|UFDC Home||myUFDC Home | Help|
This item has the following downloads:
1 EFFECTS OF PATIENT CENTERED MEDICAL HOMES ON EXPENDITURES OF HEALTHY POPULATIONS By DANIEL ESTRADA 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 2012
2 2012 DANIEL ESTRADA
3 To Aiden and Caleb, you both taught me to see without looking
4 ACKNOWLEDGMENTS I have met and been influenced my many people in my journey through life. I am indebted to all these individuals in their small contributions to my collective self. In this particular endeavor, I would like to highlight the contributions from the Departm ent of Health Services Research and Management Policy. Dr. R. Paul Duncan has been a constant throughout my many years of interaction with the department and its continued evolution. I am indebted for the stability that he has brought. Dr. Jeffery Harman, the chairman of my committee, who has shown tremendous patience in working with this slightly distracted student. I am incredibly thankful for his technical expertise and knowledge of the nuances of Econometrics. Finally, I am thankful to all the other faculty members of the department, past and present, who have always shown a welcome and open door to all the students in the program. I can only humbly thank my family for all the support, inspiration and love they have given and continue to give me. These words cannot fully convey the extent of my gratitude. I will forever be thankful to my wonderful wife, Amara, who always pushes me to do more and gives me unconditional support; my openhearted daughter, Sofia, who can mak e all troubles melt away with a simple smile and giggle; my sweet innocent boys, Aiden and Caleb, who have taught me perspective in life and the beauty of simple thin gs. I would also like to thank my parents, Robert and Diana, for their absolute love and belief that I can do no wrong and my inlaws, Marshall and Susan, f or their help and guidance and treating me like their own son along the way.
5 TABLE OF CONTENTS page ACKNOWLEDGMENTS .................................................................................................. 4 LIST OF TABLES ............................................................................................................ 7 LIST OF FIGURES .......................................................................................................... 8 LIST OF ABBREVIATIONS ............................................................................................. 9 ABSTRACT ................................................................................................................... 10 CHAPTER 1 INTRODUCTION .................................................................................................... 12 2 BACKGROUND ...................................................................................................... 17 History and Structure of Patient Centered Medical Homes ..................................... 17 Effects of Patient Centered Medical Homes ........................................................... 20 Outcomes of PCMH populations ...................................................................... 20 Expenditures of PCMH populations .................................................................. 24 Barriers to Implementation of Patient Centered Medical Homes ............................ 26 Policy Implications .................................................................................................. 29 Conceptual Model ................................................................................................... 30 3 METHODS .............................................................................................................. 39 Overview ................................................................................................................. 39 Research Questions and Hypothesis ............................................................... 39 Hypothesis 1 .............................................................................................. 39 Hypothesis 2 .............................................................................................. 39 Hypothesis 2 a ........................................................................................... 40 Data Source ............................................................................................................ 40 Outcome Measures ................................................................................................ 41 Total Expenditur es ........................................................................................... 41 Utilization of Health Care Services ................................................................... 41 Primary Explanatory Variables ................................................................................ 42 Patient Centered Medical Homes ..................................................................... 42 Pa tient Centered Medical Homes Comprehensive Definition ........................... 43 Enhanced access ....................................................................................... 43 Continuity of care ....................................................................................... 44 Comprehensiveness of care ...................................................................... 44 Coordi nation of care ................................................................................... 44 Quality and safety of care .......................................................................... 45 Patient Centered Medical Homes Simplified Definition .................................... 45
6 Office hours ................................................................................................ 45 After hours contact ..................................................................................... 45 Shared decision making ............................................................................. 46 Usual source of care .................................................................................. 46 Language ................................................................................................... 46 Healthy Individuals ........................................................................................... 47 Control Variables .............................................................................................. 48 Additional Analysis of Interest .......................................................................... 49 Methods .................................................................................................................. 49 4 RESULTS ............................................................................................................... 53 Descriptive Statistics ............................................................................................... 53 Bivariate Statistics .................................................................................................. 55 PCMH S implified Definition .............................................................................. 55 PCMH Comprehensive Definition ..................................................................... 56 Expenditures ........................................................................................................... 57 Utilization ................................................................................................................ 59 Multivariate Analysis ............................................................................................... 60 5 DISCUSSI ON ......................................................................................................... 78 Comparable Results ............................................................................................... 81 Policy Implications .................................................................................................. 86 Comparison of Patient Centered Medical Home Definitions ................................... 91 Study Limitations .................................................................................................... 92 Future Research ..................................................................................................... 95 LIST OF REFERENCES ............................................................................................... 98 BIOGRAPHICAL SKETCH .......................................................................................... 104
7 LIST OF TABLES Table page 4 1 Independent variable characteristics .................................................................. 64 4 2 Bivariate analysis with simplified PCMH status .................................................. 65 4 3 Bivariate analysis with comprehensive PCMH status ......................................... 66 4 4 GLM model predicting expenditures (Ga mma link) ............................................. 67 4 5 Log linear model predicting expenditures ........................................................... 68 4 6 GLM model predicting expenditures (Poisson) ................................................... 69 4 7 Summary of model predicted expenditures ........................................................ 70 4 8 Pregibons link test results .................................................................................. 70 4 9 Hosmer Lemeshow test ...................................................................................... 70 4 10 GLM model predicting events (Gamma) ............................................................. 71
8 LIST OF FIGURES Figure page 2 1 Relationship of Stock of Health Capital and Cost of Capital ............................... 37 2 2 Interaction of Grossmans Health Capital and Donabedians StructureProcessOutcomes with Patient Centered Medical Homes ................................ 38 4 1 Expenditures by PCMH definition ....................................................................... 72 4 2 Expenditure categories by simplified PCMH ....................................................... 73 4 5 Utilization by categories (simplified definition) .................................................... 76 4 6 Utilization by categories (comprehensive definition) ........................................... 77
9 LIST OF ABBREVIATION S CCNC Community Care of North Carolina CMS Center for Medicare and Medicaid Services HEDIS Healthcare Effectiveness Data and Information Set IOM Institute of Medicine MEPS M edical Expenditure Panel Survey NCQA National Committee for Quality Assurance PCMH Patient Centered Medical Home
10 Abstract of Disser tation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy EFFECTS OF PATIENT CENTERED MEDICAL HOMES ON EXPENDITURES OF HEALTHY POPULATIONS By Dani el Estrada December 2012 Chair: Jeffery S. Harman Major: Health Services Research The U S Healthcare system is under increasing pressure from a variety of fronts. Healthcare expenditures continue to rise at an unsustainable rate. The disease and illnesses affecti ng the U.S. population continue to shift from episodic critical events to a chro nic disease state exacerbated by the aging baby boomer population. Finally, healthcare resources such as providers and facilities have been slow to change practice patterns to account for this new frequent visit, patient management care del ivery system. Several policy proposals have been presented to address these various shortcomings, including increasing emphasis on primary care and empowering patients with more information about their care and health choices. One particular concept that has emerged as a potential solution is Patient Centered Medical Homes (PCMH). PCMHs originated as a coordinated care model for children with special health care needs. The concept is to provide care that is continuous efficient, coordinated and engaged with the patient and family. Unfortunately from a policy implementation standpoint, these broad concepts have been difficult to define and no consensus exists as to what makes a PCMH. PCMHs have proven to be effective at reducing healthcare
11 expenditures, particular ly hospital ER visits, and increasing patient satisfaction for children with special health care needs and chronic disease populations. It is for these reasons that PCMHs are being advocated as a possible solution to control expenditures and increase quality of care. The potential widespread implementation of PCMHs raise an important research question as to its effectiveness on healthy populations. This study seeks to examine the impact of PCMHs on healthy populations utilizing the Medical Expenditure Panel Survey (MEPS). The study identifies characteristics of the survey respondents to identify the existence of a PCMH along with respondent reported health status to analyze the impact of PCMHs on healthy populations. The study provides important context to the discussion of utilizing PCMHs to address our healthcare system dilemma. PCMHs inherently require significant implementation costs, either through personnel or technology. The investment of these resources might be better suited to specific subpopulations such as CSHCN or chronic diseases rather than the general population.
12 CHAPTER 1 INTRODUCTION Patient centered medical homes (PCMH) have been around as a concept of enhancing patient physician outcomes and quality since the late 1960s (Sia, Tonniges, Osterhus, & Taba, 2004) The initial application of this concept was for children with special health care needs (Sia, et al., 2004) Several recent milestones have occurred at national levels that have brought increased interest in PCMHs. The Institute of Medicine first made patient centered care as one of its six domains of quality in its 2001 report for reforming the healthcare system (Institute of Medicine (U.S.). Committee on Quality of Health Care in America., 2001) Recently in March 2007 the American Academy of Family Physicians, American Academy of Pediatrics, American College of Physicians, and the American Osteopathic Association issued the Joint Principles of the Patient Centered Medical Home. This statement continued to lend credibility to the movement and advocates for its incorporation in primary care. In addition, CMS start ed a pilot program in 2010 to enroll a subset of its population in s everal medical homes across the country (CMS, 2007) Finally, the recent healthcar e policy debate of 2009 in the United States Congress has brought increased interest in alternative approaches, such as PCMHs, to delivering healthcare that could potentially reduce or maintain costs, improve access and improve patient outcomes. Despite t his growing interest in PCMHs there are no concrete or consensus definitions of PCMHs. Several influential organizations have developed PCMH criteria that will probably influence the ultimate definition. The NCQA has developed scoring criteria similar to its accreditation process for hospitals in which physician practices can be recognized as a PCMH (Stange et al., 2010) The Commonwealth Fund developed
13 its criteria as part of a large 2007 survey t o assess the health outcomes of medical homes (Rittenhouse, Thom, & Schmittdiel, 2010) Finally CMS has modified the NCQA scorin g criteria to identify and score physician practices participating in its PCMH pilot program (Maxfield, Centers for, Medicaid Services Office of Research, Information, & Mathematica Policy Research, 2008) While the specifics attributes of a PCMH may be in short supply, several consistent broad themes emerge regardless of definition: 1. Conti nuity of care; 2. Clinical information systems; 3. Delivery system design; 4. Decision support; 5. Patient/Family engagement; and 6. Care coordination Looking past the loose definition of PCMHs, the majority of research related to PCMHs has been in regard to chronically ill or disadvantaged populations. A review of the National Survey of Children with Special Health Care Needs showed that most of this population has access to some or all components of a Patient Centered Medical Home and that the presence o f a PCMH has a positive impact on fulfilling care needs and reducing missed school days (Homer et al., 2008) Other studies have shown that PCMHs improve the access to care, the timely receipt of needed services and the management of chronic conditions for minority populations (Kane, 2007) The increasingly fractured debate on healthcare at the state and federal level has led to several proposals to incorporate PCMHs as one of the policy solutions to our healthcare dilemma. The State of Florida in Senate Bill 1986 (2009) proposed the creation of a medical home pilot project with the hopes reducing costs and improving care to the states Medicaid population. The State of Ohio in House Bil l 198 (2010) authorized the creation of a pilot project to convert some primary care practices to patient centered medical homes and overhaul payments to these practices to reflect the
14 new model. The New Jersey legislature passed a law requiring the creat ion of a three year medical home demonstration project for its Medicaid population (Kaye, 2009) There are several other states with various proposals worki ng through their legislatures or already in effect (Kaye, 2009) However, some of the proposed applications of PCMH would implement this model among healthy populations specifically Medicaid populations and little is known on whether a PCMH model is effective when applied to healthier individuals which is important given the cost of transforming a practice to a PCMH The majority of the population is heal thy and while the healthy population does not consume the majority of healthcare services, their large numbers make their overall healthcare consumption dollars significant. Therefore, a small change in healthcare consumption in the healthy population could also have a significant impact on overall healthcare costs. One study looking at a pilot program of PCMHs on a 1.5 million member panel estimated that up to 28% of total preventable paid healthcare dollars were from the population that was classified as healthy or with a minor chronic condition (Boyd et al., 2008) In addition, there are large organizational barriers to the implementation of PCMHs on top of the general lack of standards. Several papers have raised concerns about the costs (monetary and organizational) to physician practices. Some of the concerns that have been raised have been physician compensation models, implementation of expensive information technology systems, and general organizational restructure of practices (Chandran et al., 2009; Fuller, Clinton, Goldfield, & Kelly, 2010) As a result of these concerns, it is important for research to show that the costs associated with implementing Patient Centered Medical H omes are worth the benefits
15 when applied to the population as a whole. As mentioned above, the implementation of all six characteristics of a Patient Centered Medical Home can have significant tangible (IT systems) and intangible costs (practice re organi zation). It is not expected that healthy populations would utilize the same intensity of services compared to a chronically ill population. The cost of a full PCMH meeting all six criteria versus one that has some components of a PCMH applied to a healt hy population should be questioned. Answering this question may point to a scaled down implementation of Patient Centered Medical Homes for healthy populations that may only have one or two characteristics implemented. This study proposes to address this deficiency by evaluating the impact of Patient Centered Medical Homes on healthy populations including assessing the impact of each of the separate components of PCMHs The specific aims of this research topic are as follows: 1. Determine if there are differences in costs and utilization of health care services of healthy individuals in PCMHs versus healthy individuals that are not in PCMHs. 2. Among the six defining characteristics of a PCMH, determine the individual effect of each of these characteristics on costs, outcomes and utilization of healthcare services. This study will utilize data from the Household Component section of the Medical Expenditure Panel Survey ( MEPS) from the years 2004 2007. MEPS is a national survey that is designed to provide national ly representative estimates on the health care use, expenditures, insurance coverage and payment of the noninstitutionalized US population. Chapter 2 will provide background information on Patient Centered Medical Homes and their development and impact on health care expenditures. Chapter 3 will provide the theoretical and conceptual framework that will guide this study. Chapter 4
16 will present the data elements incorporated in the study and the statistical methods that will be used. Chapter 5 will present the results of the analysis. Chapter 6 provides a discussion of the results with potential policy implications as well as potential future studies.
