Defined Daily Dose (DDD) index (see Appendix D ), and then divided by the days of supply listed on that claim. Patients who received an abnormal dose (i.e. higher than the largest recommended daily dose) were excluded from the analysis The analysis also only included users who initiated stimulants within 30 days of diagnosis since the prescription filling date would be close to the time of physicians decisionmaking. Stimulant M aintenance Daily D ose The total stimulant dose prescribed to a patient within 6 months of treatment initiation was also converted to MPH equivalent s dose, and then divided by 180 to derive the average daily dose. Patients who received an abnormal dose (i.e. higher than the largest recommended daily dose) were excluded from the analysis. U sers were required to have at least six months consecutive eligibility and continue stimulant treatment (with

PAGE 47

47 PDC greater than 40%) for at least 6 months after drug initiation to calculate maintenance dose. Pre treatment Screening (for Cardiac Risk) Receiving pretreatment screening for cardiac risk refers to having an electrocardiogram (ECG) (indentified based on Current Procedural Terminology [CPT] codes see Appendix E ) within 30 days before initi ating stimulants Patients who got their first ADHD diagnosis from a hospital, or those who were hospitalized for any causes within one month prior to first diagnosis during the same period were excluded from the analysis of pretreatment screening, since their ECG charges might be aggregated into hospitalization charges and would therefore not be detectable from our data. Data A nalysis Part I : Joinpoint Analysis Joinpoint (JP) analysis was used to detect change(s) in outcome trends. The JP model characterizes trend data by a number of continuous linear segments and change points (i.e., points at which trends change): Y= E[Y|X]= 0 + 0 X + 1(X 1)+ ++ k(X k) Where + Y = monthly stimulant outcome rate X = study month (e.g., 01/2001= 1, 02/2004=2, ..., 12/2008= 96) k a s= the unknown change points + =a for a > 0, a+ =0 for a JP analysis can identify a simplest (i.e. least change point) best fitting combination of line segments and join points through a sequential hypothesis testing and a series of permutation tests, as well as, characterize the trend (i.e. locate the change points and estimate th e trend change in each segment).77 This test algorithm has been applied to

PAGE 48

48 investigate changes in cancer incidence78 and mortality trends and antidepressants utilization rate after safety warnings on drugrelated suicidal risk .79, 80Model fitting s pecification The dependent variabl es of the analyses on stimulant initiation, stimulant discontinuation, treatment switching and ECG use are the natural logarithms of monthly outcome rates (i.e. outcome event count per 1,000 patient days for initiation or discontinuation; outcome event count per 100,000 patient days for ECG use). Data were fit ted to an independent errors join point model (i.e. no correlation between successive error in the model) assuming the outcome rates follow a Poissonlike distribution, where the variance is proportional to the mean. The dependent variables for the analyses on initial or maintenance daily dose are the natural logarithms of monthly means of prescribed MPH equi valent daily dose (per patient). D ata were fit to a heteroscedastic model with inputt ed standard error (SE), derived by dividing the standard deviation of monthly mean dose by the square root of the monthly patient count. Hypothesis t esting Due to computation efficiency considerations, t he test started from setting the possible number of change points at a maximum of 3. It began from testing: H0: k0=0 (no joinpoint) vs. Ha: ka= 3 (three joinpoints) and proceeded sequentially, increasing the number of joinpoints under H0 by 1 if the null hypothesis was rejected and decreasing the number of joinpoints under the alternative hypothesis if H0 was accepted. A Bonferr oni correction was applied to ensure the probability of a type I error was less than 0.05. If the three joinpoint model were selected, a second sequential testing would be performed with the maximum change points set at 4.

PAGE 49

49 Joinpoint Software, version 3 (National Cancer Institute, Bethesda, MD) was used for data analysis. The number/ location of the probable change point(s) and the monthly percent change (MPC, calculated from the estimated slope in each segment) in trends were reported. The impact of CV safety concerns on stimulant utilization was assessed by the temporal relationship between the change point(s) and the following three events: 1) Health Canada withdrew Addreall XR in 02/ 2005; 2 ) FDA DSRM Advisory Board proposed boxedwarning in 02/ 2006; 3 ) FDA required a medication guide (MedGuide) in 02/ 2007. Adjustment for seasonality A n initial ins pection confirmed seasonality in stimulant initiation and discont inuation trends. A s suggested in the literature patients are likely to stop taking stimulant s during summer breaks, which violated the assumption of heteroscedastic data. Therefore, the trends were deseasonalized by the following steps before fitting the model: Step 1: Let Y be the mean of the observed values (of the outcome) per calendar month during the study period. Step 2: Rank the study months and let X equal to the average rank per ca lendar month. For example, if January ranks are 1,13,25,37,49,61,73, and 85, then XJanStep 3: Fit = 43. 01 by an ordinary least squares model using the 12 (Y, X) pairs. T 1Step 4: Let the calendar month with the lowest Y be the reference month. Let Y ) depicts a trend due to observed time Jan, YFeb, YDec equal to (YJanYref), (YFebYref), (Y DecYref) and XJan, XFeb, XDec equal to (XJanXref), ( XFebXref), ( X DecXrefStep 5: Calculate the deseasonalized value for each study month. ) Deseaonalized value of month i = Observed value of month i Y calendar time of i (1) I (2)

PAGE 50

50 The first step reduce s the seasonal fluctuation (i.e. impact of calendar time) on the observed value. The second step adjust s for error s due to the observed time. For example, i f the re were an increasing trend over the study period 1 > 0), Y would be larger than it would have been for the months with more observed time than the reference month, and, it would be smaller for the months with less observed time than the reference month (t he situation would be reverse if the trend were decreasing [ 1Figure 31 illustrate s the original and deseasonalized initiation and discontinuation trends. < 0]) Part II Sub group A nalysis Stratified analyses were performed based on 7 patient or provider characteristics that are associated with stimulant utilization. The selection of subgroups was based on a literature review and their availability in the datasets. Since age is a significant factor for drug use and the adult population was small, people over 21 years of age were excluded in these analyses. Subgroups were obtained based on patient and provider characteristics as described below: Patient demographic characteristic Age Gender Race Medicaid eligibility type Information on dem ographic characteristics was extracted from the insurance enrollment files for the index date (i.e., the first ADHD diagnosis for model 1 and 2 or the first stimulant claim for model 36). Four age groups were defined for the study

PAGE 51

51 population, less than 5 years of age, 5 to 9 years of age, 10 to 14 years of age, and 15 to 20 years of age. B eneficiaries were classified as Caucasian (nonHispanic), Black (nonHispanic), Hispanic, or other race (American Indian, Asian, and others) according to Medicaid categories. Medicaid eligibility st atus was classified as Temporary assistance for Needy Family (TANF), foster care, Supplemental Security Income (SSI, those who met federal qualifications for assistance through disability) or other (other programs within Medicaid). Patient clinical charact eristics Presence of mental comorbidity Presence of preexisting heart conditions Patients with mental disorders other than ADHD were identified by the presence of an inpatient or outpatient claim at any time in the study period with any Preexist ing heart disease was defined as the presence of any of the following diagnoses: adjustment disorder (identified by ICD 9 CM code 309 309.80, 309.82 309.99, and 313.89), anxiety (300 300.09, 300.2 300.29, 300.3, 309.81, and 308.xx), autism (299.0x), bipolar disorder (296 296.19, 296.4 296.99, and 301.13), conduct disorder (312 313.88 and 313.9x), depression ( 296.2 296.39, 300.4, and 311.xx) learning disorders (315.xx, 307.0, and 307.9), schizophrenia (295.0x 295.9x), and tic disorders (307.2x and 307.3x). To assure that the identified conditions presented concurrently with the diagnosis of ADHD or stimulant initiation, at least one inpatient or outpatient claim had to be present in the 180 day period before the index date and another claim for the same condition observed in the 180 day period after the index date. inpatient or outpatient claim within 6 months before the index date with any of the following codes:

PAGE 52

52 diseases of the circulatory system (390. xx to 459. xx), syncope (780.2x), tachycardia or palpitation (7 85.0x, 785.1x), chest pain (786.50), congenital anomalies of the heart and other hereditary diseases that affecting the circulatory system (hereditary hemolytic anemia [282. xx], hemophilia [286.0x to 286.4x], anomalies of bulbus cordis and cardiac septal c losure [745. xx ], other congenital anomalies of heart [746. xx )], congenital anomalies of circulatory system [747.0 747.4xx], Down syndrome [758.0x], gonadal Provider characteristics dysgenesis [758.6x], and Fragile X syndrome [759.83]). Specialty of physician who diagnosed ADHD or prescribed stimulant Provider s pecialty was identified from the medical encounter claim that was closest to (but not exceeding) the index date and was categorized into one of the following: psychiatry; primary care, including fami ly practice, general practice, a nd pediatrics; or other (including unspecified). Treatment discontinuation and average daily dose were not stratified by this variable since attribution to one provider might be invalid. Model fitting S pecification Similar to the main analyses, the dependent variables were the natural logarithms of the monthly outcome rates for stimulants initiatio n, discontinuation and ECG use. Data were fit to a heteroscedastic (i.e. uncorrelated) model assuming the outcome rates follow a Poisson distribution. The dependent variables for the analyses on initial or maintenance daily dose were the natural logarithms of monthly means of prescribed MPHequi valent daily dose (per patient). D ata were fit to a heteroscedastic model with standard e rror (SE), derived by dividing the standard deviation of monthly mean dose by the square root of the monthly patient count.

PAGE 53

53 Hypothesis Testing The study population was paired as the following for testing comparability81 Male versus Female of the outcome trends: Whites versus Blacks Whites versus Hisp anics Age < 5 years old versus 513 years of age 5 13 years of age versus age 1420 years of age TANF beneficiaries versus beneficiaries with SSI status TANF beneficiaries versus beneficiaries receiving foster care Patients with mental comorbidities versus those without (pure ADHD patients) Patients with preexisting heart disease versus those without Patient diagnosis by primary care physician versus patient diagnosis by psychiatrist A parallel test was performed for each pair: H0(i.e., share the same change point[s] and the same slope[ s ] in each segment) : The outcome trends are parallel between the two groups HaIf the test failed to reject parallelism, a coincidence test was performed : The outcome trends are not parallel between the two groups H0(i.e., share the same change point[s] and the same slope and interception in each segment ) : The outcome trends are identical between the two groups Ha : The outcome trends are not identical between the two groups Joinpoint Software, version 3 (National Cancer Institute, Bethesda, MD) was used for data analysis. A maximum number of change point max was assigned in each test based on the main analyses. The null hypothesis was rejected if the pvalue was lower than 0.05. The pvalue for the comparability tests was reported. The specification of the outcome trends (i.e., number/location of the change point(s) and MPC) were reported for the pairs with significant difference.

PAGE 54

54 Table 31 Stimulant refill patterns Refill interval drug supply (day) Cumulative percentage <=0 22 0 7 57 7 15 72 15 30 84 30 45 89 45 60 91 >60 100 Table 32 Secondline ADHD treatment refill patterns Refill interval drug supply (day) Cumulative percentage <=0 23 0 7 56 7 15 69 15 30 81 30 45 87 45 60 90 >60 100

PAGE 55

55 Figure 31. O riginal and deseaonalized trends. A) stimulant initiation. B) stimulant discontinuation 0 10 20 30 40 50 60 70 80 90 100 Per 1,000 patient months Original Deseasonalized A B

PAGE 56

56 CHAPTER 4 RESULTS Table 41 summarizes the size of the selected cohorts and the duration of the study period available for the analyses. The duration varied due to different requirements for post index date continuous eligibility and the availability of specific information needed for calculati ng outcome measures (i.e. data prior to year 2002 were excluded from the analysis of discontinuation, switching, initial dose and maintenance dose trends because the days supply field in the pharmacy claim had a significant proportion of mi ssing values). For analysis of medication switching JP model fitting was infeasible d ue to the rare occurrence of switching events, and this measure was dropped from the analyses; similarly, stratified analyses on preexisting heart condition were dropped due to the low prevalence (34%) of cardiovascular disease in the study cohorts (Table 42, 48, 412, 4 16). Stimulant Initiation Trend Table 42 summarizes the patient and provider characteristics of the cohort for stimulant initiation trend analysis (N= 44,571). There was a decrease in the eligible population during the period after stimulant CV safety warnings (570 per month in 2005, and 298 per month in 2008, Table 42 ) Patient demographic characteristics were similar to what had been previously report ed with larger representation of male s, school age d children, and Whites. However, average age was higher at the end of study period, which was attributable to the increased proportion of adult patients. There was a drop in the proportion of patients who w ere diagnosed by psychiatrists over the study period, suggesting a shift of provider type in ADHD care (Table 42).

PAGE 57

57 Thirty seven percent of newly diagnosed ADHD patients (N=16,327) initiated stimulant treatment within the first month of diagnosis (Table 42). The initiation rate ranged between 240 to 540 per 1,000 patient months (Figure 41). Four joinpoints were found in the initiation trend (Table 43). Two points were located during the prewarning periods (prior to February 2005) (Figure 41). This fluc tuation was likely due to the market redistribution. We fitted the initiation trend of atomoxetine and plotted the trend against that of stimulants (Figure 42). It appears that the stimulant initiation rate was decreasing since the time (07/2002, 95% conf idence interval [CI]: 03/200102/2003) when atomoxetine entered the market (September 2002). Because of the concern on atomoxetinerelated suicidal risk, which later led to the issued of a boxed warning(09/2005), atomoxetine initiation decreased during mi d 2005 to early 2006, while the stimulant initiation rate rebounded gradually. During the post warning period (after February 2005), stimulant initiation rate went up from 370 per 1,000 patient months to 540 per 1,000 patient months (3.9% per month, 1.0, 9.1); it then dropped from 540 per 1,000 patient months to 354 per 1,000 patient months (2.4% per month, 3.6, 1.2) in 03/2007 (08/200604/2008), closely after the introduction of a required MedGuide (Figure 41; Table 44). Since the rise in stimulant initiation was close to the time a new stimulant product, Daytrana patch, entered the market (April 2006), we further examined the initiation of only oral stimulant products (Figure 43). The joinpoint model failed to find a significant increase in initi ation in 07/2006 (Figure 43, Table 45, 46), though, the raw data trends are almost identical (Figure 41; Figure 43), suggesting issues with statistical power.