17 CHAPTER 2 BACKGROUND History and Structure of Patient Centered Medical Homes The concept of Patient Centered Medical Homes (PCMH) was introduced in 1967 by the American Academy of Pediatrics (AAP) Council on Pediatric Practice. The AAP at that time advocated for a medical home in a very narrow definition as a central source of medical records for children with special health care needs. (Sia, et al., 2004) The AAP only recommended this as a guideline for pediatric practices and not an official policy position for the Academy. The fir st formal incorporation of medical homes as a policy statement from the AAP came in 1977 under the policy Fragmentation of Health Care Services for Children. It was again reaffirmed in 1979 under the policy statement Children Having Care from Multiple Sources. Both statements still maintained the narrow definition of medical homes as a central repository for medical records for children with special health care needs. Throughout the 1980s, various states continued to work on defining the medical home from a policy standpoint with an evolving focus on total family and child health in the community setting. In 1992 the AAP published their first policy statement to define the medical home. The statement read: The AAP believes that the medical care of infants, children and adolescents ideally should be accessible, continuous, and compassionate. It should be delivered or directed by well trained physicians who are able to manage or facilitate essentially all aspects of pediatric care. The physician should be known to the family and should be able to develop a relationship of mutual responsibility and trust with the family. During the mid1990s and into the 2000s, the medical home concept began to be expanded to focus not only on children but also other at risk populations such as those with chronic illnesses. (Sia, et al., 2004) Also during this time, the structure of PCMHs
18 began to be translated from a conceptual ideal to more tangible organizational elements that made a particular provider organization and PCMH (ncqa, 2008) Initial research on organizations that incorporated medical home concepts showed reduced referrals and improved patient satisfaction(Starfield & Shi, 2004) These promising results attracted interest from employers that were beginning to experience large increases in healthcare coverage costs for their employees. In 2005 several large employers lead by IBM actively sought to collaborate with sever al primary care organizations in an attempt to address the burgeoning healthcare costs on employers (Hall & Kruse, 2010) These efforts led to the 2007 joint statement of the American Academy of Family Physicians, American Academy of Pe diatrics, American College of Physicians, and the American Osteopathic Association on Patient Centered Medical Homes ("Joint principles of the Patient Centered Medical Home," 2008) The joint statement highlighted seven principles of PCMHs: 1. Personal physician relationships; 2. Physician directed medical practice of a team of healthcare professionals; 3. Whole person orientation; 4. Care coordination; 5. Quality and safety of medical practice; 6. Enhanced access to care; 7. Revised payment mechanisms to reflect this new concept. The Patient Centered Primary Care Collaborative (PCPCC) was created shortly thereafter which established a national advocacy and lobbying organization for PCMHs. In 2008 the NCQA issued their standards and guidelines for Patient Centered Medical homes. Similar to their accreditation process for health care facilities, the process allows providers to be recognized as a PCMH. The NCQA working from the joint statement scores providers on 9 domains: Access and communication; Patient tracking and registry function; Care management; Patie nt self -
19 management support; Electronic prescribing; Test tracking; Referral tracking; Performance reporting and improvement; and Advanced electronic communications (ncqa, 2008) CMS continued to build on this standard as part of its 2010 Pilot program for PCMHs. The CMS standard built up the same 9 domains as the NCQA version, but removed or added some elements that ar e measured within those domains (Assurance, 2008b) At about the same time, the Commonwealth Fund conducted a national Health Care Quality survey in 2006 to measure, among other things, the existence of a Patient Centered Medical Home for survey respondents. The survey identified PCMHs using only four criteria: The existence of a regular provider or place of care; no difficulty in contacting provider by phone; no difficulty in getting advice or care on weekends or evenings; and office visits are organized and on time (AC Beal, 2007) It is interesting to note that this definition differs from the CMS and NCQA versions in not only its simplicity, but also the lack of incorporation of information technology which are heavily weighted in both the NCQA and CMS versions. Finally, an additional measure or definition of Patient Centered Medical Homes that was also recently developed in 2004 was the Medical Home Index (Cooley, Mc Allister, Sherrieb, & Clark, 2003) Similar to the NCQA tool, the Medical Home Index (MH I) is broken into 25 themes that are grouped into 6 domains of practice activity. It is reasonable to believe that a consensus definition of a Patient Centered Medical Home will be difficult to attain, especially if policy makers are unsuccessful in changing reimbursement mechanisms to account for the existence of a PCMH. Due to limitation of the data collected in MEPS, this study will use a definition closer to the Commonwealth fund criteria.
20 Effects of Patient Centered Medical Homes Patient Centered Medical Homes were initially developed to address concerns related to the care of children with special health care needs. CSHCN have been shown to have many unmet needs and t he most severely affected of these children have complex chronic conditions that require frequent visits to multiple providers of different specialties (CMS, 2007; Cooley, et al., 2003; Farmer, Marien, Clark, Sherman, & Selva, 2004; Hall & Kruse, 2010; Homer, et al., 2008; Kaye, 2009; Pless, Satterwhite, & Van Vechten, 1978) A natural growth of this concept is to include individuals affected by any chronic condition regardless of age, since these individuals would also experience frequent visits to multiple providers of different specialties. The intent of the medical home in this scenario is to provide oversight and coordination of these visits in such a manner that is effective and sensitive to the needs of the patient and family. The ideal result of providing care in this manner is to ensure the right services are utilized at the right time and the services that are utilized are desired and adhered to by the patient and family. It has been shown that improved patient compliance in treatment regimens and careful coordination of health care services has improved outcomes and reduced expenditures for patients. Outcomes of PCMH populations As outlined above, there are several key components related to a Patient Centered Medical Home, the potential strength in a Patient Centered Medical Home is the ability to collectively apply these individual components in a coherent fashion. These components can be highlighted as follows: The coordination of care for patients that visit multiple providers; the ease of access to provi ders for patients either during weekends or after hours; the implementation of treatment regimes that are culturally
21 sensitive and developed with input from the patient and/or family Many of these components have been previously evaluated as to their impact on patient outcomes individually Limited studies exist evaluating the collective impact of these components on patient outcomes. In addition to the limited implementation of Patient Centered Medical Homes, the lack of a consensus definition of what makes a Patient Centered Medical Home has meant that many studies have evaluated different aspects of a PCMH versus the combined effect of all components In the past several years, results of several significant pilot programs have been evaluated and reported on. All pilot programs have shown to have positive effects on some outcomes that are measured in the particular study. The different programs have incorporated different aspects of the Patient Centered Medical Home. Steiner provided a synopsis o f the Community Care program of North Carolina. (Jeffe, Whelan, & Andriole, 2010) The Community Care of North Carolina (CCNC) program was not developed as a Patient Centered Medical Home pilot, but did incorporate PC MH ideals in its development. CCNC formed as a partnership in 2000 between the state Medicaid program, primary care physicians and other providers. The formed network is comprised of approximately 1,200 primary care physicians and 750,000 Medicaid members. CCNC has several PCMH characteristics, although they were only described in t his synopsis. The characteristics described were the assignment of members to primary care physicians, the coordination of care to other providers, 24 hour oncall coverage, and continuous qual ity improvement projects. Steiner reports positive effects on disease management, particularly asthma control. Individuals with asthma saw an increase of 112% from baseline for the inoculation of
22 influenza. Children with asthma saw an 8% decrease in emergency department visits and the hospital admission rates decreased by 34% for these same children within the CCNC. Paulus described highlights of a Patient Centered Medical Home pilot as part of an overall review of innovations within the Geisinger Health Care System (Paulus, Davis, & Steele, 2008) McCarthy provided a similar review with updated data (McCarthy). The pilot program began in 2006 and with 2 sites within the integrated delivery system. The goal of the pilot was to improve care coordination and health status for the enrollees in the Geisinger Health Plan. The distinct aspect of this pilot program is the utilization of health care technology in the PCMH. The pilot, called Personal Health Navigator provides access to the electronic medical record through a web portal for both providers and patients. Paulus reported t he first year results (2006 2007) from the pilot program showed a 20% drop in all admissions and a 7% reduction in total medical costs. After these promising results, the program was expanded to include 25,000 Medicare beneficiaries who receive care at 21 sites. McCarthy reports a 5% reduction in the hospital readmission rate for a subset of Medicare members from 2007 2008 versus a 1% increase among a control group of Medicare enrollees who did not participate in the pilot program. Although the Group Health Cooperative officially implemented their pilot program in 2006, their practice redesign initiative has been ongoing since 2000 and has been widely reported on (Conrad et al., 2008; McCarthy, 2009; Ralst on et al., 2007; Reid et al., 2010; Reid et al., 2009) The pilot program focused on one clinic within the health system that serviced approximately 9,200 adult patients. A collaborative workgroup of
23 physicians, managers and researchers identified the practice changes for the clinic that would incorporate PCHMs. The use of electronic medical records, web and virtual communication between patients and providers, patient outreach, care team meetings to review care, and standardization of management practices were the components identified for implementation in the pilot program. Recently, Reid et al. reported results evaluating two years of the pilot programs implementation(Reid, et al., 2010) The study compared results from the pilot versus controls at 4 unique time periods: baseline; twelve months; twenty one months; and twenty four months. The variables evaluated were patient experience, staff burnout, clinical quality and utilization and costs. Patient experiences were better than controls although the magnitude decreased at the tw enty four month interval. A reduction in staff burnout was consistent across all time periods. Clinical quality was measured by aggregating HEDIS into four composites. The pilot clinic scored better across all four composites at all time periods versus the controls, although similar to patient satisfaction, the magnitude appeared to decrease over time. Homer et al. provides an extensive literature review of Patient Centered Medical Homes and their impact on Children with Special Health Care Needs (Homer, et al., 2008) Utilizing a systematic approach, the authors identified and reviewed 33 articles on 30 unique studies. The majority of the studies were a crosssectional study design with only six utilizing a randomized control trial format. Almost all of the studies were focused on a specific population or disease. Homer defined eight medical home activities: Care coordination with communi ty; care coordination with subspecialty care; population monitoring; physical and operational modification; clinical care; cultural competency; and connection with PCP provider practice. Consistent with the varied
24 definition of a Patient Centered Medical Home, the majority of studies evaluated only one activity of a medical home, with only nine studies evaluating more than four activities. The impact of Patient Centered Medical Homes on outcomes were reviewed and split between long term and short term out comes. The short term outcomes that were evaluated were safety; effectiveness; efficiency; family centeredness; timeliness; and equity. Long term outcomes that were reviewed were health/functional status; developmental outcome; family function; and cost. In general, all the articles reviewed showed that CSHCN in a medical home had better outcomes from children that were not in a medical home. In review, there are several challenges or limitations when discussing research on outcomes involving Patient Centered Medical Homes. While all the studies and pilot programs on PCMH appear to show a positive effect on patient outcomes, they highlight the difficulty and variety of defining what is a medical home. Some pilot programs have focused on utilizing a primary care physician with a care coordinator, others have focused on the utilization of electronic medical records or web portals. The varied nature of what researchers call a Patient Centered Medical Home limits the ability to generalize the results and their potential impacts to the overall population. Nonetheless, the research indicates that the existence of some type of medical home or charac teristics of a medical home has positive effects on all measures of patient outcomes. Expenditures of PCMH populations The analysis of the effects of Patient Centered Medical Homes and their impact on overall expenditures is limited. Several of the pilot programs have reported reduced expenditures driven through a reduction of hospitalizations and emergency department