PAGE 58

58 Thus, the marketing of Daytrana was not likely the only cause of the elevation in stimulant initiation in mid 2006. Subgroup analyses of the initiation trend excluded periods before 2004 to reduce the impact from change points that were not relevant to stimulant CV concerns. Table 47 shows the results of comparability tests for the stratified trends. Stimulant initiation rate was the highest among 5to 13year old children, and lowest among adolescents (14to 20year old). Prior to Canadas withdrawal of Adderall, the difference of initiation rates between 513 years and < 5 years ranged 205 0 per 1,000 patient months; it ranged 20170 per 1,000 patient months thereafter (Figure 44). In contrast, the gap between 513 years and adolescents ranged consistently 170270 per 1,000 patient months from 20042008 (Figure 45). The initiation rate did not change after CV warnings in younger children (< 5 years), but the change was significant and comparable between older children (513 years) and adolescents (Figure 44, 45). There was likely an interaction between age and gender regarding treatment initiation; the trends were identical among male and female only after excluding the younger children (Figure 46). The JP model has limited capability to handle interactions, thus, the stratified analyses of other subgroups were restricted to patients who were 5 to 20 years of age. Foster care children had the lowest stimulant initiation rate, followed by SSI children and the normal Medicaid population (TANF). Difference in treatment initiation rate between SSI children and TANF children ranged from 1080 per 1,000 patient months prior to CV warnings and 30310 per 1,000 patient months after the warnings (Figure 44). The gap between foster care children and TANF children ranged 160290

PAGE 59

59 per 1,000patient months from 2004 to 2008 (Figure 45). Changes in t reatment initiation were not comparable across patients eligibility status; safety warnings did not have an impact on initiation trend among SSI children (Figure 49, 410). Initiation rate was greater (160190 per 1,000 patient months before the warnings, 20190 per 1,000 patient months after the warnings) among patients with mental comorbidities compared to those without other psychiatric conditions, and the trend did not change among more complicated patients after the warnings (Figure 411). Stimulant s were initiated the most among whites and the least among Hispanics (difference in treatment initiation rates between Whites and Hispanics was 75540 per 1,000 patient months, Figure 48; between Whites and Blacks 150210 per 1,000 patient months, Figure 4 7), and more often among those who were diagnosed by primary care physicians (difference in treatment initiation rate ranged 400700 per 1,000 patient months, Figure 412). No significant difference by patients race or provider type was found in terms of the change in treatment initiation after the warnings (Figure 47, 4 8, 412). S timulant Discontinuation Trend Table 48 summarizes the patient and provider characteristics in the stimulant discontinuation trend analysis (N=16,372). Similar to the new e pisode patient cohort, there was a decrease in the eligible population during the period after stimulant CV safety warnings (250 per month in 2005, and 142 per month in 2008, Table 48). D emographic characteristics were similar to those for stimulant initi ation, except for age. Different from the initiation sample, which includes all patients with new ADHD episodes and which shows an increasing age over the study period, the discontinuation cohort includes all patients with newly initiated treatment and wit h continuous Medicaid

PAGE 60

60 coverage for more than 180 days post stimulant initiation. In this sample, younger age groups were increasingly represented in late study years. Thirty percent of new stimulant users (N=4,947) had a PDC lower than 40% within 180 days of treatment initiation (Table 48). Stimulant discontinuation rate ranged between 30 and 45 per 1,000 patient months and did not change after safety warnings on CV risk (Figure 413, Table 49, 410) The discontinuation rate was the lowest among 5to 13 year old children and was similar among those younger than 5 years or adolescents (Figure 414, 415). The difference in discontinuation rate was 15 per 1,000 patient months between children 513 years and the younger age group, and was 20 per 1,000 pati ent months between children 513 years and adolescents. The rate was highest among Hispanics, and lowest among Whites (Figure 417, 418), highest among children with SSI status and lowest among those in foster care (Figure 419, 420). Compared to Whites, the discontinuation rate was higher by 90 per 1,000 patient months for Hispanics and by 50 per 1,000 patient months for Blacks; Compared to TANF groups, the discontinuation rate was higher by 40 per 1,000 patient months for SSI youths and lower by 50 per 1,000 patient months for foster care children. Treatment discontinuation rate also was greater in males (by 5 per 1,000 patient months, Figure 416) and in patients with other mental comorbidity (by 4 per 1,000 patient months, Figure 421), however, there were less Whites and more Hispanics in male patients and more younger children and adolescents in those with other mental disorders. No significant difference among subgroups was found in terms of the change in treatment discontinuation after the warnings (Table 411).

PAGE 61

61 S timulant Initial Daily Dose Trend A total of 13,058 new episode stimulant users who initiated treatment within 30 days of diagnosis were eligible for the analy sis of initial dose trend (Table 412). Patient characteristics were similar to the discontinuation cohort (Table 48), except for race and mental comorbidity status Unlike the discontinuation cohort, which includes all new stimulant users with at least 6 months continuous eligibility, this cohort includes only users who initiated t reatment within first month of diagnosis, and included a higher proportion of Whites, as well as patients with no mental comorbidities. The initial dose sample also had a higher proportion of patients who were diagnosed by primary care physicians, as compared to the new episode ADHD patients (Table 42). The sample size decrease d during the period after CV safety warnings (186 per month in 2005, and 97 per month in 2008, Table 412). Pa tient characteristics were similar acro ss study periods, but fewer pati ents received stimulant treatment from a psychiatrist, which is consistent to the shift of care found in the new ADHD patient cohort. The twojoinpoint model was the final selected model (Table 413). Stimulant initial daily dose declined sharply ( 6% per month, 14; 1.9) from a steady 26mg MPH equivalent s since 05/2005 (03/200507/2007), closely after C anada withdrew Adderall XR. T he trend rebounded to a consistently lower dose (~23 m illi gram [mg]) since 09/2005 (07/200512/2005) ; close to the time when A dderall XR was reintroduced on the market ( Figure 42 2 Table 414). The initial treatment intensity did not differ by patient gender (Figure 425), race (Figure 426, 427), or mental comorbidity status (Figure 430), but increased with patients age. It was 9 11 mg/day higher among 513 years children than younger age groups before CV concerns were voiced, and 49 mg/day greater thereafter (Figure 4-

PAGE 62

62 23); it was 1215 mg/day higher among adolescents than 513 years old children prior to any CV warnings, and 1218 mg/day larger subsequently. Daily initial treatment strength among youths diagnosed by psychiatrists was slightly larger (~1 2 mg) than that among those diagnosed by primary care physicians (Figure 431). Treatment intensity among SSI children was also a higher (~2 mg) (Figure 428, 429), however, SSI children had the greatest proportion of adolescents. Change in initial dose trend after warnings was not significantly different among most groups, except that treatment intensity for younger children (age less than 5 years) or adolescents (age greater than 14 years) did not drop due to concerns on stimulant CV risk (Table 415). Stimulant Maintenance Daily Dose Trend A total of 5,222 new episode stimulant users who continued treatment for at least 6 months were eligible for the analysis of maintenance dose trend (Table 416). This cohort had similar inclusion criteria as the discontinuation cohort (Table 48), but differed from the latter in requiring more than 6month chronic stimulant use. Patient characteristics in the two samples were similar, but chronic stimulant users were slightly younger, had a higher proportion of Whites and patients with no comorbidities than the discontinuation sample. There was a decrease in the eligible population during the post CV warning period (163 per month in 2005, and 90 per month in 2008, Table 416). Chronic stimulant users demographics and clinical characteristics were also similar across study periods The threejoinpoint model was the final selected model. St imulant maintenance daily dose declined ( 2.1% per month, 2.6; 1.5) from a steady 26mg MPH equivalent s since 12/2004 (07/200406/2005), around the time when Canada withdrew Adderall XR The point estimated of the change point was earlier then the first C V safety warning with

PAGE 63

63 confidence interval that crossed the preand post warning periods. The trend rebounded to a consistently lower dose ( ~2223 mg) in early 2006 (09/200506/2006). The point estimate of this change point (04/2006) was close to the FDAs Advisory Boards debate about the proposal of a boxed warning, but the lower bound of the confidence interval (09/2005) was close to remarketing of Adderall XR (Figure 432, Table 417, 418) Maintenance treatment intensity did not differ by gender or pat ients mental comorbidity status (Figure 435, 4 40), but increased with age (Figure 433, 434). The maintenance daily dose in 5to 13 year old children was higher (~6 mg) than among younger children and was lower (810 mg) than in adolescents (Figure 433, 4 34). The intensity among Whites was lesser (~4mg/day) than in Hispanics andin Blacks (~2mg/day) (Figure 436, 437). The dose strength among foster care children was larger (~ 2 mg/day) than in children covered by other insurance types (Figure 438, 4 39). Change in stimulant maintenance dose was not significant different among subgroups (Table 419, Figure 4334 40). Pre treatment ECG Use Trend A total of 30,139 new stimulant users were eligibl e for the analysis of pretreatment ECG trends (Table 420). This cohort had similar, but more generous inclusion criteria as the initial dose cohort (Table 412). Different from the initial dose sample, which restricted to new episode ADHD patients who initiated treatment within the first months of diagnosis, the ECG use cohort includes all new episode ADHD patients who initiated stimulant treatment anytime after diagnosis and were covered by Medicaid a month before treatment initiation. Patient characteristics in the two cohorts were comparable, except that EC G use cohort had a slightly larger representation of

PAGE 64

64 Whites. There was a decrease in the eligible population during the period after s timulant CV safety warnings (342 per month in 2005, and 151 per month in 2008, Table 420). P a tient characteristics were similar acro ss study periods, but fewer patients received stimulant treatment from a psychiatrist, which is consistent to the shift of care found in the new ADHD patient cohort and the initial dose cohort (Table 42, 412). On average three percent of new s timulant users had an ECG test (N=852) before treatment initiation (Table 420). Two change points were detected in the trend (Table 421). The pretreatment ECG use rate was steady around 34 per 1,000 patient months, (range: 1658 per 1,000 patient months ) from 2000 to mid 2004; it increased by 3.2% (2.3, 4.2 ) per month since 06/2004 (07/200306/2005) and further increased by 13 % (4.2 23) per month after 02/2008 (07/200711/2008). The point estimate of the first change point was earlier than Canadas acti on (062/2005) with a confidence interval (07/200306/2005) that crosses the preand post CV warning periods (Figure 441, Table 422). The second change point was close to the release of an American Heart Association report (05/2008), which recommended al l ADHD children have an ECG screening prior to starting stimulants.82, 83Pre treatment ECG use was rare (30 events during the study period) among adolescents, and this age group was excluded from the stratified analysis. The pretreatment ECG use rate was similar across age groups and gender (Figure 442, 44 3). It was higher among Hispanics (Figure 444, 445), foster care children (Figure 446, 447), and patients with other mental disorders (Figure 448). The difference in pretreatment ECG use rate between Hispanics and Whites was 12 per 1,000 patient mont hs during the preCV warning period and was 1575 per 1,000 patient months

PAGE 65

65 thereafter (Figure 445). Prior to Canadas withdrawal of Adderall, foster care children had a steady 9 per 1,000 patient months higher screening rate than others. The difference ranged from 1050 per 1,000 patient months subsequently (Figure 447). The difference in ECG screening rates between patients with and without mental comorbidity was 2430 per 1,000 patient months before the warnings and 124 per 1,000 patient months after t he warnings (figure448). In most subgroups, the changes in trends after warnings were equivalent (Table 427, 428), however, change in the trend was sharper among patients without mental comorbidity (4.9, 3.76.0) than patients with more complicated ment al status (3.3, 1.84.8) and among patients diagnosed by psychiatrists (5.9, 4.5 7.4) than those diagnosed by primary care physicians (4.1, 2.55.7) (Table 423, Figure 448, 449).

PAGE 66

66 Table 41 Summary of analysis time frame, eligible population, and event count Study measures Time frame for analysis (d uration months ) Eligible patients (a verage patients per month) Event counts Stimulant initiation 01/2001 11/2008 (95 ) 44,571 (469) 16,327 Stimulant discontinuation 01/200206/2008 (78 ) 16,327 (209) 4,947 Switching to second line treatment 01/2002 08/2008 (80 ) 19,244 (240) 135 Initial daily dose 01/2002 11/2008 (83 ) 13,058 (157) Maintenance dose 01/2002 06/2008 (78 ) 10,635 (136) Pre treatment electrocardiography use 01/2001 11/2008 (95 ) 30,139 (317) 852

PAGE 67

67 Table 42. Patient and pr ovider characteristics in stimulant initiation trend analysis Pre warning Post warning Post warning Post warning Post warning Total 01/2001 01/2005 02/2005 12/2005 01/2006 12/2006 01/2007 01/2007 01/2008 11/2008 01/2001 11/2008 ADHD patients (n) 26,769 6,268 5,182 3,664 3,282 44,571 Follow up time (p t mo ) 19,227 4,464 3,729 2,528 2,375 31,911 Avg. patient per month (n 546 570 432 305 298 469 Stimulant use (n) 9,671 2,305 1,892 1,495 1,196 16,327 Initiation rate ( 1000 pt m 356 373 362 441 357 366 Demographics Mean age (years ) (95% CI ) 8.4 (8.4 8.5) 8.5 (8.3 8.6) 8.4 (8.2 8.6) 8.6 (8.4 8.8) 9.5 (9.2 9.8) 8.5 (8.5 8.6) < 5 y ea rs (%) 15 16 16 16 14 15 5 9 y ea rs (%) 54 56 57 58 58 55 10 14 y ea rs (%) 24 19 17 15 16 21 15 20 y ea rs (%) 6 6 6 6 5 6 >20 y ea rs (%) 2 3 3 4 6 2 M ale (%) 70 68 68 66 67 69 White (%) 43 41 43 41 42 42 Black (%) 23 24 23 22 21 23 Hispanic (%) 18 21 20 19 19 19 Other race (%) 16 15 14 17 18 16

PAGE 68

68 Table 42. Continued Pre Warning Post warning Post warning Post warning Post warning Total 01/2001 01/2005 02/2005 12/2005 01/2006 12/2006 01/2007 01/2007 01/2008 11/2008 01/2001 11/2008 Medicaid eligibility status Temporary assistance for needy family (%) 39 41 38 37 36 39 Supplemental security income (%) 20 17 17 18 19 19 Foster care (%) 11 10 13 13 11 11 Other category (%) 30 32 32 32 34 31 Clinical condition Pre existing CV condition (%) 4 4 4 5 5 4 Mental comorbidity (%) 34 36 38 33 32 35 Provider specialty Primary care physicians (%) 40 42 43 56 46 42 Psychiatrists (%) 31 21 17 11 9 25 Other physicians (%) 8 12 13 15 26 11 Missing (%) 21 25 27 18 19 22 *The rates presented were adjusted for seasonality. Abbreviations: pt mo, patient months, Avg. : a verage, CI: confidence interval, CV: cardiovascular