25 visits (Homer, et al., 2008) The cost of implementing a Patient Centered Medical Home is a particular gap in the literature. Pilot programs implemented to date have been implemented by mainly leveraging existing structures, such as electronic medical records, to create a medical home or aspects of a medical home. Again, the lack of a consistent definition of what makes a Patient Centered Medical Home and the sliding scale in which an organization can be a medical home contr ibutes to this problem. Fields meta analysis of seven medical home models showed reduced hospitalizations in all seven models, reduced emergency room visits in five of the seven models, and per patient savings in six of the seven models ( Fields, Leshen, & Patel, 2010) The analysis attempted to identify seven successful medical home models across the country to identify common features. The seven models showed a wide variability in implementation setting and patient population focus. All the medical home projects reported significant cost reductions or improvement in quality. Savings reported in this analysis on a per patient basis ranged from $71 to $640. The wide range of these reported cost savings is most likely due to the vari ety of medical home implementations. Some models focused solely on a chronic condition, some on pediatric patient populations and some on a Medicaid population. Regardless of the extent of medical home implementation, when costs are reported, they have shown a reduction. Unfortunately, the high administrative costs of implementation may overwhelm any cost savings acheived. Several characteristics of a medical home such as care coordination, have shown to be effective in reducing expenditures and may be the primary driver of these cost savings. Future research
26 would be well served at evaluating the costs of implementing the full practice transformation that a Patient Centered Medical Home requires. Barriers to Implementation of Patient Centered Medical Homes The focus on costs related to the implementation of medical homes is important when discussing barriers to implementation of the medical home. Looking at the Patient Centered Medical Home concept, there are several organizational and policy level barriers that will need to be address to allow widespread implementation. A major component of Patient Centered Medical Homes is the coordination of specialty care and management of conditions. Both of these activities imply the implementation or utiliz ation of electronic medical records, an organizational barrier to implementation for many primary care practices. Another central tenet of a Patient Centered Medical H ome is the care team that is led by a primary care physician. The trends in medical education have been in the opposite direction of this requirement. Despite the fact that many have called for a focus of medical training on primary care, medical students increasingly choose specialty training (Jeffe, et al., 2010) The ability to alter this trend in medical training is a significant policy barrier that will need to be addressed for the implementation of medical homes. Any program that requires practices to alter their methodology will likely need to provide some type of incentive to stimulate this change. Providers will need to be recognized for undertaking this endeavor and compensation should be adjusted accordingly. A significant policy barrier will be the alteration of entrenched reimbur sement models to compensate providers based on the utilization of a Patient Centered Medical Home concept. The challenges to widespread adoption of the Patient Centered Medical Home concept are significant and will need to be addressed in order to ensure its future success.
27 Berenson highlighted several barriers to implementation of medical homes (Berenson et al., 2008) In discussing practice size and scope, Berenson raised the issue of electronic medical records. The per patient costs of implementing an electronic medical record can be particularly high. The high costs act as an economic barrier to entry for small physician practices and therefore m ake it more likely that only large medical practices would have the economies of scale to undertake its implementation. In fact, all pilot programs cited that utilize or highlight electronic medical records have been large integrated delivery systems. Ri ttenhouse also raised the issue of electronic medical records in the context of practice redesign(Rittenhouse & Shortell, 2009) Incorporating an electronic medical record in a medical practice will require business process changes such a s patient check in, medical record collection and billing. Combined with the high cost of implementation, many small or medium size practices would be challenged to change their office practice patterns Perhaps the most significant challenge to successful adoption of Patient Centered Medical Homes is the disturbing trends in graduate medical education and primary care. Primary care providers and patients are the two parties at the center of the medical home. Brotherton reviewed characteristics of physic ians in primary care training over a 10 year period from 1995 to 2005 to identify trends (Brotherton, Rockey, & Etzel, 2005) The results from this analysis showed that the number of family medicine residents fell from 8 232 in 1998 1999 to 4, 848 in 2004 2005. Pugno reviewed results from the 2007 National Resident Matching Program (Pugno, McGaha, Schmittli ng, DeVilbiss, & Kahn, 2007) Fewer US medical senior students chose family medicine through the match program in 2007, although looking at trends, at appears that the decline in
28 interest in family medicine training has stabilized. The low level of pr imary care physician trainees presents a particular policy challenge. Despite several policy initiatives to provide incentives to medical school graduates, such as loan repayment, graduates still appear to favor specialty training over primary care traini ng. The lack of primary care physicians and providers will severely limit implementation of Patient Centered Medical Homes. A final barrier to implementation of Patient Centered Medical Homes is the alteration of reimbursement models. The reimbursement mechanisms will have a significant impact on the other barriers. The level of reimbursement will affect providers ability to recover costs of implementation of electronic medical records. Reimbursement levels will also affect the attractiveness of primary care fields and recruitment of medical school graduates into the primary care field. Primary care providers are challenged with payment systems that reward providers based on patients seen versus the care management. New rei mbursement mechanisms need to be developed and implemented that recognize the unique structure of Patient Centered Medical Homes. Both Berenson and Rittenhouse highlighted these fixed reimbursement models as a significant barrier to implementation. The cost of implementation of electronic medical records, training of primary care providers and reimbursement m odels of primary care providers are some of the significant barriers to implementation of Patient Centered Medical Homes. These are significant bar riers that will present significant costs to be addressed if Patient Centered Medical Homes are going to be a successful policy.
29 Policy Implications Patient Centered Medical Homes are promising alternatives in the US healthcare debate but need more research US policy makers continue to look for possible solutions to control healthcare expenditures while maintaining quality. Patient Centered Medical Homes have shown effectiveness controlling and reducing costs for chronically ill populations. They have also been shown to improve outcomes for the patient populations that they serve. Unfortunately, these pilot programs have been limited to patient specific populations with limited review of their effectiveness on broader subsets of populations. Policy m akers are also confronted with the confusing dilemma of defining what constitutes a Patient Centered Medical Home. There are several definitions of Patient Centered Medical Homes, from the very specific like the NCQA definition, to very broad like the Com monwealth Fund definition. Research still needs to facilitate policy development in determining the most effective components of a Patient Centered Medical Home. Given the questions outlined above regarding Patient Centered Medical Homes, this study will contribute in addressing these concerns. The effect of Patient Centered Medical Homes on the general population, of which a large portion of is considered healthy, is important to answer. If Patient Centered Medical homes are only marginally effective on healthy populations, the policy approach might be altered. Policy makers may elect to narrowly focus the utilization of Patient Centered Medical Homes on chronically ill populations versus as a broader solution to rising healthcare costs. Finally, give n the varied definition of Patient Centered Medical Homes, determining the most effective characteristics of medical homes will help policy makers. Policy makers could choose to incorporate these characteristics with other proposed solutions versus complete Patient Centered Medical Home implementations.
30 Conceptual Model The conceptual model used for this study will be based upon Grossmans model of human capital and the demand for health (Grossman, 1972) combined with Donabedians Structure Proce ss Outcome model (Donabedian, 1966) to determine health quality. Grossmans model provides the framework to evaluate the motivating factors for healthy individuals to obtain health care services. Donabedians model provides the framework to evaluate the effects of Patient Centered Medical Hom es on outcomes and expenditures. Combined, these models help explain that although Patient Centered Medical Homes might increase the efficiency of the production of healthy days, the impact of this increase in efficiency will be minimized for healthy indi viduals. Grossman provides an empirical model that helps explain the concept of an individuals demand for health. As an overview health is viewed as a durable capital stock that yields an output of healthy time. Similar to other durable capital stocks like buildings or equipment, health stock depreciates with age and can be increased with investments into this stock. Health versus health stock, in this model is defined to indicate illness free days in a given year and is both demanded and produced by individuals. Health is demanded by consumers because it provides utility in the form of satisfaction of a healthy life and illness free days that provide an indi vidual the ability to invest in other activities, specifically income generation. At the simplest level, an individual will seek not only health goods, such as visits to a doctor or medical procedures, but will also seek the production of health commodities, such as healthy eating or exercise. Individuals seek this for several reasons. First it helps reduce the impact of depreciation of their stock of health. Second, the increase or maintenance of
31 health stock provides or maintains healthy days which an individual can use to earn more income by working more or have more time to engage in other activities that the individual may enjoy and provide utility to them. To provide a more thorough explanation of the Grossman model and its application to this study, we will turn our attention to its empirical representation. The utility funct ion of an in dividual can be represented by Equation 21. = ( , ) (2 1) Equation 2 1 shows that an individuals intertemporal utility can be viewed as a function of the consumption of health services, and the consumption o f all other commodities, The consumption of health services depends on the individuals health stock and some usage rate, which produces healthy days In addition, t he relationship of health stock and depreciation of health stock can be expres sed in Equation 22 Ht + 1Ht= IttHt (2 2 ) In E quation 2 2 the difference between the health stock in the future time period and the current time period is equal to the gross investment of health during the current time period, It less the depreciation rate, t, times the health stock in the current period. Given this representation, we can see that people can negate or lessen the impact of depreciation on their future health stock by investing in their health in a given time peri od. In this manner, an individual can theoretically increase their longevity by keeping their health stock above the minimum health stock. The gross investment in health in a given time period can be represented as a set of household production functions as shown in the Equation 23
32 It = It( Mt, THt; E ) ( 2 3 ) In E quation 2 3 Mt represents medical care THt represents the time component of health investment production. E represents an exogenous variable of an individuals stock of human cap ital. Total human capital, in particular education level, indirectly influences the gross health investment by affecting how efficient the other direct inputs operate. Mt deserves additional discussion, as this function represents the premise that the investmen t in health, while typically associated with the input of medical care, is associated with a variety of inputs. Other inputs or goods that can affect health include housing, diet, exercise, smoking, alcohol, etc. In his discussion of t his model, Grossman highlights this fact and outlines that the model would retain its structure even if the primary health input associated with Mt was something other than medical care. In order to determine optimal investments and health capital amounts Grossman introduces several additional constraining equations. An individuals time constraint is the most obvious one, in that there is a definitive boundary of days that are available in a particular time period, generally one year or 365 days. This time constr aint eq uation can be represented in Equation 24. + + + = (2 4) The total time available to an individual, is equal to the total time spent working, total time spent on health activities, total time lost due to illness, and the total time spent on other activities Related to the time constraint E quation 2 4 is the goods budget equation which allows us to represent the wealth that an individual can generate with the time that they have available. This is represe nted in Equation 25. ( ) ( ) = ( ) + = (2 5)
33 Equation 25 shows that an individual cannot spend more on the purchase of medical goods, other market goods and non market production activities than the earnings that they can generate if they worked all the time combined with their initial inherited capital. Solving for first order optimality conditions and rearranging equations, we derive E quation 2 6 that represents the relationship of the optimal stock of health capital to the supply costs of capital. + + ( 1 + ) = ( ( ) + ) (2 6) represents the marginal product of the stock of health in the production of healthy days, or in other words, how much you increase your healthy days if you increase your next stock of health by one. represents the wage rate. represents the marginal utility of healthy days and represents the marginal utility of wealth. The interest rate is represented by and represents the marginal cost of gross investment in that particular period. If we divide by the marginal cost of gross investment in period 1 then we can determine the marginal monetary rate of return on an investment in health shown by Equation 27 + = (2 7) = ( ) and represents the psychic rate of return on the investment in health. Evaluating E quation 2 7 we can see that as we increase the stock of health capital, the marginal product of the stock of health in the production of healthy days will continue to decrease. It is easy to assume this because as discussed prior, there is a finite bound in the nu mber of days available, so eventually, an individual will not be able to produce any more healthy days. Therefore, as this continues to decrease, the return on investment continues to decrease. Figure 21 represents this relationship graphically