PAGE 69

69 Table 43. Testing results for stimulant initiation trend Hypothesis testing P value Significance level ** (H 0 )=0 joinpoint versus (H a )=4 joinpoints 0.0002 0.0125 (H 0 )=1 joinpoint versus (H a )=4 joinpoints 0.0002 0.0167 (H 0 )=2 joinpoint versus (H a )=4 joinpoints 0.0002 0.025 (H 0 )=3 joinpoint versus (H a )=4 joinpoints 0.022 0.05 *Selected model ; ** t ests were performed with the Boferroni correction at an overall significance level of 0.05. Table 44. Specification for stimulant initiation trend Segment MPC (95% CI ** Joinpoint (95% CI) ) I (01/2001 07/2002) 1.9 (0.7,3.0) 07/2002 (03/2001, 02/2003) II (07/2002 07/2003) 2.6 ( 4.7, 0.5) 07/2003 (06/2002, 05/2004) III (07/2003 07/2006) 0.4 ( 0.0, 0.8) 07/2006 (07/2003, 07/2007) IV (07/2006 03/2007) 3.9 ( 1.0, 9.1) 03/2007 (08/2006, 04/2008) V (03/2007 11/2008) 2.4 ( 3.6, 1.2) Abbreviations: MPC: m onthly percent change, CI: confidence i nterval Table 45. Testing results for oral stimulant+ Hypothesis testing initiation trend P value Significance level ** (H 0 )=0 joinpoint versus (H a )=4 joinpoints 0.0002 0.0125 (H 0 )=1 joinpoint versus (H a )=4 joinpoints 0.0002 0.0167 (H 0 )=2 joinpoint versus (H a )=4 joinpoints 0.0009 0.025 (H 0 )=3 joinpoint versus (H a )=4 joinpoints 0.168 0.05 + Ex cludes D aytrana patch ; selected model ; ** t ests were performed with the Boferroni correction at an overall significance level of 0.05 Table 46 Specification for oral stimulant+ Segment initiation trend MPC (95% CI) Joinpoint (95% CI) I (01/2001 07/2002) 1.8 (0.7,2.9) 07/2002 (09/2001, 02/2003) II (07/2002 09/2003) 2.4 ( 4.1, 0.7) 09/2003 (12/2002, 04/2004) III (09/2003 07/2007) 0.7 (0.3, 1.0) 05/2007 (02/2004, 07/2008) IV (07/2007 11/2008) 1.7 ( 3.2, 0.2) + E xcludes D aytrana patch Abbreviations: MPC: monthly percent change, CI: confidence interval.

PAGE 70

70 Table 47 Comparability test results for stratified stimulant initiation trend (01/200411/2008, 59 months) Test pair s Monthly sample size (range) P values* (Parallel test, Coincident test) < 5 years vs 5 13 years 69 (20 118) vs 313 (177 597) (0.002, ) 14 20 years vs 5 13 years 37 (14 62) vs 313 (177 597) (0.27, 0.0002) Female vs male ** 113 (53 173) vs 250 (138 362) (0.27, 0.24) Blacks vs Whites ** 72 (37108) vs 159 (69250) (0.96, 0.0002) Hispanics vs Whites ** 66 (39 94) vs 159 (69 250) (0.24, 0.0002) Supplemental security income vs Temporary assistance for needy family ** 58 (36 80) vs 141 (81 202) (0.001, ) Foster care vs Temporary assistance for needy family ** 39 (21 58) vs 141 (81 202) (0.94, 0.0002) Primary care physician vs psychiatrist ** 71 (16 126) vs 172 (88 256) (0.05, 0.0002) ADHD+ mental comorbidities vs ADHD only ** 108 (46 170) vs 255 (145 365) (0.03, ) *S ignificance level= 0.05; ** e xcluded patients less than 5 years of age.

PAGE 71

71 Table 48 Patient and provider characteristics in stimulant discontinuation trend analysis Pre warning Post warning Post warning Post warning Post warning Total 01/2002 01/2005 02/2005 12/2005 01/2006 12/2006 01/2007 01/2007 01/2008 06/2008 01/2001 06/2008 ADHD patients (n) 8,587 2,750 2,379 1,807 849 16,372 Follow up time (pt mo ) 51,522 16,500 14,274 10,842 5,094 98,232 Avg. patient per month (n) 232 250 198 151 142 210 Stimulant discontinuation (n) 2,549 871 699 538 290 4,947 Discontinuation rate (1000 pt mo ) 40 45 38 40 42 40 Demographic Mean age (years old) (95% C I ) 8.2 (8.1 8.3) 8.2 (8.0 8.4) 7.9 (7.7 8.1) 7.9 (7.7 8.1) 7.7 (7.4 8.0) 8.1 (8.0 8.2) < 5 y ea rs (%) 11 11 11 11 12 11 5 9 y ea rs (%) 61 62 66 67 71 63 10 14 y ea rs (%) 23 20 18 16 13 21 15 20 y ea rs (%) 4 4 4 4 4 4 >20 y ea rs (%) 1 2 1 1 1 1 Male (%) 71 69 70 68 69 70 White (%) 42 41 41 42 40 42 Black (%) 25 26 25 25 26 25 Hispanic (%) 17 18 21 18 18 18 Other race (%) 16 15 14 15 15 15

PAGE 72

72 Table 48 Continued Pre warning Post warning Post warning Post warning Post warning Total 01/2002 01/2005 02/2005 12/2005 01/2006 12/2006 01/2007 01/2007 01/2008 06/2008 01/2001 06/2008 Medicaid eligibility status Temporary assistance for needy family (%) 43 45 43 43 40 43 Supplemental security income (%) 18 15 16 16 16 17 Foster care (%) 9 10 11 13 13 10 Other category (%) 30 29 30 28 31 30 Clinical condition P re existing CV condition (%) 3 4 3 4 3 3 M ental comorbidity (%) 41 47 44 43 42 42 *The rates presented were adjusted for seasonality. Abbreviations: pt mo, patient months, Avg. : a verage, CI: confidence interval, CV: cardiovascular

PAGE 73

73 Table 49 Testing results for stimulant discontinuation trend Hypothesis testing P value Significance level ** (H 0 )=0 joinpoint versus (H a )=3 joinpoints 0.21 0.0167 (H 0 )=0 joinpoint versus (H a )=2 joinpoints 0.08 0.0167 (H 0 )=0 joinpoint versus (H a )=1 joinpoints 0.05 0.0167 *Selected model ; ** t ests were performed with the Boferroni correction at an overall significance level of 0.05 Table 41 0 Specification for stimulant discontinuation trend Segment MPC (95% CI) Joinpoint (95% CI) I (01/2002 06/2008) 0.0 ( 0.1,0.1) Abbreviations: MPC: monthly percent change, CI: confidence interval.

PAGE 74

74 T able 41 1 Comparability test results for stratified stimulant discontinuation trend Test pair s Monthly sample size (range) P values (Parallel test, coincident test) < 5 years vs 5 13 years 22 (21 24) vs 109 (43 176) (0.14, 0.0002) 14 20 years vs 5 13 years 15 (7 23) vs 109 (43 176) (0.43, 0.0002) Female vs male 37 (24 51) vs 104 (44 164) (0.22, 0.003) Blacks vs Whites 36 (1953) vs 58 (2988) (0.65, 0.0002) Hispanics vs Whites 25 (12 38) vs 58 (29 88) (0.64, 0.0002) Supplemental security income vs Temporary assistance for needy family 27 (9 46) vs 63 (30 97) (0.08, 0.0002) Foster care vs Temporary assistance for needy family 15 (7 23) vs 63 (30 97) (0.31, 0.0002) ADHD+ mental comorbidities vs ADHD only 56 (33 80) vs 85 (35 135) (0.13, 0.001) S ignificance level= 0.05

PAGE 75

75 Table 412. Patient and provider characteristics in stimulant initial daily dose trend analysis Pre warning Post warning Post warning Post warning Post warning Tatal 01/2002 01/2005 02/2005 12/2005 01/2006 12/2006 01/2007 01/2007 01/2008 11/2008 01/2002 11/2008 ADHD patients (n) 6866 1968 1787 1368 1069 13058 Avg. patient per month (n) 186 179 149 114 97 157 Mean dose (mg/day) (95% CI) 26.9 (26.5 27.2) 23.8 (23.2 24.5) 21.1 (20.5 21.7) 22.6 (21.8 23.3) 22.2 (21.4 23.0) 24.8 (24.6 25.0) Demographic Mean age (years old) (95% CI) 7.8 (7.7 7.9) 7.9 (7.7 8.1) 7.7 (7.5 7.9) 7.7 (7.4 8.0) 7.5 (7.2 7.8) 7.8 (7.7 7.9) < 5 y ea rs (%) 13 13 12 13 13 13 5 9 y ea rs (%) 62 65 67 68 70 65 10 14 y ea rs (%) 20 17 16 13 12 17 15 20 y ea rs (%) 3 4 3 4 3 3 >20 y ea rs (%) 1 2 2 2 2 1 Male (%) 70 67 70 65 70 69 White (%) 52 52 51 51 50 51 Black (%) 21 23 22 21 23 21 Hispanic (%) 12 12 14 13 12 12 Other race (%) 15 14 13 15 16 15

PAGE 76

76 Table 41 2. Continued Pre warning Post warning Post warning Post warning Post warning Tatal 01/2002 01/2005 02/2005 12/2005 01/2006 12/2006 01/2007 01/2007 01/2008 11/2008 01/2002 11/2008 Medicaid eligibility status Temporary assistance for needy family (%) 43 44 43 45 43 43 Supplemental security income (%) 16 14 13 13 15 15 Foster care (%) 8 8 10 11 9 9 Other category (%) 33 34 34 31 33 33 Clinical condition P re existing CV condition (%) 3 3 3 3 3 3 M ental comorbidity (%) 29 29 30 27 28 29 Provider specialty Primary care physicians (%) 61 64 65 74 63 64 Psychiatrists (%) 19 14 11 6 4 15 Other physicians (%) 8 11 13 12 25 11 Missing (%) 11 10 11 8 7 10 Abbreviations: Avg.: average, CI: confidence interval, CV: cardiovascular

PAGE 77

77 Table 4 13. Testing res ults for stimulant initial daily dose trend Hypothesis testing P value Significance level ** (H 0 )=0 joinpoint versus (H a )=3 joinpoints 0.0002 0.0167 (H 0 )=1 joinpoint versus (H a )=3 joinpoints 0.0002 0.025 (H 0 )=2 joinpoint versus (H a )=3 joinpoints 0.21 0.05 *Selected model ; ** t ests were performed with the Bonferroni correction at an overall significance level of 0.05 Table 414. Specificat ion for stimulant initial daily dose trend Segment MPC (95% CI) Joinpoint (95% CI) I (01/2002 05/2002) 0.0 ( 0.1,0.1) 05/2005 (03/2005, 07/2005) II (05/2005 09/2005) 6.4 ( 14, 1.9) 09/2005 (07/2005, 12/2005) III (09/2005 11/2007) 0.3 (0.1, 0.4) Abbreviations: MPC: monthly percent change, CI: confidence interval.

PAGE 78

78 Table 41 5. Comparability test results for st ratified stimulant initial daily dose trend Test pairs Monthly sample size (range) P values (Parallel test, Coincident test) < 5 years vs 5 13 years 19 (7 31) vs 127(73 181) (0.002, ) 14 20 years vs 5 13 years 9 (7 12) vs 115 (49 181) (0.03, ) Female vs male 71 (50 92) vs 105 (65 146) (0.14, 0.20) Blacks vs Whites 37 (15 59) vs 73 (44 102) (0.34, 0.40) Hispanics vs Whites 17 (5 30) vs 73 (44 102) (0.83, 0.84) Supplemental security income vs Temporary assistance for needy family 32 (19 46) vs 72 (46 99) (0.79, 0.03) Foster care vs Temporary assistance for needy family 31 (16 46) vs 72 (46 99) (0.45, 0.49) Primary care physician vs psychiatrist 88 (49 127) vs 28 (2 54) (0.11, 0.01) ADHD+ mental comorbidities vs ADHD only 42 (24 61) vs 111 (59 163) (0.14, 0.05) S ignificance level= 0.05

PAGE 79

79 Table 4 16. Patient and provider characteristics in stimulant maintenance daily dose trend analysis Pre warning Post warning Post warning Post warning Post warning Total 01/2002 01/2005 02/2005 12/2005 01/2006 12/2006 01/2007 01/2007 01/2008 06/2008 01/2002 06/2008 ADHD patients (n) 5522 1789 1602 1183 539 10635 Avg. patient per month (n) 149 163 134 99 90 138 Mean dose (n) (95% CI) 26.0 (25.6 26.3) 22.8 (22.2 23.4) 21.4 (20.8 22.0) 22.6 (21.9 23.4) 21.4 (20.2 22.5) 24.1 (23.9 24.4) Demographic Mean age (years old) (95% CI) 7.9 (7.8 8.0) 7.9 (7.7 8.1) 7.6 (7.4 7.8) 7.8 (7.5 8.1) 7.3 (7.0 7.7) 7.8 (7.7 7.9) < 5 y ea rs (%) 10 11 11 11 11 11 5 9 y ea rs (%) 65 65 69 68 74 67 10 14 y ea rs (%) 21 19 16 16 11 19 1520 y ears (%) 3 4 3 3 3 3 >20 y ea rs (%) 1 1 1 2 1 1 Male (%) 71 69 70 67 68 70 White (%) 45 46 44 47 44 45 Black (%) 25 25 25 24 27 25 Hispanic (%) 16 16 18 16 16 16 Other race (%) 15 13 13 13 14 14

PAGE 80

80 Table 416. Continued Pre warning Post warning Post warning Post warning Post warning Total 01/2002 01/2005 02/2005 12/2005 01/2006 12/2006 01/2007 01/2007 01/2008 06/2008 01/2002 06/2008 Medicaid eligibility status Temporary assistance for needy family (%) 43 46 43 43 43 44 Supplemental security income (%) 16 14 13 15 14 15 Foster care (%) 10 12 13 15 14 12 Other category (%) 30 29 30 28 29 30 Clinical conditions P re existing CV condition (%) 3 3 2 4 3 3 M ental comorbidity (%) 34 40 38 37 37 36 Abbreviations: Avg.: average, CI: confidence interval, CV: cardiovascular

PAGE 81

81 Table 4 17. Testing results for stimulant maintenance daily dose trend Hypothesis testing P value Significance level ** (H 0 )=0 joinpoint versus (H a )=3 joinpoints 0.0002 0.0167 (H 0 )=1 joinpoint versus (H a )=3 joinpoints 0.0002 0.025 (H 0 )=2 joinpoint versus (H a )=3 joinpoints 0.0068 0.05 *Selected model ; ** t ests were performed with the Bonferroni correction at an overall significance level of 0.05 Table 4 18. Specification for stimu lant maintenance daily dose trend Segment MPC (95% CI) Joinpoint (95% CI) I (01/2002 12/2004) 0.1 ( 0.1,0.2) 12/2004 (07/2004, 06/2005) II (12/2004 04/2006) 2.1 ( 2.6, 1.5) 04/2006 (09/2005, 06/2006) III (04/2006 07/2006) 6.6 ( 13, 31) 07/2006 (05/2006, 02/2008) IV (07/2006 06/2008) 0.1 ( 0.5,0.3) Abbreviations: MPC: monthly percent change, CI: confidence interval.

PAGE 82

82 Table 419. Comparability test results for stratified stimulant maintenance daily dose trend Test pairs Monthly sample size (range) P values (Parallel test, coincident test) < 5 years vs 5 13 years 14 (13 15) vs 73 (32 115) (0.13, 0.0002) 14 20 years vs 5 13 years 3 (2 4) vs 73 (32 115) (0.75, 0.0002) Female vs male 24 (19 30) vs 67 (32 102) (0.13, 0.14) Blacks vs Whites 24 (13 36) vs 40 (24 57) (0.22, 0.0002) Hispanics vs Whites 14 (8 21) vs 40 (24 57) (0.20, 0.0002) Supplemental security income vs Temporary assistance for needy family 14 (6 23) vs 42 (24 61) (0.4, 0.31) Foster care vs Temporary assistance for needy family 9 (5 14) vs 42 (24 61) (0.46, 0.0002) Presence of mental comorbidities vs ADHD only 29 (24 35) vs 62 (27 97) (0.68, 0.86) S ignificance level= 0.05

PAGE 83

83 Table 4 2 0. Patient and provider characteristics in pretreatment electrocardiograph (ECG) use trend analysis Pre warning Post warning Post warning Post warning Post warning Total 01/2001 01/2005 02/2005 12/2005 01/2006 12/2006 01/2007 01/2007 01/2008 11/2008 01/2001 11/2008 ADHD patients (n) 18964 3760 3374 2377 1664 30139 Follow up time (pt mo ) 11159 2099 1819 1107 658 16842 Avg. patient per month (n) 387 342 281 198 151 317 ECG use (n) 382 94 137 109 130 852 ECG use rate (1000 pt mo ) 34 45 75 98 197 51 Demographic Mean age (years old) (95% CI) 7.8 (7.7 7.8) 7.7 (7.5 7.8) 7.5 (7.4 7.7) 7.5 (7.4 7.7) 7.6 (7.4 7.8) 7.7 (7.7 7.8) < 5 yrs (%) 16 16 17 15 14 16 5 9 yrs (%) 59 62 63 67 67 61 10 14 yrs (%) 21 16 15 13 14 19 15 20 yrs (%) 3 4 3 4 3 3 >20 yrs (%) 1 2 2 2 2 1 Male (%) 71 70 70 67 70 71 White (%) 46 45 46 46 47 46 Black (%) 22 24 23 22 22 22 Hispanic (%) 16 16 17 16 15 16 Other race (%) 16 14 14 16 17 16

PAGE 84

84 Table 420. Continued Pre warning Post warning Post warning Post warning Post warning Total 01/2001 01/2005 02/2005 12/2005 01/2006 12/2006 01/2007 01/2007 01/2008 11/2008 01/2001 11/2008 Medicaid eligibility status Temporary assistance for needy family (%) 41 43 41 42 41 41 Supplemental security income (%) 18 15 15 14 16 17 Foster care (%) 10 10 12 13 10 10 Other category (%) 31 32 33 31 33 31 Clinical conditions M ental comorbidity (%) 29 29 31 25 27 29 Provider specialty Primary care physicians (%) 48 52 54 66 59 51 Psychiatrists (%) 27 17 14 8 5 22 Other physicians (%) 9 13 14 14 26 11 Missing (%) 17 18 19 12 10 16 Abbreviations: pt mo, patient months, Avg.: average, CI: confidence interval.

PAGE 85

85 Table 4 21. Testing results for pretreatment electrocardiograph use trend Hypothesis testing P value Significance level ** (H 0 )=0 joinpoint versus (H a )=3 joinpoints 0.0002 0.0167 (H 0 )=1 joinpoint versus (H a )=3 joinpoints 0.003 0.025 (H 0 )=2 joinpoint versus (H a )=3 joinpoints 0.051 0.05 *Selected model ; ** t ests were performed with the Bonferroni correction at an overall significance level of 0.05. Table 4 2 2 Specification for pretreatment electrocardiograph use trend Segment MPC (95% CI) Joinpoint (95% CI) I (01/2001 06/2004) 0.6 ( 1.5,0.4) 06/2004 (07/2003, 06/2005) II (06/2004 02/2008) 3.2 (2.3, 4.2) 02/2008 (07/2007, 11/2008) III (02/2008 11/2008) 13.2 (4.2 23) Abbreviations: MPC: monthly percent change, CI: confidence interval.

PAGE 86

86 Table 423. Comparability test results for stratified pretreatment electrocardiograph use trend Test pair s Monthly sample size (range) P values (parallel test, c oincident test) < 5 years vs 5 13 years 50 (8 61) vs 246 (83 341) (0.09, 0.07) 14 20 years vs 5 13 years 17 (3 26) vs 246 (83 341) Female vs male 90 (18 99) vs 222 (76 329) (0.79, 0.83) Blacks vs Whites 70 (17 96) vs 142 (47 187) (0.82, 0.92) Hispanics vs Whites 50 (9 57) vs 142 (47 187) (0.40, 0.006) Supplemental security income vs Temporary assistance for needy family 52 (12 122) vs 129 (49 144) (0.30, 0.37) Foster care vs Temporary assistance for needy family 33 (8 59) vs 129 (49 144) (0.26, 0.03) Primary care physician vs psychiatrist 160 (51 173) vs 67 (2 163) (0.01, ) ADHD+mental comorbidities vs ADHD only 89 (23 134) vs 224 (71 294) (0.05, ) S ignificance level= 0.05

PAGE 87

87 Figure 41 Stimulant initiation trend 0 100 200 300 400 500 600 Jan 01 Jan02 Jan 03 Jan 04 Jan05 Jan 06 Jan 07 Jan08 per 1,000 patient months Adderall Boxed warning MedGuide withdrawal debate distributed

PAGE 88

88 Figure 42 Stimulant and atomoxetine initiation trend 0 100 200 300 400 500 600 Jan 01 Jan 02 Jan 03 Jan 04 Jan 05 Jan 06 Jan 07 Jan 08 Per 1,000 patient months 1 2 1 I 2 II 3 Stimulant Stimulant fitted model Atomoxetine Atomoxetine fitted model 1 2 3 Adderall withdrawal Boxed warning debate MedGuide distributed I II Atomoxetine boxed warning Daytrana entered the market

PAGE 89

89 Figure 43. Stimulant initiation trend (oral products only) 0 100 200 300 400 500 600 Jan 01 Jan 02 Jan 03 Jan 04 Jan05 Jan 06 Jan 07 Jan08 per 1,000 patient months Adderall Boxed warning MedGuide withdrawal debate distributed

PAGE 90

90 Figure 4 4 Stimulant initiation trend by age (< 5 years versus 513 years) Figure 45 Stimulant initiation trend by age (1420 years old versus 513 years old) 0 100 200 300 400 500 600 700 Jan 04 Jan 05 Jan 06 Jan 07 Jan 08 Per 1,000 patient months 0 100 200 300 400 500 600 700Jan04 Jan05 Jan 06 Jan 07 Jan 08 Per 1,000 patient months 1 2 3 5 13 years 5 13 years fitted model <5 years <5 years fitted model Adderall withdrawal Boxed warning debate MedGuide distributed 1 2 3 1 2 3 1 2 3 5 13 years 5 13 years fitted model 1420 years 1420 years fitted model Adderall withdrawal Boxed warning debate MedGuide distributed

PAGE 91

91 Figure 46 Stimulant initiation trend by gender 0 100 200 300 400 500 600 700 800 900 Jan 04 Jan 05 Jan 06 Jan 07 Jan 08 Per 1,000 patinet months 1 2 3 1 2 3 Male Male fitted model Female Female fitted model Adderall withdrawal Boxed warning debate MedGuide distributed

PAGE 92

92 Figure 47 Stimulant initiation trend by race (Blacks versus Whites) Figure 48 Stimulant initiation trend by race (Hispanics versus Whites) 0 100 200 300 400 500 600 700 800 900 1000 Jan 04 Jan 05 Jan 06 Jan 07 Jan 08 Per 1,000 patient months 1 2 3 Whites Whites fitted model Blacks Blacks fitted model Adderall withdrawal Boxed warning debate MedGuide distributed 1 2 3 0 100 200 300 400 500 600 700 800 900 1000 Jan 04 Jan 05 Jan06 Jan 07 Jan 08 Per 1,000 patient months 1 2 3 1 2 3 Whites Whites fitted model Hispanics Hispanics fitted model Adderall withdrawal Boxed warning debate MedGuide distributed

PAGE 93

93 Figure 49 Stimulant initiation trend by eligibility ( Supplemental security income [SSI] v er sus Temporary assistance for needy family [TANF] ) Figure 410 Stimulant initiation trend by eligibility (Foster care v s Temporary assistance for needy family [TANF] ) 0 100 200 300 400 500 600 700 800 900 1000Jan04 Jan 05 Jan 06 Jan 07 Jan 08 Per 1,000 patient months 1 2 3 1 2 3 TANF TANF fitted model SSI SSI fitted model Adderall withdrawal Boxed warning debate MedGuide distributed 0 100 200 300 400 500 600 700 800 900 1000 Jan 04 Jan 05 Jan06 Jan 07 Jan 08 Per 1,000 patient months 1 2 3 1 2 3 TANF TANF fitted model Foster care Foster care fitted model Adderall withdrawal Boxed warning debate MedGuide distributed

PAGE 94

94 Figure 411 Stimulant initiation trend by comorbidity status 0 100 200 300 400 500 600 700 800 900 1000 Jan 04 Jan 05 Jan 06 Jan 07 Jan 08 Per 1,000 patient months 1 Adderall withdrawal 2 Boxed warning debate 3 MedGuide distributed 1 2 3 1 2 3 ADHD only ADHD only fitted model ADHD+comorbidity ADHD+comorbidity fitted model Adderall withdrawal Boxed warning debate MedGuide distributed

PAGE 95

95 Figure 412 Stimulant initiation trend by provider type 0 200 400 600 800 1000 1200 1400 Jan 04 Jan 05 Jan 06 Jan07 Jan08 Per 1,000 patient months 1 2 3 1 2 3 Primary care physicians Primary care physicians fitted model Psychiatrists Psychiatrists fitted model Adderall withdrawal Boxed warning debate MedGuide distributed

PAGE 96

96 Figure 413 Stimulant discontinuation trend 0 10 20 30 40 50 60 70 80 90 100 Jan 02 Jan 03 Jan 04 Jan 05 Jan 06 Jan 07 Jan 08 Per 1,000 patient months Adderall Boxed warning MedGuide withdrawal debate distributed

PAGE 97

97 Figure 414 Stimulant discontinuation trend by age group (< 5 years versus 513 years) Figure 415 Stimulant discontinuation trend by age group (513 years versus 1420 years) 0 20 40 60 80 100 120 Jan 02 Jan 03 Jan 04 Jan 05 Jan 06 Jan07 Jan 08 Per 1,000 patient months 1 2 3 5 13 years 5 13 years fitted model <5 years <5 years fitted model Adderall withdrawal Boxed warning debate MedGuide distributed 0 20 40 60 80 100 120Jan02 Jan 03 Jan04 Jan 05 Jan 06 Jan 07 Jan 08 Per 1,000 patient month 1 2 3 5 13 years 5 13 years fitted model 1420 years 1420 years fitted model Adderall withdrawal Boxed warning debate MedGuide distributed 1 2 3 1 2 3

PAGE 98

98 Figure 416 Stimulant discontinuation trend by gender 1 2 3 0 10 20 30 40 50 60 70 Jan 02 Jan 03 Jan 04 Jan 05 Jan06 Jan 07 Jan 08 Per 1,000 patient months 1 2 3 Male Male fitted model Female Female fitted model Adderall withdrawal Boxed warning debate MedGuide distributed

PAGE 99

99 Figure 417 Stimulant discontinuation trend by race (Blacks versus Whites) Figure 418 Stimulant discontinuation trend by race (Hispanics versus Whites) 0 20 40 60 80 100 120 Jan 02 Jan 03 Jan 04 Jan 05 Jan 06 Jan 07 Jan 08 Per 1,000 patient months 0 10 20 30 40 50 60 70 Jan 02 Jan 03 Jan 04 Jan 05 Jan 06 Jan 07 Jan 08 Per 1,000 patient months 1 2 3 1 2 3 Whites Whites fitted model Blacks Blacks fitted model Adderall withdrawal Boxed warning debate MedGuide distributed 1 2 3 1 2 3 Whites Whites fitted model Hispanics Hispanics fitted model Adderall withdrawal Boxed warning debate MedGuide distributed

PAGE 100

100 Figure 419 Stimulant discontinuation trend by eligibility status ( Supplemental security income [SSI] v er sus Temporary assistance for needy family [TANF] ) Figure 420 Stimulant discontinuation trend by eligibility status (Foster care versus Temporary assistance for needy family [TANF] ) 0 10 20 30 40 50 60 70 80 90 Jan 02 Jan 03 Jan 04 Jan 05 Jan 06 Jan 07 Jan08 Per 1,000 patient months 0 10 20 30 40 50 60 70Jan02 Jan 03 Jan 04 Jan 05 Jan 06 Jan 07 Jan 08 Per 1,000 patient months 1 2 3 1 2 3 1 2 3 TANF TANF fitted model SSI SSI fitted model Adderall withdrawal Boxed warning debate MedGuide distributed 1 2 3 TANF TANF fitted model Foster care Foster care fitted model Adderall withdrawal Boxed warning debate MedGuide distributed