34 along wi th the supply curve which shows an individuals optimal amount of stock of health given the cost of capital. Now that we can represent an individuals demand for health we can discuss how an individuals demand for health might be affected within a Patient Centered Medical Home It is in this context in which we introduce Donabedians model. This model was originally developed as an approach for quality assurance in the delivery of medical care. The model is made up of three distinct approaches that are related in a sequential order. Taken together, the three approaches provide an assessment of the quality of medical care delivered. The three approaches identified by Donabedian are Structure, Process, and Outcome (SPO) Structure is meant to designate the conditions under which care is provided. Process relates to the acti vities that make up health care. Process activities can include the activities by families and individuals that might affect health in addition to traditional medical procedures Th e third approach, Outcomes, refer to the changes that are attributed to the care delivered. Under this model, there is an interactive component between the three approaches such that elements of structure influence elements of process which in turn influences outcome s. It is helpful to use the SPO model to stratify how PCMHs might affect the production of health capital of an individual. Specifically, individuals within PCMHs will be more efficient producers of health capital. Structural characteristics of PCMHs include the use of electronic medical records (EMR), the mix of physicians and physician extenders and the use of disease management databases. Process characteristics include the use of email and other forms of communi cation to increase and facilitate access to providers of care. The tracking of and dissemination of patient
35 specialt y visits is a key process for PCMHs. Outcomes and the feedback loop into the other approaches is the final critical component of PCMHs. A nother critical component of PCMHs is to track outcomes of interventions and utilize this tracking for continuous improvement. It is clear that i ndividuals that are within PCMHs will be aff ected by all these approaches in the production of health capital versus indi viduals that are not in PCMHs. Using these two models, we can describe the effect s of Patient Centered Medical Homes on healthy individuals and the utilization of health care services. Healthy and ill individuals will seek to maintain or improv e their health stock through a variety of means. One can view this as investing in their current health so that they might extend their health and hence their longevity. Referring to the Grossman model, these investments can be activities such as healthy eating, exercising, physician services, etc. Individuals seek to increase their stock of health because it allows them to produce healthy time that leads to additional utility through income generation or other activities. Referring to Donabedians model the effectiveness of the production of healthy time will be influenced by Patient Centered Medical Homes. The StructureProcessOutcomes approaches used in organizing Patient Centered Medical Homes will lead to a more efficient production of healthy tim e. The more efficient production of healthy time will in turn cause an individuals health productio n curve to be shifted. Figure 22 shows that this shift will cause a decrease in the amount of stock of health demanded since a lower stock of health capi tal will produce the same amount of health days. When specifically evaluating a health individual, the number of healthy days produced before the shift of the marginal production curve is already high. We know that the marginal production of
36 health days decreases at an increasing rate the closer you get to the upper bound of total health days available. Therefore even though Patient Centered Medical Homes will shift the production curve, its impact on a healthy individual would be expected to be minimal since they are already at the top of the production curve in regards to health days produced.
37 Figure 21. Relationship of Stock of Health Capital and Cost of Capital ( ) + H S H*
38 A B Figure 22 Interaction of Grossmans Health Capital and Donabedians StructureProcessOutcomes with Patient Centered Medical Homes ; A) Health stock over time; B) Investment in Healt h Stock, Influence of StructureProcessOutcomes. Patient Centered Medical Home H t Individual Health Stock H t +1 Individual Health Stock Investment in Health Stock I t Human Capital Other Direct Inputs Structure Process Outcome Non PCMH PCMH H 365
39 CHA PTER 3 METHODS Overview Research Questions and Hypothesis The purpose of this study is to evaluate the effects of Patient Centered Medical Homes on healthy populations. To review, t he general approach of the study will be to identify characteristics of Patient Centered Medical Homes and healthy individuals that can be classified in Patient Centered Medical Homes Once these individuals are classified in these categories, the study w ill evaluate: 1. The effect of Patient Centered Medical Homes on total expenditures 2. The effect of Patient Centered Medical Homes on utilization The study used longitudinal data from a national expenditure survey from the years 2005 to 2009. In addition, t he study attempt ed to distinguish between characteristics of Patient Centered Medical Homes and their individual effects on claims expenditures. Hypothesis 1 Patient Centered Medical Homes will not have a significant effect on expenditures due to their pos ition on the marginal product health curve. Since a healthy individual sits high on the marginal product curve, any additional investment in health will produce minimal returns on addi tional health capital. Therefore healthy individuals will not make sig nificant investments in the form of additional expenditures in their health capital. Hypothesis 2 Patient Centered Medical Homes will not have a significant effect on utilization of health care services for healthy individuals due t o the same reasons outli ned in H ypothesis 1.
40 Hypothesis 2 a At an aggregate level, total utilization of health care services between healthy individuals in Patient Centered Medical Homes and healthy individuals that are not in Patient Centered Medical Homes are expected to be sim ilar. Variations between these two groups are expected within the types of utilization. Healthy individuals in Patient Centered Medical Homes are less likely to have specialty office visits due to the increased efficiency of Medical Homes which would al leviate their desire to seek specialty services. This is in contrast to individuals that are not in Patient Centered Medical Homes. Data Source The study relied on data from the Medical Expenditure Panel Survey or MEPS. MEPS is a large national survey of individuals, families, medical providers and other key health care entities that began in 1996 on a mandate from the Agency for Healthcare Research and Quality (AHRQ). The survey is comprised of three related surveys cal led components. The household component surveys households from a subsample of the annual National Health Interview Survey and is designed to provide estimates on health care utilization, expenditures, payments and sources of care. The provider component is a survey of providers and facilities that provided care to members of the household component survey. It is intended solely to augment the household component responses on health care utilization and expenditures and impute values for survey nonresponse in the household component The final component is the insurance component. The insurance component is a mail survey of approximately 30,000 businesses with the objective to provide estimates on costs of employee sponsored health insurance. The hous ehold survey is an overlapping panel design, which
41 conducts a computer assisted interview 5 times over 30 months to collect expenditure and utilization data for a 2 year period. In 2008 the household survey had a sample of 12,316 families representing 31, 262 individuals. Of particular interest in this study, the household survey incorporates elements of the Consumer Assessment of Health Plans Study (CAHPS) and the Medical Outcomes Study, short form or SF 12. The CAHPS elements are intended to measure quality of care and patient satisfaction and will be used in this study to help identify the existence of Patient Centered Medical Homes. The SF 12 is intended to measure health status and will facilitate the identification of healthy individuals for the pur pose of this study. Finally, MEPS oversamples policy relevant groups of Hispanics, Blacks, Asians, persons with disabilities and low income households. Outcome Measures Total Expenditures The primary dependen t variable in this study is total health care expenditures for healthy individuals as indicated in the MEPS file. These expenditures were evaluated from the years 2005 to 2009, the latest av ailable data file in MEPS. Total expenditures for healthy individuals were compared between individuals in Patient Centered Medical Homes and individuals that are not in Patient Centered Medical Homes. This comparison will help evaluate potential monetary impacts of Patient Centered Medical Homes which would help answer if the costs of implementation are worth the benefits gained, in this case, reduced claim expenditures. Utilization of Health Care Services A seco ndary dependent variable was overall outpatient visits for healthy individuals. Furthermore, the study segment ed those outpatient visits between primary
42 care and specialty care visits. The evaluation of this variable will help determine if the existence of Patient Centered Medical Homes help reduce the number of outpatient visits for healthy individuals Primary Explanatory Variables There are several explanatory variables that are unique to this study that are used to identify healthy individuals in Patient Centered Medical Homes Since this study is focused on the effects of Patient Centered Medical Homes, the first primary explanatory va riable is a categorical variable indicating the existence of a Patient Centered Medical Home. The second primary explanatory variable is another categorical variable indicating a healthy individual. Patient Centered Medical Homes One of the biggest challenges facing researchers in dealing with Patient Centered Medical Homes is the lack of an accepted definition of what makes a Patient Centered Medical Home. We discussed earlier the various entities that have presented their criteria, and highlighted some of their similarities and differences. This study is presented with the same dilemma. The MEPS does not specifically survey for the existence of Patient Centered Medical Homes which presents additional challenges. In order to address this challenge, the study will conduct a sensitivity analysis on two different methods to classify a Patient Centered Medical Home. The first method will use a comprehensive index of multiple variables in MEPS developed by Thompson in her research on Patient Centered Medi cal Homes Using the principles from the Joint Statement on Pat ient Centered Medical Homes, we use d five categories to group multiple responses from MEPS. The categories that are used are Comprehensive Care, Continuity of Care, Enhanced Access, Coordinat ion of Care,
43 and Quality and Safety in care. The responses will be tabulated and an average score will be calculated to create an index indicating the medical homeness that exists for the respondent. Similar to the NCQA Medical Home Certification (Assurance, 2008a) and the work done by Cooley (Cooley, et al., 2003) the creation of an index allows us to account for varying practice patterns, in which some areas of a particular practice may be str onger than others. For example, some practices might be more efficient at Care Coordination while limited in Quality and Safety. The average score of greater than 25% will indicate the existence of a Medical Home. Bethell (Bethell, Read, & Brockwood, 2004) showed the use of a continuum score to be valuable in determining the existence of a Medical Home using MEPS data. The second method that was used was a simplified definition of the existence of a Medical Home. A minimal number of identified survey response elements that are directly related to a Patient Centered Medical home would indicate the presence of a Patient Centered Medical Home. For the purpose of this study, only the positive response o f all these survey elements classified an individual in a Patient Centered Medical Home. These elements for both the comprehensive and simplified method are outlined as follows: Patient Centered Medical Homes Comprehensive Definition Enhanced a ccess Usual Source of Care (USC) A yes response is scored as one point. Go to USC for preventative care A yes response is scored as one point Go to USC for referrals A yes response is scored as one point
44 How difficult to contact USC by phone A response of not at all difficult or not too difficult is scor ed as one point. How difficult to receive after hours advice A response of not at all difficult or not too difficult is scored as one point. USC has night and weekend office hours A yes response is scored as one point. How often do you get an appointment when needed A response of Always or Usually is scored as one point. Continuity of care Continuity of care w as scored based upon a ratio of ambulatory visits. The number of ambulatory visits that an individual makes to their usual source of care provider will be compared to the total number of ambulatory visits that the individual makes. The higher the ratio of usual source of care ambulatory visits to total ambulatory visits, the closer that we can postulate that the individual has continui ty of care through their usual source of care. Any ratio greater than 50% was scored as one point. Comprehensiveness of care How much a problem getting needed medical treatment A response of not a problem is scored as one point How much of a problem seeing a specialist A response of not a problem is scored as one point. How much of a problem getting prescription medication A response of not a problem is scored as one point. Coordination of care The usual source of care provides referrals A r esponse of yes is scored as one point. The usual source of care provider asks about other treatments A response of yes is scored as one point.