PAGE 101

101 Figure 421 Stimulant discontinuation trend by comorbidity status 0 10 20 30 40 50 60 70 Jan 02 Jan03 Jan 04 Jan05 Jan06 Jan 07 Jan08 Per 1,000 patient months 1 2 3 1 2 3 ADHD only ADHD only fitted model ADHD+comorbidity ADHD+comorbidity fitted model Adderall withdrawal Boxed warning debate MedGuide distributed

PAGE 102

102 Figure 422 Stimulant initial daily dose trend 10 15 20 25 30 35 40 45 50 55 60Jan02 Jan 03 Jan 04 Jan 05 Jan 06 Jan07 Jan 08 Milligram/day Adderall Boxed warning MedGuide withdrawal debate distributed

PAGE 103

103 Figure 423 Stimulant initial daily dose trend by age group (< 5 years versus 5 13 years) Figure 424 Stimulant initial daily dose trend by age group (1420 years versus 513 years) 0 5 10 15 20 25 30 35 Jan 02 Jan03 Jan 04 Jan 05 Jan 06 Jan07 Jan 08 Milligram/ day 1 2 3 0 10 20 30 40 50 60 70 Jan 02 Jan03 Jan 04 Jan 05 Jan 06 Jan 07 Jan 08 Milligram/day 1 2 3 1 2 3 5 13 years 5 13 years fitted model <5 years <5 years fitted model Adderall withdrawal Boxed warning debate MedGuide distributed 1 2 3 5 13 years 5 13 years fitted model 1420 years 1420 years fitted model Adderall withdrawal Boxed warning debate MedGuide distributed

PAGE 104

104 Figure 425 Stimulant initial daily dose trend by gender 0 5 10 15 20 25 30 35 Jan 02 Jan 03 Jan 04 Jan 05 Jan 06 Jan 07 Jan 08 Milliramg/ day 1 2 3 1 2 3 Male Male fitted model Female Female fitted model Adderall withdrawal Boxed warning debate MedGuide distributed

PAGE 105

105 Figure 426 Stimulant initial daily dose trend by race (Blacks versus Whites) Figure 427 Stimulant initial daily dose trend by race (Hispanics versus Whites) 0 5 10 15 20 25 30 35 40 Jan 02 Jan 03 Jan 04 Jan 05 Jan 06 Jan 07 Jan 08 Milligram/day 0 5 10 15 20 25 30 35 40Jan02 Jan 03 Jan 04 Jan 05 Jan 06 Jan 07 Jan08 Milligram/ day 1 2 3 Whites Whites fitted model Blacks Blacks fitted model Adderall withdrawal Boxed warning debate MedGuide distributed 1 2 3 Whites Whites fitted model Hispanics Hispanics fitted model Adderall withdrawal Boxed warning debate MedGuide distributed 1 2 3 1 2 3

PAGE 106

106 Figure 428 Stimulant initial daily dose trend by eligibility status ( Supplemental security income [SSI] ver sus Temporary assistance for needy family [TANF] ) Figure 429 Stimulant initial daily dose trend by eligibility status (Foster care versus Temporary assistance for needy family [TANF] ) 0 5 10 15 20 25 30 35 40 Jan 02 Jan 03 Jan 04 Jan 05 Jan06 Jan07 Jan08 Milligram/ day 0 5 10 15 20 25 30 35 40Jan02 Jan03 Jan 04 Jan 05 Jan 06 Jan 07 Jan 08 Milliramg/ day 1 2 3 TANF TANF fitted model SSI SSI fitted model Adderall withdrawal Boxed warning debate MedGuide distributed 1 2 3 TANF TANF fitted model Foster care Foster care fitted model Adderall withdrawal Boxed warning debate MedGuide distributed 1 2 3 1 2 3

PAGE 107

107 Figure 430 Stimulant initial daily dose trend by comorbidity status 0 5 10 15 20 25 30 35 Jan 02 Jan 03 Jan 04 Jan 05 Jan06 Jan07 Jan 08 Milligram/ day 1 2 3 ADHD only ADHD only fitted model ADHD+comorbidity ADHD+comorbidity fitted model Adderall withdrawal Boxed warning debate MedGuide distributed 1 2 3

PAGE 108

108 Figure 431 Stimulant initial daily dose trend by provider type 0 5 10 15 20 25 30 35 40 Jan 02 Jan 03 Jan 04 Jan 05 Jan 06 Jan 07 Jan 08 Milligram/day 1 2 3 Primary care physicians Primary care physicians fitted model Psychiatrists Psychiatrists fitted model Adderall withdrawal Boxed warning debate MedGuide distributed 1 2 3

PAGE 109

109 Figure 432 Stimulant maintenance daily dose trend 10 15 20 25 30 35 40 45 50 55 60Jan02 Jan 03 Jan 04 Jan 05 Jan 06 Jan 07 Jan 08 Milligram/day Adderall Boxed warning MedGuide withdrawal debate distributed

PAGE 110

110 Figure 433 Stimulant maintenance daily dose trend by age group (< 5 years versus 513 years) Figure 434 Stimulant maintenance daily dose trend by age group (1420 years versus 513 years) 0 5 10 15 20 25 30 35Jan02 Jan03 Jan04 Jan05 Jan06 Jan07 Jan08 Milligram/day 0 10 20 30 40 50 60Jan02 Jan 03 Jan 04 Jan05 Jan 06 Jan 07 Jan 08 Milligram/day 1 2 3 5 13 years 5 13 years fitted model <5 years <5 years fitted model Adderall withdrawal Boxed warning debate MedGuide distributed 1 2 3 5 13 years 5 13 years fitted model 1420 years 1420 years fitted model Adderall withdrawal Boxed warning debate MedGuide distributed 1 2 3 1 2 3

PAGE 111

111 Figure 435 Stimulant maintenance daily dose trend by gender 0 5 10 15 20 25 30 35Jan02 Jan 03 Jan 04 Jan05 Jan 06 Jan 07 Jan 08 Milligram/day 1 2 3 Male Male fitted model Female Female fitted model Adderall withdrawal Boxed warning debate MedGuide distributed 1 2 3

PAGE 112

112 Figure 436 Stimulant maintenance daily dose trend by race (Blacks versus Whites) Figure 437 Stimulant maintenance daily dose trend by race (Hispanics versus Whites) 0 5 10 15 20 25 30 35 Jan 02 Jan 03 Jan 04 Jan 05 Jan06 Jan 07 Jan 08 Milligram/day 1 2 3 0 5 10 15 20 25 30 35 Jan 02 Jan03 Jan04 Jan05 Jan 06 Jan 07 Jan 08 Milligram/day 1 2 3 Whites Whites fitted model Blacks Blacks fitted model Adderall withdrawal Boxed warning debate MedGuide distributed 1 2 3 Whites Whites fitted model Hispanics Hispanics fitted model Adderall withdrawal Boxed warning debate MedGuide distributed 1 2 3

PAGE 113

113 Figure 438 Stimulant maintenance daily dose trend by eligibility status ( Supplemental security income [ SSI ] versus Temporary assistance for needy family [ TANF ] ) Figure 439 Stimulant maintenance daily dose trend by eligibility status (Foster care versus Temporary assistance for needy family [ TANF ] ) 0 5 10 15 20 25 30 35Jan02 Jan 03 Jan 04 Jan05 Jan 06 Jan 07 Jan 08 Milligram/day 0 5 10 15 20 25 30 35 40 Jan 02 Jan 03 Jan 04 Jan 05 Jan 06 Jan 07 Jan 08 Milligram/ day 1 2 3 TANF TANF fitted model SSI SSI fitted model Adderall withdrawal Boxed warning debate MedGuide distributed 1 2 3 TANF TANF fitted model Foster care Foster care fitted model Adderall withdrawal Boxed warning debate MedGuide distributed 1 2 3 1 2 3

PAGE 114

114 Figure 440 Stimulant maintenance daily dose trend by comorbidity status 0 5 10 15 20 25 30 35 Jan 02 Jan 03 Jan 04 Jan 05 Jan 06 Jan 07 Jan 08 Milligram/day 1 2 3 ADHD only ADHD only fitted model ADHD+comorbidity ADHD+comorbidity fitted model Adderall withdrawal Boxed warning debate MedGuide distributed 1 2 3

PAGE 115

115 Figure 441 Pre treatment ECG use trend 0 50 100 150 200 250 300 350 400 450 Jan 01 Jan02 Jan 03 Jan 04 Jan 05 Jan06 Jan 07 Jan 08 Per 1,000 patient months Adderall Boxed warning MedGuide withdrawal debate distributed

PAGE 116

116 Figure 442 Pre treatment electrocardiography use trend by age group (< 5 years versus 513 years) 0 100 200 300 400 500Jan01 Jan 02 Jan03 Jan 04 Jan 05 Jan 06 Jan 07 Jan 08 per 1,000 patient months 1 2 3 1 2 3 5 13 years 5 13 years fitted model <5 years <5 years fitted model Adderall withdrawal Boxed warning debate MedGuide distributed

PAGE 117

117 Figure 443 P re treatment electrocardiography use trend by gender 0 100 200 300 400 500 Jan 01 Jan02 Jan 03 Jan 04 Jan 05 Jan 06 Jan 07 Jan 08 Per 1,000 patient months 1 2 3 1 2 3 Male Male fitted model Female Female fitted model Adderall withdrawal Boxed warning debate MedGuide distributed

PAGE 118

118 Figure 444 Pre treatment electrocardiography use trend by race (Black versus Whites) Figure 445 Pre treatment electrocardiography use trend by race (Hispanics versus Whites) 0 100 200 300 400 500 600 Jan 01 Jan02 Jan 03 Jan 04 Jan 05 Jan06 Jan 07 Jan 08 Per 1,000 patient months 0 100 200 300 400 500 600 700 800 Jan 01 Jan02 Jan 03 Jan 04 Jan 05 Jan06 Jan 07 Jan 08 Per 1,000 patient months 1 2 3 Whites Whites fitted model Blacks Blacks fitted model Adderall withdrawal Boxed warning debate MedGuide distributed 1 2 3 Whites Whites fitted model Hispanics Hispanics fitted model Adderall withdrawal Boxed warning debate MedGuide distributed 1 2 3 1 2 3

PAGE 119

119 Figure 446 Pre treatment electrocardiography use trend by eligibility status ( Supplemental security income [ SSI ] versus Temporary assistance for needy family [ TANF ] ) Figure 447 Pre treatment electrocardiography use trend by eligibility status (Foster care versus Temporary assistance for needy family [ TANF ] ) 0 100 200 300 400 500 600 700 800Jan01 Jan 02 Jan 03 Jan04 Jan 05 Jan 06 Jan 07 Jan 08 Per 1,000 patient months 0 100 200 300 400 500Jan-01 Jan 02 Jan 03 Jan 04 Jan-05 Jan-06 Jan 07 Jan 08 Per 1,000 patinet months 1 2 3 TANF TANF fitted model SSI SSI fitted model Adderall withdrawal Boxed warning debate MedGuide distributed 1 2 3 TANF TANF fitted model Foster care Foster care fitted model Adderall withdrawal Boxed warning debate MedGuide distributed 1 2 3 1 2 3

PAGE 120

120 Figure 448 Pre treatment electrocardiography use trend by comorbidity status 0 100 200 300 400 500 Jan 01 Jan 02 Jan 03 Jan 04 Jan 05 Jan 06 Jan07 Jan 08 Per 1,000 patient months 1 2 3 ADHD only ADHD only fitted model ADHD+comorbidity ADHD+comorbidity fitted model Adderall withdrawal Boxed warning debate MedGuide distributed 1 2 3

PAGE 121

121 Figure 449 Pre treatment electrocardiography use trend by provider type 0 100 200 300 400 500 600Jan01 Jan02 Jan03 Jan04 Jan05 Jan06 Jan07 Jan 08 Per 1,000 Patient months 1 2 3 Primary care physicians Primary care physicians fitted model Psychiatrists Psychiatrists fitted model Adderall withdrawal Boxed warning debate MedGuide distributed 1 2 3

PAGE 122

122 CHAPTER 5 DISCUSSIONS Stimulant U tilization after Hea lth Canadas W ithdrawal of Adderall While prior research on the impact of drug safety warnings found treatment decisionmaking altered soon after the first signal58, 61, 84, we did not observe an immediate decrease i n stimulant initiation or discontinuation after Canadas action; we, however, found the prescribed dose significantly reduced. The initial daily dose decreased 6% per month from 26 milligram (mg) (MPH equivalents), and the reduction coincidentally stopped after the time Adderall reentered the Canadian market at an average of about 20 mg. The maintenance daily dose declined consistently through this time period with a rate of 2% per month from 26 mg per day. The absolute reduction in stimulant treatment intensity was approximately 6 mg, which equates to a 3060% decrease of the recommended initial dose (5 mg twice daily for immediaterelease formulations; 18 mg once daily for extendedrelease tablets; 20 mg once daily for extendedrelease capsules) or 20 % of the maintenance dose (30.5 mg) in the landmark ADHD treatment trial, the Multimodal Treatment Study of Children with ADHD (MTA). It raises the question whether treatment effectiveness was compromised in resp onse to Canadas action, given that the maintenance dose before Canadas action ( 6 mg ) was already lower than that in the MTA study (30.5mg). However, it should be noted that the clinical trial population might represent patients with more pronounced sym ptoms, and that lower doses might suffice for real world populations of children with mixed symptom severity.

PAGE 123

123 Stimulant U tilization after the Debate on Boxed W arnings The stimulant CV safety issue was addressed extensively in the U.S. during t he debate ov er the boxed warning proposal. The FDAs final evaluation of this safety signal was not clear, largely because the available evidence is inconclusive. On one hand, rejection of the boxed warning proposal by the Drug Safety and Risk Management Advisory Boar d implies the notion that stimulant treatment is generally safe. On the other hand, the requirement of a MedGuide suggests that the safety risk was not was not considered minor, following regulatory guidance that MedGuide are issued for drugs when their s ide effect possess serious and significant public health concerns 85, 86The controversial opinions on stimulant CV safety are reflected in the utilization patterns. The observed changes during this period wer e different from the impact usually attributed to a safety signal. We found stable treatment discontinuation ; no further change in initial treatment strength; a 50% (370540 per 1000 patient month) increase in stimulant initiation rate, and, a 4mg MPH equi valents rebound in the maintenance daily dose from the initial decrease after Canadas action. The increase in treatment initiation and intensity could be attributed to the FDAs rejection of the boxed warning proposal or the latent impact from the remark eting of Adderall. Moreover, a new stimulant product (i.e. Daytrana; approval date= 04/2006) was released in this period. The introduction of new products have shown to positively alter overall use and patients and providers perception about stimulant safety might have been mitigated with new FDA approvals and new marketing efforts. Stimulant U tilizatio n after the D istribution of MedGuides The distribution of MedGuides marked a big step in communicating stimulant CV safety concern in a systematic fashion. The MedGuide is a consumer directed drug

PAGE 124

124 information pamphlet distributed by pharmacies when consumers fill or refill a prescription for the target drug. It is supposed to assure that patients receive FDA approved information regarding risk of severe C V side effects prior to taking stimulants. Nevertheless, the effectiveness of MedGuides as a risk communication tool has been questioned, although it has not been formally assessed. A study that evaluated reading difficulty, content, and format of 40 MedGuides reported the Guides were generally written at a higher reading level than the federal recommendations, including extensive quantities of information (the average length of Medication Guides was 2208 words, or 6 10 pages) and lack of summaries highlighting the most important information for patients.87 Moreover, interviews of 251 primary care patients at a public hospital clinic suggested that only few (less than a quarter of) patients had looked at MedGuides.87Given concerns about the effectiveness of MedGuides, the decrease in stimulant initiation rate after the release of MedGuides was unexpected. A causal relationship between the release of MedGuides and the reduction in stimulant initiation, however, seems less plausible if we factor in the way the guides were distributed. Currently, MedGuides are delivered at the time of dispensing; they are enclosed with the dispensed medication. It is unlikely that patients are able to make a treatment decision based on the guide for the first prescription since they woul d not have the chance to read it before they receive the dispensed drug. Only a comparison of prescribing versus dispensing information would be able to discern whether MedGuides influenced prescribers or patients in deciding to initiate therapy. To provide perspective, we compare our results to those of a study evaluating the impact of regulatory actions related to s electi ve serotonin reuptake inhibitor (SSRI)

PAGE 125

125 antidepressants suicidal ideation.84We found no change in the discontinuation rate after the distribution of MedGuides, similar to what we observed in earlier time periods, which suggests that the primary effect of drug safety warnings occurred at the point of treatment initiation. The two warnings have similar target populations (children and adolescent with mental disorder) and similar severity of the proposed adverse effect, yet differences in the strength of evidence supporting a causal association. This study, which was also conducted in a Medicaid population, found a decrease in treatment initiation of similar magnitude as the trend in our study (1.6% per month versus 2%). Since the evidence is more concrete, regulatory agencies issued warnings in a stronger tone. For example, the drug regulatory agency in the UK declared the risk benefits profile of most SSRI antidepressants were unfavorable for the treatment of major depressive disorder in children and adolescents T he US FDA issued a boxed warning. It is therefore surprising that a minor warning might have a comparable effect. However, because evidence on the effectiveness of stimulant treatment is fragmented and because many reports have questioned whether stimulants are over used, providers and patients might have readjusted their understanding of the risk benefit profile entirely. Potentially, the observed decrease in stimulant utilization might reflect those patients who received only marginal benefits from treatment. Overall, the reduction in stimulant initiation rate (34%) that could be associated with CV safety warnings was clinically significant. 84 It also i mplies that provision of a MedGuide at the time of refilling a prescription might not have a significant impact on treatment decisionmaking. Thus, emerging safety issues may not be effectively communicated via MedGuides to prevalent drug users. It is also

PAGE 126

126 noteworthy that 30% or one in three stimulant users stopped treatment within 6 months of initiation, throughout the entire study period, raising questions about treatment effectiveness, suboptimal ADHD care or excessive stimulant use. We found no change in treatment strength after the distribution of MedGuide, which is expected, since dosing decisions are typically made by prescribers, yet the target audiences of the Guide are consumers. ECG Utilization after Stimulant CV Safety W arnings It is arguable if the first change point observed in the pretreatment ECG trend is attributable to the CV safety warning since the point estimate is more than half a year earlier than Canadas action and the confidence interval is wide. In contrast, the increase in the ECG utilization in early 2008 seems more closely tied to the AHAs re commendation on pretreatment ECG screening. This finding is encouraging, in particular because prior research has not found that a drug safety warning increases a preventive action.57, 88 Nevertheless, while ECG utilization significantly rose from 34 per 1,000 patient months to 400 per 1,000 patient months, the clinical benefit from such marginal improvement is probably negligible given low ECG test sensitivity.8991Change in Trends among S ubgroups Overall, the change in stimulant utilization trends after CV safety warnings was not different among most subgroups. It should be noted that the sample size of the subgroups were small, resulting is limited power to detect differences. We did not observe a decrease in stimulant initiation trend after safety warnings among younger (< 5 years old) patients, patients with other mental comorbidities, or patients receiving Medicaid benefits due to SSI status (disability). Previous research offer s support for an association between age and ADHD severity; patients diagnosed

PAGE 127

127 with ADHD at a younger age were found to have poorer functioning and more aggression than their older counterparts by teacher report and clinic ian rating.92 In addition, an early age of onset of ADHD was correlated with a greater rate of parent reported child externalizing comorbid symptoms93, including several aggressive behaviors. Given the more severe ADHD symptoms, it is likely more difficult to abandon medication treatment among younger patients, and thus, their stimulant initiation rate did not change in comparison to older patients. Likewise, patients with ment al comorbidities or disability generally have more complex psychiatric conditions39Change in pretreatment ECG utilization trend after CV safety warnings differed by patients clinical condition and provider type. The higher E CG screening rate prior to warnings among patients with other mental conditions might be due to concerns about their more fragile health status. The increase in ECG use in turn, was sharper among patients with only ADHD. The more pronounced increase in EC G screening among psychiatrists might be due to the fact that specialists adopt new medical information more rapidly than generalists do. and control of ADHD symptoms is probably given higher priority than in otherwise healthy patients. Nevertheless, among those patients where ADHD treatment benefit might out weigh risk, stimulants were prescribed at a lower strength after concerns on CV risk were voiced, consistently with their comparison groups. 9497 Psychiatrists might also feel a greater need for general health assessments since their medical involvement is typically not as holistic as care provided by a primary care provider who might feel in a better position to evaluate presence of cardiac risk factors.

PAGE 128

128 It is noteworthy t hat stimulant maintenance dose was consistently lower among minorities and higher in foster care children; especially, Hispanic patients received treatment at a 4mg lower strength than Whites. It is not clear if race were a modifier of ADHD severity or stimulant efficacy, however, given that the initial dose is similar across ethnic groups, it is likely that minorities either adjust doses downward or adhere less to a regular dosing schedule. This observation is consistent with different attitudes toward psy chotropic use across cultures, and raises the concern that treatment outcomes may be less optimal among minorities. Foster care status has been found previously to be a positive predictor for psychotropic utilization27Other Important Findings and similarly, could be a predictor for more aggressive dosing. It is possible that the hurdle of getting foster care children into th e health care system and of assuring regular follow up lead providers to more aggressive treatment regimens to assure positive treatment outcomes. We noticed the number of newly diagnosed ADHD patients was shrinking during the post CV warning period (from 2005 to 2008). Enrollment records from Florida Medicaid showed the cumulative size of feefor service beneficiaries per calendar year varied only slightly from 2001 to 2008, as did the age composition of beneficiaries (Table 51 and 52). Thus, the decrease in the size of new episode patients was probably driven by a reduction of ADHD diagnoses, predominately among the pediatric population (from 0.57% at the end of 2004 to 0.25% at the end of 2008). The association between ADHD diagnosis trend and stimulant CV warnings is not the scope of this study; however, it is possible that drug safety concerns may have affected the disease diagnosis rate as well. The pediatric depression diagnosis rate was lowered significantly after the FDA advisory issued a warning on SSRI suicidality risk60. It is

PAGE 129

129 reasonable to suspect the same applied to ADHD diagnosis pattern and raises questions whether the disorder had been over diagnosed before or is under recognized now. The Application of Joinpoint A nalysis Our study is one of few that applied JP analysis to assess the impact of new evidence or regulatory changes on health service utilization. It is unique in that the research questions could not have been answered without JP analysis Studies using other approaches in quasiexperimental design, including the strongest conventional method interrupted timeseries analysis require a definite hypothesis on the timing and latency of event(s) that might influence the trend. In our case, t he multitude of events and uncertainty of their relative magni tude and latency, as well as the weak evidence and controversial action on stimulants CV risk prohibited any explicit a priori hypotheses about change points. The findings have to be interpret ed with caution. The estimation method currentl y used in the analytical software (i.e. the grid research method) assumes that the joinpoints occur at the observed data points. Therefore, exact timing (i.e. the location) of change in trend can be inaccurate. Also, latent effects are not easily captured by the JP model. As we test for trend change after multiple events, we are not able to attribute a change to a single occasion. However, t he immediacy of change and how closer it is related to the event could point to a causal relationship Study Limitations In addition to the limitation due to the statistical technique, our study is also restricted by the data source and nature of administrative data. First, this study was based on the Florida Medicaid population, and thus, limited to a single state and biased

PAGE 130

130 towards low income groups and minorities. The practice of Medicaid providers may not be representative of providers under contract with other types of health insurance programs. Second, claim data reflect only dispensing information and cannot descri be the actual medication administr ation; drug utilization and persistence may be lower than what were found in this study. Nevertheless the fact that prescriptions were filled suggests some intent to administer medication. Third, patients diagnostic info rmation was not confirmed through chart review or other clinical assessments The requirement of a second di agnostic code increases specificity but under or mis coding of health care encounter information may result in the inclusion of patients who did not have ADHD or in miss classification of patients comorbidity status Miss classification decreases our ability to detect a change in utilization pattern; however, it would need to change over time to bias the trend analyses. Forth, w e were not able to address disease severity, which is likely to influence treatment decisions The data source also lacks important contextual factors on patient, family, provider, and treatment systems which may help to explain sociocultural contributions to the observed t rends. Lastly, some measures were based on small numbers per time unit; this makes the identification of change points arbitrary and highly sensitive to the chosen pvalue cut off, especially in the subgroup analyses. Also, we are not able to run multivari ate analysis in JP software, and the ability to stratify analyses was again limited by sample size. Summary and Future Research Canadas withdrawal of Adderall had no effect on stimulant initiation but did permanently decrease treatment intensity Furthermore, a slightly smaller porti on of ADHD patients initiated drug treatment in 2008 after release of MedGuides. It is

PAGE 131

131 important to assess how the decrease in stimulant use or intensity translated into clinical outcomes. Subpopulations with unusual trends, such as, young ageof diagnosis and presence of mental comorbidities or physical illness, should be especially studied. The findings did not support that the publicity or regulatory actions on stimulant CV safety concerns, including the distribution of MedGuides, altered prevalent drug users treatment decisions, as stimulant discontinuation rates remained stable dur ing the study period. The effectiveness of MedGuides as a risk communication tool needs further examination. Lastly, the declining new ADHD visits and treatment intensity, the differential stimulant maintenance strength among subpopulations and the subopt imal treatment persistence underscore the importance of investigations on the appropriateness of ADHD diagnoses and stimulant treatment. Conclusions Practitioners reacted to stimulant CV safety concerns with immediate 6mg methylphenidate equivalent reduction in dosing and an increase in ECG screening, affecting however only a marginal proportion of patients. While treatment discontinuation remained stable, treatment initiation decreased after the requirement for MedGuides. This decline differed by patient age, presence of mental comorbidities or disability; while the reduction in treatment intensity did not show significant difference by patient or provider characteristics. A more pronounced elevation in ECG use was observed among patients wi th less complicated mental conditions or patients who were diagnosed by psychiatrists. Clinical consequence s of these changes are uncertain.

PAGE 132

132 Table 5 1. Age distribution of Florida Medic aid feefor service population during 20012008. 2001 2002 2003 2004 2005 2006 2007 2008 Age N % N % N % N % N % N % N % N % 0 4 428,085 19 446,610 19 452,292 19 460,044 19 464,440 19 460,519 19 468,102 20 512,296 19 5 9 283,543 13 294,255 13 293,195 12 294,390 12 305,415 13 294,435 12 290,979 12 310,246 12 10 14 241,772 11 261,471 11 259,769 11 253,128 11 256,882 11 237,029 10 229,273 10 246,139 9 15 20 218,671 10 236,995 10 243,330 10 245,882 10 262,461 11 255,596 11 248,579 10 268,850 10 21+ 1,030,371 47 1,089,834 47 1,131,951 48 1,112,316 47 1,141,590 47 1,151,660 48 1,150,241 48 1,294,996 49 <21 1,172,071 53 1,239,331 53 1,248,586 52 1,253,444 53 1,289,198 53 1,247,579 52 1,236,933 52 1,337,531 51 Total 2,202,442 100 2,329,165 100 2,380,537 100 2,365,760 100 2,430,788 100 2,399,239 100 2,387,174 100 2,632,527 100 Cumulative count per calendar year

PAGE 133

133 Table 5 2 Age distribution of newly diagnosed ADHD patient* in Florida Medicaid feefor service population during 20012008. 2001 2002 2003 2004 2005 2006 2007 2008 Age N % ** N % ** N % ** N % ** N % ** N % ** N % ** N % ** 0 4 864 0.20 833 0.19 1,065 0.24 1,180 0.26 1,001 0.22 858 0.19 605 0.13 496 0.10 5 9 3,205 1.13 3,253 1.11 3,809 1.30 3,998 1.36 3,544 1.16 3,010 1.02 2,147 0.74 2,054 0.66 10 14 1,737 0.72 1,480 0.57 1,581 0.61 1,558 0.62 1,236 0.48 913 0.39 583 0.25 581 0.24 15 20 426 0.19 345 0.15 412 0.17 402 0.16 413 0.16 332 0.13 236 0.09 206 0.08 21+ 70 0.01 81 0.01 101 0.01 190 0.02 173 0.02 153 0.01 145 0.01 202 0.02 <21 6,232 0.53 5,911 0.48 6,867 0.55 7,138 0.57 6,194 0.48 5,113 0.41 3,571 0.29 3,337 0.25 Total 6,302 0.29 5,992 0.26 6,968 0.29 7,328 0.31 6,367 0.24 5,266 0.22 3,716 0.16 3,539 0.13 Cumulative count during calendar year ; **proportion of newly diagnosed patients in Florida Medicaid fee for service population.