45 Quality and safety of care How often does the usual source of care ask your help in making medical decisions A response of always or usually is scored as one point. Does the usual source of care give you control over treatment A response of always or usually is scored as one point. Does the usual source of care have respect for what you say A resp onse of always or usually is scored as one point. Does the usual source of care listen to you A response of always or usually is scored as one point. Does the usual source of care spend enough time with you A response of always or usually i s scored as one point. Does the usual source of care explains treatment that is understood A response of always or usually is scored as one point. Does the usual source of care speak the same language or provides interpreter A response of yes is scored as one point. Patient Centered Medical Homes Simplified Definition Office h ours A provider that has office hours on nights and weekends is indicative of a Patient Centered Medical Home. This characteristic falls under the broad principal of enhanced access to care. After hours c ontact An additional variable that is consistent with the existence of a Patient Centered Medical Home is the ease at which an individual can contact a provider after normal business hours. Individuals that can contact a provider with ease after normal business hours have an enhanced access to care, a core principle of Patient Centered Medical Homes.
46 Shared decision m aking Another survey element that can be used to indicate the presence of a Patient Centered Medical Home is a respondent who is usually or always involved in the medical decision process. Providers who take a whole person orientation in providing care are practicing one component of a Patient Centered Medical Home. This principle advocates not only coordinating care across multiple specialties, if needed, but also making care decisions that are culturally appropriate and sensitive to the individual. Therefore, it can be inferred that individuals that share decision making responsibilities with a provider may be in a Patient Centered Medical Home. Usual source of c are The most consistent principle cited by all the various definitions of Patient Centered Medical Homes is having a usual source of care for the individual. The MEPS questionnaire captures respondents who indicate that they have a usual source of care. The study include d individuals that have a usual source of care that is not a facility or emergency room. Language The final criteria that will be included in determining if an individual exists in a Patient Centered Medical Home is the if the respondent indic ates that the provider speaks the same language or provides an interpreter for the individual. The positive response to this criterion would be consistent with the whole person orientation that is required of a Patient Centered Medical Home. Medical deci sions would be evaluated by the individual with full understanding of any treatment in which they may or may not be entering into. In addition, it would be reasonable to expect any cultural desires would also be taken into account in any treatment plan.
47 H ealthy Individuals The identification of healthy individuals within the Medical Expenditure Panel Survey is an additional challenge. A s part of ARHQs continuing effort to measure the nations health care quality and changes to overall health status as a result of ongoing programs, the complete set of SF 12 questions have been included in the MEPS. The SF 12 is a widely validated and accepted measure of health status and is a subset of the larger more comprehensive SF 36. The SF 12 is a short self admin istered, 12 question survey that measures both mental and physical health status. These are given as a 0 100 scale score in the Physical Component Summary (PCS) and Mental Component Summary (MCS) respectively with a higher score indicating a healthier status The study wil l only focus on the physical health status. The identification of healthy people using the SF 12 or SF 36 has been done by Harman et al. (Harman et al., 2010) in their study of health plans healthcare effectiveness with outcomes of enrollees with Diabetes. In their study, they attempted to predict the probability of an indi vidual being healthy at follow up, where healthy was classified as being in the top 75th percentile on PCS or MCS. The study used the same approach and classified individuals as healthy that are in the top 75th percentile on PCS score. Juxtaposed against this definition of healthy individuals is an alternative definition of healthy individuals using the survey respondents self reported health status. A sensitivity analysis was conducted against the two methods of classification of health status to determ ine if there is an effect on the analysis depending upon which definition is incorporated.
48 Control V ariables Healthcare expenditures and utilization are influenced by many variables. Some of these variables are observed and known to have a positive or negative effect. When these variables are known and included in the data set, it is important that we recognize and account for them. Controlling for these variables will help isolate the true effect of Patient Centered Medical Homes on healthy populations. Several of these variables are discussed below: Age Age is known to influence both expenditures and utilization of healthcare services. Age was stratified into 10 year bands. Gender G ender is also a known variable to influence healthcare expenditures and utilization. Marital Status An individuals marital status has the ability to influence healthcare expenditures and utilization. A married individual is more likely to seek care for services than a non married individual. Location. The location of an individual will influenc e access to health care and was segmented between urban and rural. Race. Race plays an important part in an individuals perceived need for health care as well as utilization of health care services. Race was segmen ted between the categories presented in MEPS. Family Size The size of a family will also influence healthcare expenditures and uti lization. Family status was segmented to <3 and 3+. Income Income, like any other constrained resource, is a significant variable in influencing healthcare expenditures and utilization. As an individual is increasingly
49 constrained by less income, they will trade off expenditures on healthcare for basic needs like f ood and shelter. Income was based upon percentage of poverty level. Education. Education plays a significant part in the utilization of health care services. Education also plays a part in the Grossman model of health capital. Education was segmented based upon the following categories: no education; high school education; college or greater. Additional Analysis of Interest Medical Home Selection The continued development of Patient Centered Medical Homes into mainstream health care present s the direct question as to why individuals migh t select PCMHs. The study conduct ed a cursory review of characteristics of individuals in Patient Centered Medical Homes. The intention was to provide an improved understanding if certain individual charact eristics can predict one selecting a Patient Centered Medical Home Methods The primary dependent variable of total health expenditures creates some methodological challenges. It is well known that health expenditures are heavily skewed, meaning that a small portion of the population have the highest expenditures. Additionally, there are many observations in which individuals will have zero expenditures. Finally, the data is censored in that only positive values will be observed, since it is impossible to report negative medical expenditures. Given these challenges, the utiliza tion of traditional regression techniques to estimate the effect of Patient Centered Medical Homes on healthy populations would create biased and imprecise estimates. There are several econometric models that have been thoroughly tested to address these d ata characteristics (Blough, Madden, & Hornbrook, 1999; Duan, 1982;
50 Jones, 2000; Manning, 1998; Mullahy, 1998) The most wi dely used approach is a two part model in which the first part determines the probability of having expenditures. The basic E quation 3 1 can be represented as follows: Prob ( > 0 ) = ( ) (3 1 ) Where ( ) represents a distribution function that best represents the dependent variable or expenditures or utilization. Since we know that both expenditures and utilization tend to be heavily skewed, the study will use a logis tic distribut ion or logit model Equation 32 can then be represented as follows: Prob ( > 0 ) = (3 2 ) This first part addresses the challenge of a large number of zeros in the expenditure data. Once the first p art is evaluated and we know the probabili ty of expenditures or utilization, the second part is to apply the estimation model given the probability of expenditures or utilization of healthcare services occurring. This second part is traditionally an Ordinary Least Squares (OLS) regression in whic h the dependent variable is log (ln) transformed to account for the skewness of the observed expenditure values that are greater than zero. This c an be specified in Equation 33 ln ( y ) = + MH + Controls + (3 3 ) Where the dependent variable y represents either expenditures or utilization, represents the theoretical intercept or starting point of expenditures or utilization before any of the variables are introduced. MH represents a dichotomous variable of the existence of a Patient Centere d Medical Homes MH = 1 or not MH = 0 represents the coefficient of interest for the study and indicates the effect of a Patient Centered Medical Home on total expenditures or utilization. Controls represent the control
51 variables incorporated in the study. Finally, represents the error term of all the other unobserved variables or factors that can influence expenditures or utilization. The use of ln expenditures or utilization creates additional challenges in that the results will need to be transformed back into a value that is pertinent to the study (ln expenditures or utilization provide little insight into effects of Patient Centered Medical Homes). In addition to being a complex process, t his re transformation can be sensitive to mis interpr etation if there is heteroscedasticity in the data. This study used a methodological approach as presented by Manning et al. (Manning & Mullahy, 2001) The approach is intended to provide the best estimator for a given set of data while potentially avoiding the cumbersome and potential error inducing retransformation that is required of using log dependent va riables. The study used a Generalized Linear Model (GLM) based upon a generalized Gamma distribution for the second part of the model The central structure of the GLM model is a log link to the conditional mean. ln ( | ) = + MH + Controls + (3 4 ) or ( | ) = exp ( MH ) = ( MH ; ) (3 5 ) In this model, the choice of a mean and variance function for the observed raw scale dependent variables determines how the model will predict results. A flexible way of representing the potential variance function is as follows: ( y | x ) = ( ( ) ) (3 6 ) Where is some proportion to the mean and is some factor that is finite and nonnegative. We can simply plug in various values of and to represent different
52 distrib ution functions. The study use d a = 2 o r a generalized Gamma distribution. The generalized Gamma distribution is chosen based upon its flexibility in providing multiple modeling alternatives (Manning, Basu, & Mullahy, 2005) Using this initial model the log scale residuals were evaluated to see if they were skewed or if the kurtosis value was greater than 3. Based upon the analysis, the kurtosis was less than 3, the refore an GLM model was used. T he study also evaluated the results of a Poisson distribution and Gamma distribution for comparison. The results, while similar, st ill indicated the use of a Poisson distribution based upon the model fit tests Although both the OLS models and the GLMs have been thoroughly validated and would be expected to produce acceptable results, this methodol ogical approach to the study is believed to have yielded the most precise and unbiased results.
53 CHAPTER 4 RESULTS The results from this analysis are presented in two sections. Descripti ve data analysis look s at descriptive statisti cs of the sample. Modeling look ed at the several models evaluated during the study to determine the bes t fit and prediction. This section will include bivariate and multivariate analysis results. Descriptive Statisti cs The total number of respondents in the sample for the period of the s tudy from 2005 to 2009 was 168, 991. The Table 41 presents the characteristics for the independent variables During the period used for this analysis, the respondent sex was narrowl y even, with 48% of the respondents being Male and 52% F emale. Thirty eight percent were considered married while 62% were considered unmarried. The unmarried category included those that were never married, widowed, divorced or separated. The race of study population was 73% white; 19% white and 8% other. The other race category included American Indian, Asian, Pacific Islander and other. Age among the respondents was broken up into 4 categories. Thirty percent of the respondents were under age 18 a t the time of the survey. Twenty three percent of the respondents were bet ween the ages of 18 and 34. Thirty six percent of the respondents were between the ages of 35 and 64. The final category of those over 65 comprised 11% of the sample. The final demographic variable was family size. Families of two or less comprised 34% of the sample. Families of 3 or more individuals comprised 66% of the sample. Moving to other variables of interest, the study looked at respondents Metropolitan Statistical Area (MSA), their income level, and the highest degree obtained. The
54 respondents reported being in a MSA 84% of the time. Sixteen percent of the respondents were not in an MSA. Less than 1% of the sample did not respond or were in a classifiable MSA area. T he income of the respondents was broken into five categories: poor/negative; near poor; low income; middle income; high income. These categories were represented in the sample in the following percentages: poor/negative income was 21%; near poor income was 7%; low income was 17%; middle income was 29%; and high income was 27%. The final additional variable related to socioeconomic status included in the study was the highest degree obtained. The respondents reported that 18% had no degree at all. 31% o f the respondents had successfully obtained a high school diploma, while 3% had completed the requirements to obtain a GED. Ten percent of the respondents had obtained a bachelors degree. Finally, only 4% and 1% had obtained a masters degree or a doctorate degree respectively. Five percent of the respondents had obtained so other form of degree and 28% of the respondents were under 16 and therefore their degree status was inapplicable. Insurance status for the sample population appears consistent wit h the general population. Fifty five percent of the sample population has private insurance. Seventeen percent of the sample population has no insurance. Public insurance is split with 8% under Medicare coverage and 20% under Medicaid coverage. The fina l variables of interest relate to the existence of Patient Centered Medical Home s and the health status of the respondents. Applying the criteria outlined previously in this study, 16% of the respondents were classified as healthy. In other words, 16% of the respondents had PCS scores greater than the 75th percentile. Forty five percent of the respondents were classified as sick. Thirty nine percent of the
55 respondents did not provide enough information to classify their health status in the SF 12 section and therefore were considered mi ssing. Using the simplified definition of the existence of a Patient Centered Medical Home in which five characteristics had to be present only 8% of the respondents were considered to be involved in a Patient Centered M edical H ome. Using the mor e compreh ensive scoring definition of a P atient Centered Medical H ome, 76% of the respondents were considered to be in some version of a Patient Centered Medical H ome based upon displaying a 25% degree of Patient Centered Medical Home ness. Bivariate Statistics The results of bivariate analysis of research variables and their relation to Patient Centered Medical Homes, using both the simplified and comprehensive definition, can be found in Table 42 and Table 43 respectively PC MH Simplified Definition The simplified definition of Patient Centered Medical Homes identifies its existence based on the positive response of 5 criteria (office hours, after hours contact, usual source of care, provider speaks language, shared decision making) All five criteria must be present in order for the respondent to be considered in a Patient Centered Medical Home. Pearsons Chi squared and Cram rs V was calculated for all the research variables to evaluate the relationship between them and the existence of Patient Centered Medical Homes. Evaluating gender there appears to be no difference between males and females selecting a Patient Centere d Medical Home. There appears to be significance between gender and PCMH (Pr = 0.002). Cram rs V (0.0076) indicates that the relationship is weak and more influenced by the large sample size. Marital status also appears to have significance with Patient Centered Medical
56 Home s (Pr < 0.001 ) but an overall weak relationship with Cram rs V = 0.0250. Race, while not showing any break between the PCMH and no PCMH, is not significant (Pr = 0.640). The breakdown of age and PCMH shows a moderate level of significance with a Chi squared test Pr = 0.000 and a Cramrs V = 0.0976. The family size relationship with a Patient Centered Medical Home was significant (Pr < 0.001) with a weak relationship Cramrs V = 0.0591. MSA also has a significant weak relationshi p with the Chisquared test Pr < 0.001 and Cram rs V = 0.0430. Income, like most of the variables already presented, has a weak significance with Pr = 0.000 and Cramrs V = 0.0617. Education is the only variable that shows something of a moderate significance with Patient Centered Medical Homes with a Chi squared test result Pr = 0.000 and Cram rs V = 0.1004. This is probably the result of a higher educated individual aware of and seeking the characteristics of a Patient Centered Medical Home. PCMH Comprehensive Definition The comprehensive definition used seventeen characteristics of Patient Centered Medical Homes. A percentage of those characteristics (25%) only had t o be present to qualify a respondent as in a Patient Centered Medical Home. Using this comprehensive definition allows a more nuanced approach versus the strict either/or approach with the simplified definition. As a result, the comprehensive definition identifies a significant number of additional respondents in Patient Centered Medical Homes versus the simplified definition. The additional respondents in this definition reveal some slightly different results in the bivariate analysis. The relationship between gender and PCMH results is the same with the comprehensive definition as with the simplified definition. The results of the Chi -
57 squared test and Cram rs V are Pr = 0.000 and 0.0967 respectively. Marital status results also appear to be consisten t between the simplified and comprehensive definitions (Pr = 0.000; 0.0264). The comprehensive definition showed significance between race and Patient Centered Medical homes (Pr = 0.000 0.0181). The results are different from the simplified definition, which showed no significance between the two variables. Family size also showed significance and a weak relationship (Pr = 0.000; 0.0304), which is consistent with both definitions. MSA showed significance with a weak association approaching a moderate association (Pr = 0.000; 0.1223). Income also shows significance with a weak to moderate association with Patient Centered Medical Homes (Pr = 0.000; 0.1268). Similar to the simplified definition of Patient Centered Medical Homes, education appears to be significant with a moderate association (Pr = 0.000; 0.1616). The most significant difference between the simplified definition of Patient Centered Medical Homes and the comprehensive definition is the age variable. Age appears to be significant with a moderate association with Patient Centered Medical Homes (Pr = 0.000; 0.2313) this is contrasted with a weak to no association for the simplified Patient Centered Medical Home Definition. Expenditures Figure 4 1, Figure 4 2, Figure 4 3 and Figure 4 4 display health care expenditure information from the study sample. The results are stratified by Patient Centered Medical Home definition (simplified and comprehensive) and health status. The expenditures are further broken down by expenditure categ ory. Interestingly enough, the simplified definition of Patient Centered Medical Homes yielded a higher mean expenditure ($3,355) for individuals identified as being within a PCMH versus those that were identified as not being in a PCMH ($2,773). Looking at
58 the comprehensive definition of Patient Centered Medical Homes, the data was slightly different. Individuals identified as being in a Patient Centered Medical Home under the comprehensive definition had slightly less expenditures ($4,109) than those no t in a medical home ($4,330). The difference between these two results can be attributed to the comprehensive definition capturing a larger portion of the sample population. The expectation would be that individuals in Patient Centered Medical Homes woul d have lower overall expenditures since their care is carefully managed. Again, t he difference could be attributable to the fact that the comprehensive definition captures a larger sample than the simplified definition. The comprehensive definition captures a larger sample due to having 20 elements. Only 5 of those 20 elements need to be present to fall under the Patient Centered Medical Home definition. Compared to the simplified definition, all 5 specific elements must be present to classify an indivi dual in a Patient Centered Medical Home. The breakdown of expenditures by expenditure category provides additional information on the differences between the simplified Patient Centered Medical Home definition and the comprehensive definition. The top thr ee categories for both the comprehensive and simplified definition are the same. Facility hospital, office events and prescription expenditures are the top three expenditure categories for both definitions. The lower expenditures for individuals in a Pat ient Centered Medical Home using the comprehensive definition appear to be attributed to lower expenditures in the bottom categories. Office events and prescription drugs are higher by a large margin using the simplified definition as compared to the comprehensive definition for individuals not in a PCMH ($916; $852 vs $461; $307 respectively ). All the lower
59 categories (Dental, Hospital Physician, ER Facility, Home Health, Outpatient Facility) exhibit the same difference by a lesser margin. The simplifie d definition presents the same expenditure categ ories in the same order. The expenditure for those in Patient Centered Medical Homes using both the simplified definition and comprehensive definition are comparatively close. The top three categories Hospi tal Facility, Office Events and Scripts are $1,096, $944, and $818 for the simplified definition and $1,100, $1,008, and $955 for the comprehensive definition for individuals in Patient Centered Medical Homes. Utilization The breakdown of utilization by c ategories between Patient Centered Medical Home (simplified definition) and no Patient Centered Medical Home is shown in Figure 4 5. The average number of events for the sample population is 26.3 with a standard deviation of 60.32. The number of events w as broken in 6 different categories for the analysis The number of prescriptions was included as events since they would imply events to pharmacies. The majority of the categories averaged less than 1 visit. The six categories that averaged more than 1 visit are shown in Figure 45. The six categories are ordered the same from highest to lowest for both no Patient Centered Medical Home and Patient Centered Medical Home. The highest category is prescription drugs (9.4 and 9.5, no PCMH and PCMH respecti vely). The next highest categories are provider office events (4.3;4.9) and physician office events (2.8; 3.4). Evaluating this graph shows that individuals in Patient Centered Medical Homes appear to have slightly higher prescription drugs, provider and physician office events Patient Centered Medical Homes have improved access to providers which would reasonably explain why individuals in these models would have more office events
60 The breakdown of utilization by categories between Patient Centered Medical Home (comprehensive definition) and no Patient Centered Medical Home is shown in Figure 46. The top categories are the same for both Patient Centered Medical Home definitions. T he most compelling statement of F igure 4 6 is the disparity between the events for those that are in a Patient Centered Medical Home and those that are not. The possible explanation being that the comprehensive definition includes more respondents in a Pa tient Centered Medical Home. Multivariate Analysis Three separate models were used to predict expenditures. All statistics were run on Stata SE version 12. The survey procedures were utilized with the subpopulation of healthy individuals. The total expenditures were skewed near zero, however there were few expenditures that were actually zero (less than 1%) As a result, a two part model was not utilized. Two Generalized Linear Models were estimated to determine the model with the best fit One utiliz ed a Gamma error distribution, and the other utilized a Poisson distribution. Finally a log linear model was also utilized to predict expenditures based upon Patient Centered Medical Home and health status. Pregibons Link test and Hosmer Lemeshow test w as conducted on all the models to determine the best fit. The results of both of the fit tests can be found on Table 48 and Table 49. Based upon these results, the best predictor model that was selec ted was the GLM Poisson family and log link The mode l evaluated 96,262 observations. The subpopulation of the model representing the healthiest individuals had 19,731 observations. The paper will focus discussion on the results from the GLM Poisson family with log link results. The PCMH variable shows a slight positive effect on total expenditures (0.1 0 3 ; P = 0.098). The
61 positive effect implies individuals in Patient Centered Medical Homes will have higher expenditures than those that are not in PCMHs. An individual in a Patient Centered Medical Home wi ll increase expenditures by $0.120. Evaluating the results of the model shows that several of the variables did not have any significance. The metropolitan statistical area did not appear to have any significance (P = 0.128). Only n ear poor status appeared to have any signi ficance (P = 0.031). Interestingly, all the educational levels did not have any significance except the master s level degree. Since all the educational categories appear to have a small t value coupled with large P values, it would indicate that these coefficients are not significantly different from zero. Several variables are significant in this model. Gender ( P < 0.001 ) and marital status (P < 0.001) show significance. The race categories all appear to be significant. The age categories are also all significant. Income for the middle income and near poor categories appears to be significant (P = 0.011 ) Overall, the model indicates that keeping all variables constant, healthy individuals that are in a Patient Centered Medical Home will have slight ly higher expenditures based upon the model ing equation PCMH coefficient ( 0. 1 0 3 ). Married males have the effect of lower expenditures ( = 0.040). As expected, individuals under the age of 65 also had lower expenditures. Individuals under the age of 65 have not developed chronic conditions and have lower incidence of high cost hospital events The model predicts overall higher total expenditures for healthy individuals in Patient Centered Medical Homes. Comparing treatment effec ts of Patient Centered Medical Homes for healthy individuals the model predicts mean expenditures of $2,501 for Patient Centered Medical Homes versus $2,215 for no Patient Centered Medical Homes.
62 The Gamma model with log link predicts similar results for utilization as it does for expenditures. Utilization for healthy individuals in Patient Centered Medical Homes is predicted to be higher than utilization for healthy individuals outside of Patient Centered Medical Homes. Table 410 contains results from the model. The coefficient for the Patient Centered Medical Home variable is slightly positive (0.186, P<0.01) which would indicate that being in a Patient Centered Medical Home would slightly increase your total number of events The existence of a Pat ient Centered Medical Home will increase your events by 18.6%. All the control variables appear to be statistically significant with P<0.01 with the exception of MSA, Income and Education. The predicted events 14.58, are more than the actual events 13.7 9. In the same manner as expenditures, utilization for healthy individuals in Patient Centered Medical Homes are higher than healthy individuals that are not in Patient Centered Medical Homes. The inclusion of insurance variables in the model does not change the results obtained. The results of the models with the insurance variables can be found in Table 4 11 through Table 414. In the case of both expenditures and events, the predicted results are higher than the actual results. The insurance variables are broken down by no insurance, Medicare insurance and Medicaid insurance. Private insurance serves as the reference variable. In all models the insurance variables are statistically significant (P < 0.05). Evaluating the insurance variables, the coefficients are consistent with expected results. The model indicates that individuals without insurance would have lower expenditures since the sign of the coefficient is negative ( 0.510 Poisson). Individuals that have some form of insurance have a p ositive coefficient and therefore a positive effect on expenditures and events. The coefficient signs are consistent with
63 expected results in that individuals without insurance would have a difficult time accessing care compared to individuals with some f orm of insurance. Overall, however, the presence of insurance does not appear to influence to results of the model.