PAGE 134

134 APPENDIX A N ATIONAL D RUG CODE (NDC): STIMULANTS 00555 0765 005550766 005550767 005550768 005550764 005550762 005550763 005550790 005550792 005550788 005550791 005550787 005550789 66336*970 63552*035 63552*036 63552*037 63552*385 63552*389 63552*381 63552*669 63552*670 63552*031 63552*032 63552*033 63552*034 63552*671 548684760 548683674 548685470 548685368 548684410 548684640 548685007 548685142 548685140 54092*371 54092*376 54092*373 54092*377 54092*387 54092*383 54092 *381 54092*389 54092*385 54092*374 54092*391 16590*739 16590*738 173140586 173145853 173145850 173145851 173145852 173140585 173140588 636293733 636293735 636293734 636293736 003394169 66336*971 66336*956 59604*585 59604*588 59604*587 59604*586 50580*852 50580*851 50580*853 50580*850 68071*384 68071*383 68071*386 68071*385 50458*585 50458*586 55289*835 55289 *859 55289*975 55289*854 548684957 548684759 548684789 54868 4489 65084*388 65084*387 65084*389 57616*152 57616*153 57616*155 57616*154 54092*552 54092*553 54092*554 54092*555 000743377 67386*102 000743241 551548654 551548655 11014*502 11014*501 11014*500 548683811 548683403 548683402 000073513 000073514 63552*160 63552*161 46672*180 000780493 000780432 000780430 000780431 000780381 000780380 000780382 59917*514 59917*513 548684718 548680668 548685397 53014*576 53014*574 53014 *584 53014*583 53014*582 53014*581 53014*580 53014*579 53014*575 53014*594

PAGE 135

135 APPENDIX B STIMULANT DEFINED DAILY DOSE (DDD) Stimulants DDD A mphetamine 15mg M ethylphenidate 15mg D exmethylphenidate 30mg D extroamphetamine 15mg A tomoxetine 80mg

PAGE 136

136 APPENDIX C N ATIONAL D RUG C ODE (NDC): SECOND LI NE ADHD TREATMENT (BUPROPION) 00024 5810 62660*581 000245811 62660*811 62660*812 24236*446 35356*087 511294067 548685927 000935501 000935502 18837*246 21695*577 001152442 001152444 511293066 648962442 648962444 511293192 551548250 21695*578 000935350 16590*589 18837*354 648960682 000935351 35356*918 68071*679 000935703 001152333 648962333 006155535 16590*474 55887*496 67544*647 11014*522 550453197 11014*523 18837*287 21695*295 58016*881 21695 *019 21695 *020 24236*491 58016*240 67544*499 63552*555 005910839 005913385 511293273 68115*447 511292479 63552*553 682589238 21695*017 21695*018 24236*191 33358*053 511293653 511293691 545695391 550453195 58016*584 58864*808 66267*614 67544*547 68115*412 68115*445 68115*848 10370*101 10370*102 636292873 001791629 001791634 005910867 672280244 672280245 672280246 68071*122 62660*043 62660*042 12280*165 511294058 49999 *349 67767 *117 67767*142 67816*507 006156561 58437*023 24236*910 260530420 24236*808 58437*024 58437*025 58437*026 61392*520 61392*521 66336*897 63552*552 61392*996 000930280 000930290 003780433 003780435 004800280 004800290 006155554 007811053 007811064 178560435 18837*298 234901071 234905173 24236*018 24236*019 24236*543 43806*280 43806*290 498560157 498560158 51079*943 51079*944 511292390 511294206 548684134 54868 4550 60505 0157 605050158 61392*997 63739*317 63739*318 66267*700 67544*299 67544*406 67544*673 67544*892 001850415 006156553 007811529 65084*414 67767*133 67767*141 67816*508 234901072 68116*676 005910858 234907689 58864*840 647250390 234901073 548684892 260530075 54348*293 548685377 001737020 001737019 260530076 63552*554 55887*151 55887*650 636293233 636293449 548685736 67544*464 58016*024 24236*138 21695*137 21695 *138 67801 *433 001737042 001737050 548683984 548684505 550452631 55289*905 578660901 58016*599 58016*734 58864*625 66105*474 66105*480 68115*901 511291308 63552*135 548684763 001730722 63552*137 63552*133 001730177 001730178 548681449 548681450 63552*177 63552*178 68115*926 551541129 58016*671 636293315 001730947 001730135 584864935 49999*774 001730730 001730731 62660*037 16590*526 178560730 511292672 511293631 55289 *900 51129 3693 511293714 68084*252 001851111 008551597 008551598 004907163 001155445 001850410 005913331 005913332 178560415 551545472 55289*733 58016*722 58864*794 60429*746 60429*747 62660*038 55289*922 58016*031 68115*810 68071*478 68115*825 538730135 001737061 58016*871 66105*153 511291340 63552*556 001730556

PAGE 137

137 APPENDIX D NATIONAL DRUG CODE: SECOND LINE ADHD TREATMENT (OTHERS) 54868 2546 009041570 009041571 009041572 009041573 511291208 511291749 511291943 511293598 511293765 52959*128 548681019 548682548 550451612 550451718 550451719 58016*502 58016*503 66267*619 66267*626 68071*210 68071*439 68115*482 68115*509 68115*591 511291579 539782076 539782077 62584*661 62584*663 001791557 005910808 005910809 006033166 006033167 006033168 006771198 511291967 516724001 516724002 516724003 516724004 525494001 52549 4002 00677 1199 007811973 21695*428 24236*772 260530105 511293054 52152*341 52152*342 52152*343 52152*344 52152*345 52152*346 61392*891 61807*130 676510011 676510012 676510013 001850019 001850029 001850721 001850722 001850736 001850760 007811971 007811972 007811974 007811975 007811976 51079*490 511293243 55887*143 55887*144 16590*484 58016*853 000680720 000680007 000680011 24236*478 24236*566 24236*567 545693849 545694146 548682481 54868 2482 00068 0015 000680019 000680020 000680021 511293819 001791797 006770421 007811762 009040925 24236*941 49999*400 511291443 511291445 511292854 511293668 511293955 53489*330 53489*331 53489*332 54738*913 54738*914 548681344 548682221 550451794 578663930 672280069 68071*474 68115*439 007811764 007811766 62584*751 62756*896 55289*144 006034044 006034045 006770422 006770423 009045030 009045032 18837*114 18837*310 234906020 24196*517 24196 *518 00904 0927 009040929 24236*075 24236*152 24236*601 24236*813 49884*054 49884*055 49884*056 511292471 511293056 52152*323 52152*324 52152*330 530021066 539780073 545690194 54738*912 548682571 55289*149 58864*292 61392*025 61392*026 61392*027 66116*253 672280068 000932111 000932113 000932117 006034043 62584*750 62756*895 004069931 004069932 004069933 004069934 664060032 636292833 49999*538 511293048 003781410 003782325 003783250 00378 4175 66406 0033 664060034 664060035 55887*288 16590*577 234900332 51655*148 550451722 55289*839 58016*841 58016*866 60760*117 66267*266 62584*749 146560065 000280022 000280040 000280045 000280020 004069923 004069924 004069925 004069926 664060028 664060029 664060030 664060031 000280032 000280136 000280140 004069920 004069921 004069922 664069920 664069922 664069921 58864*844 66336*973 68071*325 68071*326 68115*261 68115*262 68115*418 68387 *330

PAGE 138

138 APPENDIX D NATIONAL DRUG CODE: SECOND LINE ADHD TREATMENT (OTHERS) 52549 4001 525494002 525494003 525494004 55160*116 55160*117 55289*099 58016*491 60760*811 61392*370 63739*189 63739*190 66116*048 66336*757 67544*874 687885786 687885787 000930810 000930811 000930812 000930813 006151306 006151307 009047939 009047940 00904 7941 54868 2481 548682482 55160*114 55160*115 009047942 51079*803 51079*805 001210678 516724119 525494119 63304*202 67544*832 511292990 004069913 000780016 000780078 000780086 000780087 004069910 004069911 004069912 664060036 664060037 664060038 664060039 00043 6239 24196 *517 24196*518 24196*519 24196*520 004069918 61392*361 61392*364 61392*367 49999*215 55887*439 55887*498 55887*518 12634*680 16590*435 16590*510 548682480 548682835 58016*875 67544*078 68084*031 68084*032 001791445 511291611 003395804 005915788 00591 5789 00378 3250 003784175 005915786 005915787 66336*621 009045031 13411*404 13411*405 21695*093 21695*094 24236*858 511291259 511291578 511293704 511293884 52959*358 52959*359 52959*519 52959*840 550451920 550451956 550451982 55289*586 58016*508 58016*519 60760 *508 68115 *418 68387*330 68071*323 009045029 66267*484 61807*142 66267*483

PAGE 139

139 APPENDIX E CPT (CURRRENT PROCEDURAL TERMINOLOGY) CODES: E LECTRO C ARDIO G RAPHY (ECG) ECG 93224 93237 93000 93010 93014 93270 93272 93271 93042 93025 93041 93040 93278 93012 93025 93278

PAGE 140

140 LIST OF REFERENCES 1. Health Canada. Health Canada Endorsed Important Safety Information on Adderall XR. http://www.hc sc.gc.ca/ahc asc/media/advisories avis/_2005/2005_01eng.php. Accessed June 13, 2010. 2. Health Canada. Report of the "Adderall XR New Drug Committee". http://www.hc sc.gc.ca/dhp mps/prodpharma/activit/new drugcom nouvelledrogue/ndca_rep_cnma_rap_20050825eng.php. Accessed June 13, 2010. 3. Health Canada. Health Canada Important Safety Information on ADDERALL XR TM (amphetamine salts). http://www.hc sc.gc.ca/dhpmps/medeff/advisories avis/prof/_2005/adderall_xr_hpc cps eng.php. Accessed June 13, 2010. 4. F ood and D rug A dm inistration. Public Health Advisory for Adderall and Adderall XR. http: //www.fda.gov/CDER/Drug/advisory/adderall.html Accessed April 13, 2009. 5. Nissen SE. ADHD Drugs and Cardiovascular Risk. N Engl J Med. April 6, 2006 2006;354(14):14451448. 6. Newsbrief. FDA Pediatric Advisers Say Nay to ADHD Drug Black Box. http://www.medpagetoday.com/Psychiatry/ADHD ADD/2921. Accessed June 13, 2010. 7. Kernan WN, Viscoli CM, Brass LM. Phenylpropanolamine and the risk of hemorrhagic stroke. N Engl J Med. 2000;343:18261832. 8. F ood and D rug A dministration. FDA Directs ADHD Drug Manufacturers to Notify Patients about Cardiovascular Adverse Events and Psychiatric Adverse Events. http://www.fda.gov/NewsEvents/ Newsroom/PressAnnouncements/2007/ucm108 849.htm. Accessed June 13, 2010. 9. CDC. Mental health in the United States. Prevalence of diagnosis and medication treatment for attentiondeficit/hyperactivity disorder --United States, 2003. MMWR Morb Mortal Wkly R ep. Sep 2 2005;54(34):842847. 10. Wysowski DK, Corken A, Gallo Torres H, Talarico L, Rodriguez EM. Postmarketing reports of QT prolongation and ventricular arrhythmia in association with cisapride and Food and Drug Administration regulatory actions. Am J Gastroenterol. Jun 2001;96(6):16981703. 11. Rossouw JE, Anderson GL, Prentice RL, et al. Risks and benefits of estrogen plus progestin in healthy postmenopausal women: principal results From the Women's Health Initiative randomized controlled trial. JAM A Jul 17 2002;288(3):321333.

PAGE 141

141 12. Hammad TA, Laughren T, Racoosin J. Suicidality in pediatric patients treated with antidepressant drugs. Arch Gen Psychiatry. Mar 2006;63(3):332339. 13. Administration UFaD. Relationship between psychotropic drugs and pediatric suicidality: review and evaluation of clinical data. http://www.fda.gov/ohrms/dockets/ac/04/briefing/20044065b110TAB08 Hammads Review.pdf Acce ssed May 18, 2010. 14. Nissen SE, Wolski K. Effect of rosiglitazone on the risk of myocardial infarction and death from cardiovascular causes. N Engl J Med. Jun 14 2007;356(24):24572471. 15. Kline GS. Study No.: ZM2005/0018/01: Avandia Cardiovascular Ev ent Modeling Project Study No.: HM2006/00497/00 /WEUSRTP866: Coronary Heart Disease Outcomes in Patients Receiving Antidiabetic Agents. http://ctr.gsk.co.uk/Summary/ Rosiglitazone/III_CVmodeling.pdf. Accessed Jul y 7, 2008. 16. APA. Diagnostic and statistical manual of mental disorders: DSM IV TR 4th ed. Washington, DC: American Psychiatric Association; 2000. 17. Elia J, Ambrosini PJ, Rapoport JL. Treatment of attentiondeficit hyperactivity disorder. N Engl J Med. Mar 11 1999;340(10):780788. 18. DiScala C, et al. Injuries to Children With Attention Deficit Hyperactivity Disorder. Pediatrics. December 1998;102(6):14151421. 19. Clinical practice guideline: treatment of the school aged child with attentiondeficit/hyperactivity disorder. Pediatrics. Oct 2001;108(4):10331044. 20. Lerner M, Wigal T. Longterm safety of stimulant medications used to treat children with ADHD. Pediat r Ann. Jan 2008;37(1):3745. 21. King S, Griffin S, Hodges Z, et al. A systematic review and economic model of the effectiveness and cost effectiveness of methylphenidate, dexamfetamine and atomoxetine for the treatment of attention deficit hyperactivity disorder in children and adolescents. Health Technol Assess. Jul 2006;10(23):iii iv, xiii146. 22. Brown RT, Amler RW, Freeman WS, et al. Treatment of attentiondeficit/hyperactivity disorder: overview of the evidence. Pediatrics. Jun 2005;115(6):e749757. 23. Russell B. AttentionDeficit Hyperactivity Disorder: A Handbook for Diagnosis and Treatment New York: The Guildford Press; 2006.