64 Table 41. Independent v ariable characteristics Variable Level % Gender Male 48 Female 52 Marital s tatus Married 38 Un married 62 Race White 73 Black 19 Other 8 Age Under 18 30 18 34 23 35 64 36 65+ 11 Family size <3 34 3 66 MSA Non MSA 16 MSA 84 Inapplicable <1 Income Poor/negative 21 Near poor 7 Low income 17 Middle income 29 High income 27 Highest degree No degree 18 GED 3 High school diploma 31 Bachelors degree 10 Masters degree 4 Doctorate degree 1 Other degree 5 Under 16 inapplicable 28 Missing 1 Healthy Healthy 16 Sick 45 Missing* 39 Insurance Private Insurance 55 No Insurance 17 Medicare Insurance 8 Medicaid Insurance 20 PCMH 8 PCMH comprehensive 76 missing values correspond to unanswered questions from survey respondents
65 Table 42. Bivariate analysis with simplified PCMH status PCMH Frequencies Variable Level No Yes (Pr = ) Cramrs Gender Male 74,399 6,213 0.002 0.0076 Female 81,202 7,177 Marital status Married 59,022 4,478 0.000 0.0250 Un married 96,579 8,912 Race White 113,114 9,766 0.640 0.0023 Black 29,212 2,513 Other 13,275 1,111 Age Under 18 44,733 5,980 0.000 0.0976 18 34 36,516 1,952 35 64 57,244 4,323 65+ 17,108 1,135 Family size <3 54,513 3,302 0.000 0.0591 3 101,088 13,390 MSA Non MSA 24,955 1,507 0.000 0.0430 MSA 129,333 11,870 Inapplicable 1,313 13 Income Poor/negative 33,291 2,206 0.000 0.0617 Near poor 10,494 662 Low income 26,978 1,897 Middle income 44,326 3,951 High income 40,512 4,674 Highest degree No degree 29,430 1,572 0.000 0.1004 GED 4,909 318 High school diploma 48,421 3,313 Bachelors degree 15,105 1,132 Masters degree 5,682 498 Doctorate degree 1,600 141 Other degree 7,600 646 Under 16 inapplicable 41,597 5,739 Healthy Healthy 71,577 4,954 0.717 0.0011 Sick 24,924 1,707
66 Table 43. Bivariate analysis with comprehensive PCMH status PCMH Frequencies Variable Level No Yes (Pr = ) Cramrs Gender Male 22,975 57,637 0.00 0 0.0967 Female 17,862 70,517 Marital status Married 14,420 49,080 0.000 0.0264 Un married 26,417 79,074 Race White 29,111 93,769 0.000 0.0181 Black 8,067 23,658 Other 3,659 10,727 Age Under 18 9,439 41,274 0.000 0.2313 18 34 15,652 22,816 35 64 14,176 47,391 65+ 1,570 16,673 Family size <3 12,928 44,887 0.000 0.0304 3 27,909 128,154 MSA Non MSA 5,495 20,967 0.000 0.1223 MSA 34,264 106,939 Inapplicable 1,078 248 Income Poor/negative 10,239 25,258 0.000 0.1268 Near poor 3,171 7,985 Low income 8,534 20,341 Middle income 11,750 36,527 High income 7,143 38,043 Highest degree No degree 10,787 20,215 0.000 0.1616 GED 1,614 3,613 High school diploma 13,813 37,921 Bachelors degree 3,154 13,083 Masters degree 905 5,275 Doctorate degree 303 1,438 Other degree 1,646 6,600 Under 16 inapplicable 7,829 39,507 Healthy Healthy 16,010 60,521 0.000 0.1045 Sick 8,270 18,361 Insurance Private Insurance 84,418 9,041 0.0000 No Insurance 27,589 852 0.0000 Medicare Insurance 12,099 798 0.0000 Medicaid Insurance 31,495 2,699 0.816
67 Table 44. GLM model predicting expenditures (Gamma link) Total Events Coefficient SE t P>|t| [95% C.I.] PCMH 0.106 0.060 1.760 0.079 0.012 0.225 Gender Female Reference Male 0.328 0.038 8.540 0.000 0.403 0.252 Race White Reference African American 0.223 0.063 3.530 0.000 0.347 0.099 Other 0.222 0.067 3.320 0.001 0.353 0.090 Marital Status Un married Reference Married 0.043 0.012 3.560 0.000 0.066 0.019 Age <18 Omitted 18 34 0.445 0.073 6.120 0.000 0.589 0.302 35 64 0.206 0.069 3.000 0.003 0.341 0.071 65+ Reference Family Size Under 3 Reference 3 and over 0.107 0.041 2.600 0.010 0.188 0.026 MSA Reference Non MSA 0.083 0.062 1.340 0.181 0.203 0.038 SF 12 Physical 0.017 0.007 2.340 0.020 0.003 0.031 Income Level Poor/Negative Reference Near Poor 0.277 0.116 2.380 0.018 0.505 0.048 Low Income 0.018 0.138 0.130 0.898 0.289 0.254 Middle Income 0.137 0.088 1.550 0.121 0.311 0.036 High Income 0.057 0.083 0.690 0.488 0.220 0.105 Education No Degree Reference GED 0.022 0.075 0.290 0.773 0.170 0.126 High School Diploma 0.081 0.128 0.630 0.526 0.170 0.332 Bachelors Degree 0.036 0.045 0.800 0.427 0.125 0.053 Masters Degree 0.170 0.059 2.890 0.004 0.055 0.286 Doctorate Degree 0.002 0.083 0.020 0.984 0.162 0.165 Other Degree 0.012 0.065 0.180 0.855 0.139 0.115 Insurance Status Private Insurance Reference No Insurance 0.502 0.115 4.370 0.000 0.728 0.276 Medicare Insurance 0.474 0.137 3.470 0.001 0.205 0.742 Medicaid Insurance 0.337 0.137 2.450 0.015 0.067 0.607 Constant 0.665 0.450 1.480 0.140 0.220 1.550
68 Table 45. Log linear model predicting expenditures Total Events Coefficient SE t P>|t| [95% C.I.] PCMH 0.125 0.041 3.070 0.002 0.045 0.205 Gender Female Reference Male 0.467 0.024 19.710 0.000 0.513 0.420 Race White Reference African American 0.304 0.040 7.640 0.000 0.383 0.226 Other 0.276 0.045 6.120 0.000 0.365 0.188 Marital Status Un married Reference Married 0.037 0.009 4.370 0.000 0.054 0.021 Age <18 Omitted 18 34 0.715 0.062 11.620 0.000 0.836 0.594 35 64 0.463 0.058 7.970 0.000 0.578 0.349 65+ Reference Family Size Under 3 Reference 3 and over 0.226 0.026 8.700 0.000 0.277 0.175 MSA Reference Non MSA 0.085 0.041 2.050 0.041 0.166 0.004 SF 12 Physical 0.0370 0.0059 6.2900 0.0000 0.0255 0.0486 Income Level Poor/Negative Reference Near Poor 0.205 0.083 2.470 0.014 0.367 0.042 Low Income 0.073 0.056 1.320 0.189 0.183 0.036 Middle Income 0.025 0.056 0.450 0.651 0.136 0.085 High Income 0.111 0.057 1.940 0.053 0.001 0.224 Education No Degree Reference GED 0.193 0.045 4.270 0.000 0.282 0.104 High School Diploma 0.096 0.089 1.070 0.283 0.271 0.079 Bachelors Degree 0.134 0.035 3.830 0.000 0.203 0.065 Masters Degree 0.128 0.044 2.900 0.004 0.041 0.215 Doctorate Degree 0.040 0.072 0.550 0.582 0.102 0.181 Other Degree 0.085 0.048 1.780 0.075 0.179 0.009 Insurance Status Private Insurance Reference No Insurance 0.628 0.044 14.210 0.000 0.715 0.541 Medicare Insurance 0.167 0.104 1.610 0.107 0.037 0.371 Medicaid Insurance 0.135 0.075 1.800 0.073 0.013 0.282 Constant 7.383 0.464 15.910 0.000 6.471 8.294
69 Table 46. GLM model predicting expenditures (Poisson) Total Events Coefficient SE t P>|t| [95% C.I.] PCMH 0.103 0.062 1.660 0.098 0.019 0.226 Gender Female Reference Male 0.297 0.040 7.520 0.000 0.375 0.219 Race White Reference African American 0.191 0.063 3.020 0.003 0.316 0.067 Other 0.201 0.075 2.690 0.007 0.348 0.054 Marital Status Un married Reference Married 0.040 0.012 3.360 0.001 0.064 0.017 Age <18 Omitted 18 34 0.368 0.089 4.150 0.000 0.542 0.194 35 64 0.161 0.090 1.790 0.074 0.337 0.016 65+ Reference Family Size Under 3 Reference 3 and over 0.107 0.037 2.860 0.004 0.181 0.033 MSA Reference Non MSA 0.092 0.061 1.520 0.128 0.211 0.027 SF 12 Physical 0.016 0.008 2.070 0.039 0.001 0.031 Income Level Poor/Negative Reference Near Poor 0.247 0.114 2.160 0.031 0.472 0.022 Low Income 0.005 0.111 0.040 0.965 0.223 0.213 Middle Income 0.086 0.082 1.060 0.291 0.246 0.074 High Income 0.022 0.077 0.290 0.773 0.172 0.128 Education No Degree Reference GED 0.012 0.080 0.150 0.883 0.168 0.145 High School Diploma 0.038 0.100 0.370 0.708 0.160 0.235 Bachelors Degree 0.046 0.044 1.050 0.295 0.132 0.040 Masters Degree 0.173 0.061 2.860 0.004 0.054 0.292 Doctorate Degree 0.003 0.086 0.030 0.973 0.166 0.172 Other Degree 0.027 0.063 0.430 0.670 0.150 0.097 Insurance Status Private Insurance Reference No Insurance 0.510 0.131 3.890 0.000 0.768 0.253 Medicare Insurance 0.394 0.149 2.640 0.009 0.100 0.688 Medicaid Insurance 0.256 0.114 2.240 0.026 0.031 0.481 Constant 7.383 0.464 15.910 0.000 6.471 8.294
70 Table 47. Summary of model predicted expenditures GLM (Gamma) GLM (Poisson) Log linear Variable Mean S.D. Mean S.D. Mean S.D. Total Expenditures 2,194 5,052 2,194 5,052 2,194 5,052 Predicted Expenditures 2,237 842 2,214 748 9,537 5,24 8 Residual 42 5,006 20 5,001 7,343 6,54 7 Table 48. Pregibons link test results Variable GLM (Gamma) GLM (Poisson) Log linear Y hat (coefficient) 1.223813 1.129569 0.9710974 Y hat 2 (coefficient) 0.0000902 0.0000521 0.0037467 F() 44.34 8.64 1.52 Prob > F 0.0000 0.0033 0.2169 Table 49. Hosmer Lemeshow test Variable GLM (Gamma) GLM (Poisson) Log linear Category 1 12.74267 40.06009 2592.066 Category 2 129.4564 189.6231 4428.992 Category 3 282.4192 338.8425 5316.073 Category 4 179.4008 104.6276 6132.499 Category 5 57.18888 60.35447 7070.828 Category 6 223.3272 247.0423 7972.33 Category 7 73.53076 76.56908 9286.148 Category 8 15.9368 42.67284 10670.17 Category 9 95.12493 100.6758 13113.44 Category 10 446.3861 264.6951 18259.74 F() 3.07 4.32 6,512 .1 0 Prob > F 0.000 7 0.000 0 0.0000
71 Table 410. GLM model predicting events (Gamma) Total Events Coefficient SE T P>|t| [95% C.I.] PCMH 0.145 0.045 3.210 0.001 0.056 0.233 Gender Female Reference Male 0.552 0.029 19.290 0.000 0.608 0.496 Race White Reference African American 0.398 0.040 9.870 0.000 0.477 0.319 Other 0.394 0.056 7.050 0.000 0.504 0.284 Marital Status Un married Reference Married 0.025 0.010 2.620 0.009 0.044 0.006 Age <18 Omitted 18 34 0.842 0.109 7.750 0.000 1.055 0.628 35 64 0.488 0.098 4.970 0.000 0.681 0.295 65+ Reference Family Size Under 3 Reference 3 and over 0.325 0.029 11.140 0.000 0.383 0.268 MSA Reference Non MSA 0.006 0.049 0.120 0.901 0.103 0.091 SF 12 Physical 0.059 0.007 8.090 0.000 0.044 0.073 Income Level Poor/Negative Reference Near Poor 0.032 0.103 0.310 0.760 0.234 0.171 Low Income 0.093 0.072 1.290 0.197 0.048 0.234 Middle Income 0.108 0.057 1.890 0.059 0.004 0.220 High Income 0.244 0.057 4.250 0.000 0.131 0.357 Education No Degree Reference GED 0.338 0.056 6.000 0.000 0.449 0.228 High School Diploma 0.107 0.103 1.040 0.300 0.309 0.096 Bachelors Degree 0.177 0.035 5.030 0.000 0.246 0.108 Masters Degree 0.079 0.045 1.750 0.081 0.010 0.168 Doctorate Degree 0.045 0.082 0.550 0.581 0.116 0.207 Other Degree 0.027 0.063 0.430 0.670 0.150 0.097 Insurance Status Private Insurance Reference No Insurance 0.828 0.056 14.840 0.000 0.937 0.718 Medicare Insurance 0.506 0.141 3.600 0.000 0.230 0.783 Medicaid Insurance 0.403 0.083 4.860 0.000 0.240 0.565 Constant 0.359 0.444 0.810 0.420 0.514 1.231
72 Figure 41. E xpenditures by PCMH definition 2773 3355 4330 4109 0 1,000 2,000 3,000 4,000 Mean Total Expenditures ($)Simplified PCMH Comprehensive PCMH No Yes No Yes
73 Figure 4 2. Expenditure categories by simplified PCMH 135 164 186 258 274 852 916 1108 140 164 172 303 306 818 944 1096 0 500 1,000 No PCMH PCMH Dental Script Outpatient Facility Office Visits Hospital Facility Hospital Physician ER Facility Home HealthExpenditure Categories
74 Figure 43. Expenditure categories by comprehensive PCMH 128 140 143 171 180 307 461 1145 133 162 193 278 306 955 1008 1100 0 500 1,000 1,500 No PCMH PCMH Dental Script Outpatient Facility Office Visits Hospital Facility Hospital Physician ER Facility Home HealthExpenditure Categories
75 Figure 44. Expenditure categories by health status 182 236 274 285 436 1365 1404 1614 26 87 92 154 290 370 412 646 0 500 1,000 1,500 Not Healthy Healthy Dental Script Outpatient Facility Office Visits Hospital Facility Hospital Physician ER Facility Home HealthExpenditure Categories
76 Figure 45. Utilization by categories (simplified definition) 9.4 4.3 2.8 1.7 1.4 1.3 9.5 4.9 3.4 1.7 1.5 1.2 0 2 4 6 8 10 No PCMH PCMH # Prescibed Meds # Office Provider Visits # Office Physician Visits # Home Health Provider Days # Office Non Physician Visits # Agency Home Health Days
77 Figure 46. Utilization by categories (comprehensive definition) 2.5 1.5 .97 .79 .53 .56 12 5.2 3.5 2.1 1.7 1.5 0 5 10 15 0 1 # Prescibed Meds # Home Health Provider Days # Office Provider Visits # Office Non Physician Visits # Office Physician Visits # Agency Home Health Days
78 CHAPTER 5 DISCUSSION The impact of Patient Centered Medical Homes on healthy populations appears to have a counter intuitive effect based upon the results of this stu dy. Several studies documented in this paper and other research have shown that Patient Centered Medical Homes are consistent in lowering health care expenditures for chronically ill populations. The compounded results of this published research creates the impression that any individual in this model of care should experience the same results. The results of this study indicate that healthy individuals in Patient Centered Medical Homes actually have higher expenditures and utilization on average ($2,237) than those that are not in Patient Centered Medical Homes ($2,194). The traits of Patient Centered Medical Homes may be most effective with chronically ill in regards to utilization and expenditures. This may be due to the fact that t he chronically ill are already consuming a high level of health services and Patient Centered Medical Homes may be able to eliminate some redundant services. These results raise interesting policy questions about Patient Centered Medical Homes and their i mplementation as a tool to control health care expenditures. An unintended byproduct of the comprehensive coordinated care model of Patient Centered Medical Homes could be the triggering effect of higher health care utilization among healthy individuals w ithin this model of care delivery. In contrast, c hronically ill populations appear to lower their health care utilization within the Patient Centered Medical Home care delivery model Patient Centered Medical Homes for chronically ill populations are att ractive because they ensure that those services are appropriately targeted to the patients needs. A result of this care management is that expenditures for the chronically ill is lower in this model since they are receiving the