PAGE 142

142 24. Sankaranarayanan J, Puumala SE, Kratochvil CJ. Diagnosis and treatment of adult attentiondeficit/hyperactivity dis order at US ambulatory care visits from 1996 to 2003. Curr Med Res Opin. Aug 2006;22(8):14751491. 25. Zito JM, Safer DJ, dosReis S, Magder LS, Gardner JF, Zarin DA. Psychotherapeutic medication patterns for youths with attentiondeficit/hyperactivity dis order. Arch Pediatr Adolesc Med. Dec 1999;153(12):12571263. 26. Winterstein AG, Gerhard T, Shuster J, et al. Utilization of pharmacologic treatment in youths with attention deficit/hyperactivity disorder in Medicaid database. Ann Pharmacother. Jan 2008;42(1):2431. 27. Chen CY, Gerhard T, Winterstein AG. Determinants of initial pharmacological treatment for youths with attentiondeficit/hyperactivity disorder. J Child Adolesc Psychopharmacol. Apr 2009;19(2):187195. 28. Dosreis S, Zito JM, Safer DJ, Soe ken KL, Mitchell JW, Jr., Ellwood LC. Parental perceptions and satisfaction with stimulant medication for attentiondeficit hyperactivity disorder. J Dev Behav Pediatr. Jun 2003;24(3):155162. 29. Bussing R, Zima BT, Gary FA, Garvan CW. Barriers to detect ion, helpseeking, and service use for children with ADHD symptoms. J Behav Health Serv Res. Apr Jun 2003;30(2):176189. 30. Bussing R, Koro Ljungberg ME, Gary F, Mason DM, Garvan CW. Exploring helpseeking for ADHD symptoms: a mixedmethods approach. Har v Rev Psychiatry. Mar Apr 2005;13(2):85101. 31. Bussing R, Gary FA, Mills TL, Garvan CW. Parental explanatory models of ADHD: gender and cultural variations. Soc Psychiatry Psychiatr Epidemiol. Oct 2003;38(10):563575. 32. Stevens J, Harman JS, Kelleher KJ. Race/ethnicity and insurance status as factors associated with ADHD treatment patterns. J Child Adolesc Psychopharmacol. Feb 2005;15(1):8896. 33. Zito JM, Safer DJ, Riddle MA, Johnson RE, Speedie SM, Fox M. Prevalence variations in psychotropic treatment of children. J Child Adolesc Psychopharmacol. 1998;8(2):99105. 34. Cox ER, Motheral BR, Henderson RR, Mager D. Geographic variation in the prevalence of stimulant medication use among children 5 to 14 years old: results from a commercially insured US sample. Pediatrics. Feb 2003;111(2):237243.

PAGE 143

143 35. Sherman M, Hertzig ME. Prescribing practices of Ritalin: the Suffolk County, New York Study. In: L G, B O, eds. Ritalin: Theory and Management. New York, NY: Many Ann Liebert Inc; 1991:187193. 36. Thiruchelvam D, Charach A, Schachar RJ. Moderators and mediators of longterm ad herence to stimulant treatment in children with ADHD. J Am Acad Child Adolesc Psychiatry. Aug 2001;40(8):922928. 37. Gau SS, Chen SJ, Chou WJ, et al. National survey of adherence, efficacy, and side effects of methylphenidate in children with attentiond eficit/hyperactivity disorder in Taiwan. J Clin Psychiatry. Jan 2008;69(1):131140. 38. Faraone SV, Biederman J, Zimmerman B. An analysis of patient adherence to treatment during a 1year, openlabel study of OROS methylphenidate in children with ADHD. J Atten Disord. Sep 2007;11(2):157166. 39. dosReis S, Zito JM, Safer DJ, Soeken KL. Mental health services for youths in foster care and disabled youths. Am J Public Health. Jul 2001;91(7):10941099. 40. Guo JJ, Curkendall S, Jones JK, Fife D, Goehring E, She D. Impact of cisapride label changes on codispensing of contraindicated medications. Pharmacoepidemiol Drug Saf. Jun 2003;12(4):295301. 41. Olfson M, Marcus SC, Zhang HF, Wan GJ. Continuity in methylphenidate treatment of adults with attentiondefic it/hyperactivity disorder. J Manag Care Pharm. Sep 2007;13(7):570577. 42. Marcus SC, Wan GJ, Kemner JE, Olfson M. Continuity of methylphenidate treatment for attentiondeficit/hyperactivity disorder. Arch Pediatr Adolesc Med. Jun 2005;159(6):572578. 43. F ood and D rug A dministration. Adderall XR label. http://www.fda.gov/cder/foi/label/2007/021303s015lbl.pdf. Accessed May 18, 2009. 44. F ood and D rug A dministration. Adderall label. http://www.fda.gov/cder/foi/label/2007/011522s040lbl.pdf. Accessed May 18, 2009. 45. F ood and D rug A dministration. Strattera labe l. http://www.fda.gov/cder/foi/label/2008/021411s024s025s026lbl.pdf. Accessed April 17 2009. 46. F ood and D rug A dministration. Ritalin label. http: //www.fda.gov/cder/foi/label/2007/010187s069,018029s040,021284s011lbl.p df Accessed April 13, 2009.

PAGE 144

144 47. Findling RL, Short EJ, Manos MJ. Short term cardiovascular eff ects of methylphenidate and Adderall. J Am Acad Child Adolesc Psychiatry. 2001;40:525529. 48. Stowe CD, Gardner SF, Gist CC. 24 hour ambulatory blood pressure monitoring in male children receiving stimulant therapy. Ann Pharmacother. 2002;36(1142 1149). 49. Wesler RH, Biederman J, Spencer TJ. Longterm cardiovascular effects of mixed amphetamine salts extended release in adults with ADHD. CNS Spectr. 2005;10 (12 suppl29):3543. 50. Wilens TE, Hammerness PG, J B. Blood pressure changes associated with medication treatment of adults with ADHD. J Clin Psychiatry. 2005;66:253259. 51. Biederman J, Mick E, Surman C. A randomized, placebocontrolled trial of OROS methlphenidate in adults with attentiondeficit/hyperactivity disorder. Biol Psychiatry. 2006;59:829835. 52. Samuels JA, Franco K, Wan F. Effect of stimulants on 24h ambulatory blood pressure in children with ADHD: a doubleblind, randomized, cross over trial. Pediatr Nephrol. 2006;21:9295. 53. Wilens TE, Prince JB, Spencer TJ, Biederman J. Stimulants and sudden death: what is a physician to do? Pediatrics. 2006;118:12151219. 54. Winterstein AG, Gerhard T, Shuster J, Johnson M, Zito JM, Saidi A. Cardiac safety of central ner vous system stimulants in children and adolescents with attentiondeficit/hyperactivity disorder. Pediatrics. Dec 2007;120(6):e14941501. 55. FDA. Guidance: Drug Safety Information FDA's Communication to the Public. 2007. 56. Smalley W, Shatin D, Wysowski DK, et al. Contraindicated use of cisapride: impact of food and drug administration regulatory action. JAMA. Dec 20 2000;284(23):30363039. 57. Graham DJ, Drinkard CR, Shatin D, Tsong Y, Burgess MJ. Liver enzyme monitoring in patients treated with troglitazone. JAMA. Aug 15 2001;286(7):831833. 58. Hersh AL, Stefanick ML, Stafford RS. National use of postmenopausal hormone therapy: annual trends and response to recent evidence. JAMA. Jan 7 2004;291(1):4753.

PAGE 145

145 59. Valuck RJ, Libby AM, Orton HD, Morrato EH, Allen R, Baldessarini RJ. Spillover effects on treatment of adult depression in primary care after FDA advisory on risk of pediatric suicidality with SSRIs. Am J Psychiatry. Aug 2007;164(8):11981205. 60. Libby AM, Brent DA, Morrato EH, Orton HD, All en R, Valuck RJ. Decline in treatment of pediatric depression after FDA advisory on risk of suicidality with SSRIs. Am J Psychiatry. Jun 2007;164(6):884891. 61. Starner CI, Schafer JA, Heaton AH, Gleason PP. Rosiglitazone and pioglitazone utilization from January 2007 through May 2008 associated with five risk warning events. J Manag Care Pharm. Jul Aug 2008;14(6):523531. 62. McKinlay JB. Some approaches and problems in the study and use of services -and overview. J Health Soc Behav. 1972;13:115152. 63. Pescosolido BA. Of pride and prejudice: the role of sociology and social networks in integrating the health sciences. J Health Soc Behav. Sep 2006;47(3):189208. 64. Williams JW, Torrens PR. Introduction to health services New York: Wiley; 1988. 65. Anderson. Revisiting the behavioral model and access to medical care: does it matter? J Health Soc Behav. 1995;36:110. 66. Anderson R. A Behavioral Model of Families' Use of Health Services. Research Series No.25. Chicago, IL; 1968. 67. Pescosolido BA. How people get into mental health services: stories of choice, coercion and "muddling through" from "first timers". Soc Sci Med. 1998;2:275286. 68. Pescosolido BA. Beyond rational choice: The social dynamics of how people see help. American Journal of S ociology. 1992;97:10961138. 69. Pescosolido BA. Illness, carrers and network ties: Aconceptual model of utilization and compliance. Advances in Meddical Sociology. 1991;2:161184. 70. Becker MH. The health belif model and personal health behavior. Healt h Educ Monogr. 1974;2:324473. 71. Rosenberg IM. The health belif model and preventive health behavior. Health Educ Monogr. 1974;2:354386. 72. Pescosolido BA, Baoyer C. How people come to use mental health services? Current knowledge and changing perspectives. New York: Cambridge University Press; 1999.

PAGE 146

146 73. Guay MP, Dragomir A, Pilon D, Moride Y, Perreault S. Changes in pattern of use, clinical characteristics and persistence rate of hormone replacement therapy among postmenopausal women after the WHI publication. Pharmacoepidemiol Drug Saf. Jan 2007;16(1):1727. 74. Gerend MA, Aiken LS, Erchull MJ, Lapin A. Women's use of hormone therapy before and after the Women's Health Initiative: a psychosocial model of stability and change. Prev Med. Sep 2006;43(3):158164. 75. McKenzie DA, Semradek J, McFarland BH, Mullooly JP, McCamant LE. The validity of medicaid pharmacy claims for estimating drug use among elderly nursing home residents: The Oregon experience. J Clin Epidemiol. Dec 2000;53(12):12481257. 76. FDA. Atomoxetine (marketed as Strattera) Information. 2005. 77. Kim HJ, Fay MP, Feuer EJ, Midthune DN. Permutation tests for joinpoint regression with applications to cancer rates. Stat Med. Feb 15 2000;19(3):335351. 78. Weir HK, Thun MJ, Hankey BF, et al. Annual report to the nation on the status of cancer, 19752000, featuring the uses of surveillance data for cancer prevention and control. J Natl Cancer Inst. Sep 3 2003;95(17):12761299. 79. Martin RM, May M, Gunnell D Did intense adverse media publicity impact on prescribing of paroxetine and the notification of suspected adverse drug reactions? Analysis of routine databases, 20012004. Br J Clin Pharmacol. Feb 2006;61(2):224228. 80. Nemeroff CB, Kalali A, Keller MB, et al. Impact of publicity concerning pediatric suicidality data on physician practice patterns in the United States. Arch Gen Psychiatry. Apr 2007;64(4):466472. 81. Kim HJ, Fay MP, Yu B, Barrett MJ, Feuer EJ. Comparabil ity of segmented line regression models. Biometrics. Dec 2004;60(4):10051014. 82. Vetter VL, Elia J, Erickson C, et al. Cardiovascular monitoring of children and adolescents with heart disease receiving medications for attention deficit/hyperactivity dis order [corrected]: a scientific statement from the American Heart Association Council on Cardiovascular Disease in the Young Congenital Cardiac Defects Committee and the Council on Cardiovascular Nursing. Circulation. May 6 2008;117(18):24072423. 83. Ame rican Academy of Pediatrics/American Heart Association clarification of statement on cardiovascular evaluation and monitoring of children and adolescents with heart disease receiving medications for ADHD. 2008.

PAGE 147

147 84. Kurian BT, Ray WA, Arbogast PG, Fuchs DC, Dudley JA, Cooper WO. Effect of regulatory warnings on antidepressant prescribing for children and adolescents. Arch Pediatr Adolesc Med. Jul 2007;161(7):690696. 85. Medication Guides for Prescription Drug Products. Code of Federal Regulations 2004. Tit le 21;Pt 208:111 14. 86. Prescription drug product labeling: FDA Medication Guide requirements. Final rule.Fed Regist. 1998;63:66378400. 87. Wolf MS, Davis TC, Shrank WH, Neuberger M, Parker RM. A critical review of FDAapproved Medication Guides. Patie nt Educ Couns. Sep 2006;62(3):316322. 88. Morrato EH, Libby AM, Orton HD, et al. Frequency of provider contact after FDA advisory on risk of pediatric suicidality with SSRIs. Am J Psychiatry. Jan 2008;165(1):4250. 89. Rijnbeek PR, van Herpen G, Kapusta L, Ten Harkel AD, Witsenburg M, Kors JA. Electrocardiographic criteria for left ventricular hypertrophy in children. Pediatr Cardiol. Sep 2008;29(5):923928. 90. Ramaswamy P, Patel E, Fahey M, Mahgerefteh J, Lytrivi ID, Kupferman JC. Electrocardiographic predictors of left ventricular hypertrophy in pediatric hypertension. J Pediatr. Jan 2009;154(1):106110. 91. Fogel MA, Lieb DR, Seliem MA. Validity of electrocar diographic criteria for left ventricular hypertrophy in children with pressureor volumeloaded ventricles: comparison with echocardiographic left ventricular muscle mass. Pediatr Cardiol. Nov Dec 1995;16(6):261269. 92. Posner K, Melvin GA, Murray DW, e t al. Clinical presentation of attentiondeficit/hyperactivity disorder in preschool children: the Preschoolers with AttentionDeficit/Hyperactivity Disorder Treatment Study (PATS). J Child Adolesc Psychopharmacol. Oct 2007;17(5):547562. 93. Connor DF, E dwards G, Fletcher KE, Baird J, Barkley RA, Steingard RJ. Correlates of comorbid psychopathology in children with ADHD. J Am Acad Child Adolesc Psychiatry. Feb 2003;42(2):193200. 94. Pugh MJ, Anderson J, Pogach LM, Berlowitz DR. Differential adoption of pharmacotherapy recommendations for type 2 diabetes by generalists and specialists. Med Care Res Rev. Jun 2003;60(2):178200.

PAGE 148

148 95. 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):526. 96. Chin MH, Zhang JX, Merrell K. Specialty differences in the care of older patients with diabetes. Med Care. Feb 2000;38(2):131140. 97. Chin MH, Friedmann PD, Cassel CK, Lang RM. Differences in generalist and specialist physicians' knowledge and use of angiotensinconverting enzyme inhibitors for congestive heart failure. J Gen Intern Med. Sep 1997;12(9):523530.

PAGE 149

149 BIOGRAPHICAL SKETCH Chih ying Chen was born and raised in Taipei, Taiwan. She received her bachelors degree in Pharmacy from N ational Taiwan University in 2001 and worked at the National Adverse Drug Reaction Reporting Center for one year. In 2004, after she recieved her Master of Health Administration degree from University of Pittsburgh she joined the Department of P harmaceutical Outcomes & Policy at the University of Florida where she was trained as a pharmacoepidemiologist. Her research interests focus on drug uti lization, safety and effectiveness