79 right services at the right time. Healthy individuals, on the other hand, in this model may feel the encouragement to seek services. The seeking of services would be consistent with Grossmans Human Capital model. The Human Capital model posits that individuals will seek to inves t in their human capital by consuming health services which allows them to maintain or increase their human capital. The higher the stock of human capital, the more healthy days are produced. Individuals will seek to increase their number of healthy days because it allows them to devote more time to other activities from which they derive utility Therefore, the ease of access to health care services and the derived benefit (human capital) of consuming health care services influences the higher expenditures and utilization seen in this study. One of the primary themes of Patient Centered Medical Homes is increased access to care. PCMHs accomplish this through improved office hours, email communication and easy phone access. Another primary theme of Patient Centered Medical Homes is coordination of professional and community resources. The activity of coordinating services increases awareness of potential services available to patients. The heightened awareness will make it easier for individuals to s eek services. Patient Centered Medical homes could provide an enabling environment from the traits highlighted above for healthy individuals to consume health services. The enabling effect will in turn lead to higher expenditures for healthy individuals w ithin Patient Centered Medical Homes without significant improvement in health. The standard of care may also be driving the results seen in this study. Healthy individuals generally do not have a compelling reason to visit a provider unless ill. One of the primary concepts of Patient Centered Medical Homes is the care coordination
80 and management of the population. The care coordination activity implies a PCMH will reach out to their patient population to ensure the appropriate standard of care activiti es are being completed, such as annual visits, diag nostic exams, etc. A healthy individual that is not in a Patient Centered Medical Home would not access all these services and therefore would have a lower utilization and expenditures of health care serv ices. The results of this study imply that a blanket approach to P atient Centered Medical Homes may not be the best approach. Patient Centered Medical Homes may not be an ideal primary care model for everyone. The results of this study indicate that healthy individuals behave differently in Patient Centered Medical Homes in terms of increased expenditures and events Chronically ill individuals in Patient Centered Medical Homes appear to have lower expenditures and utilization based upon published re search. Combining the results of this study along with published studies may indicate that the implementation of Patient Centered Medical Homes would benefit from a tiered approach. In this manner, chronically ill individuals would receive the full suite of services of a Patient Centered Medical Home. Healthy individuals on the other hand would continue to receive the usual standard of care. The author would like to point out that even in light of the results of this study, there still a re other compelli ng economic benefits to Patient Centered Medical Homes that are difficult to quantify. The comprehensiveness of care that is delivered in Patient Centered Medical Homes may produce other economic benefits that are difficult to capture an d represent in this study. Patient Centered Medical Homes, by their design, presents a onestop shop for medical care. Individuals in an appropriately structured Medical Home could see a health care provider, fill a script and complete a diagnostic
81 tes t in one visit. Compared to individuals that are not in a Medical Home, each of t hose activities could present as distinct visits to a provider, pharmacy or testing facility. The economic costs of missed workdays or childcare needed for these visits coul d be substantial. In other words, individuals in Patient Centered Medical Homes could accomplish three things in one visit or day off versus three separate visits or days off. Comparable Results The impact of Patient Centered Medical Homes on healthy indi viduals is a minimally studied area. At the time of this analysis, there were no pu blished studies on healthy individuals and Patient Centered Medical Homes. Despite the lack of published research on healthy individuals, there are several studies of Pati ent Centered Medical Homes and commercially insured or pediatric populations. These studies are primarily concerned with the implementation of some type of Patient Centered Medical Ho me model and the outcomes of the implementation in the context of expenditures and/or utilization In general, the results of these studies have been consistent with the res ults of studies looking at the e ffects of Patient Centered Medical Homes on chronically ill populations such as children with special health care needs and diabetics Patient Centered Medical Homes have been shown to be effective in reducing hospital and emergency room events and improving self reported patient satisfaction. A recent published study, while not directly evaluating healthy populations and Pa tient Centered Medical Homes, reported results that could be seen as inconsistent with the results of this analysis. DeVries, et al. evaluated the effects of Patient Centered Medical Homes in a commercial insurance environment on quality, utilization and costs (Devries et al., 2012) Their study utilized claim s data from Empire Blue Cross and Blue Shield during 2010. The cross sectional study evaluated 31,032
82 patients in Patient Centered Medical Homes versus 350,015 patients not in Patient Centered Medical Homes. The results of the analysis showed that adults and children in PCMHs had 12% and 23% lower odds of hospitalization respectively. The adults and children also had 11% and 15% lower odds of Emergency room events Risk adjusted total per member per month costs 14.5% and 8.6% lower for adults and children respectively. The reported results from this study, while consistent with published research, is inconsistent with this papers analysis. There are several issues to point out when trying to compare the results of the DeVries study to the results of this papers study. First is the issue of generalizability The DeVries group did not compare health status of the study population. Therefore while it is reasonable to expect that anyone who is in a Patient Centered Medical Home to have lower health care expenditures, we cannot with certainty confirm that based upon the results of the DeVries study without isolating healthy versus sick populations The range of expenditures between healthy and sick individuals could be quite large. The large range could have an averaging effec t on the expenditure outcomes. The large cost savings from the sicker population can mask the increased costs from the healthier population when averaged together. The other issue with the study involves the definition of Patient Centered Medical Homes. As has been highlighted, there are no consistent criteria to determine a Patient Centered Medical Home. The study affiliated patients to a provider based upon claim data during 2007 2008. Using this affiliation, the study then determined whether or not a provider was considered a PCMH based upon NCQA recognition of NCQA status in 2010. In addition, the study did not clarify if the Patient Centered Medical Homes in question were identified as Level 1, 2 or
83 3. There is a significant distinction between these levels, as Level 1 only requires a score of 25/100 to receive NCQA certification as a Patient Centered Medical Home. The difference in timing between provider affiliation and PCMH status coupled with nonspecificity of PCMH status certainly raises the question of correlation between PCMHs and utilization, costs and quality. Another recently published study presents some results that could be seen as consistent with the results of this analysis. Romaire et al. looked at a pediatric population (0 17) to evaluated the effects of Patient Centered Medical Homes on use and expenditures (Romaire, Bell, & Grossman, 2012) While this study examines the pediatric population, the utilization of the Medical Expenditure Panel Survey is the same data source as this study. MEPS does not include survey questions specifically to identify Patient Centered Medical Homes. The published study and this study both had to infer the existence of a medi cal home based upon survey responses that mapped to some Patient Center Medical Home domain. Romaire found that children in a medical home had a greater incidence of preventative events (incidence rate ratio 1.11) compared to children without a medical home. In addition, children in medical homes also had greater odds of incurring total, outpatient, prescription medication and dental expenditures, with odds ratios ranging from 1.09 1.38. Finally, the published study found that despite these greater o dds of incurring expenditures, the expenditures were no different for children in a medical home versus children without a Patient Centered Medical Home. The results of the Romaire study can be seen as consistent with the results of this study. Individual s in Patient Centered Medical Homes have greater odds to incur total
84 outpatient, prescription medication and dental expenditures than those that are not in Patient Centered Medical Homes. Similar to the Romaire study this study shows that healthy indivi duals in Patient Centered Medical Homes incur higher total expenditures than healthy individuals that are not in Patient Centered Medical Homes. There are some differences between this study and the Romaire study that should be discussed First, the Roma ire study only focused on a pediatric population and just controlled for parent self reported child health status. There could be differences between the pediatric population and adult population that lead to differences between utilization and expenditur es. Children, for example, tend to have more met health care needs than adults. Another issue with drawing a relationship between the published study and this study is that despite the greater odds of use, Romaire did not find any differences in expendit ures between children in and out of Patient Centered Medical Homes. The lack of variance between expenditures of those in and out of Patient Centered Medical Homes could be attributed to the difference in study populations, pediatric versus adult. As men tioned above, pediatric populations tend to have more health care needs met which could indicate that children, on the whole, consume the same level of health care services. The difference between children in a Patient Centered Medical Home and those that are not, is the coordination and quality of those services. Another recently published study also appears to have results that are consistent to the results of this study. Flottemesch et al. evaluated the results of clinical medical home scores to health expenditures for a insured population(Flottemesch, Fontaine, Asche, & Solberg, 2011) Their study found that Patient Centered Medical Home scores were insignificantly associated in changes to total expenditures ( $75/person,
85 1.1%). The study also found that Patient Centered Medi cal Homes appeared to have significant decreases in total expenditures ($2,378, 4.4%) for individuals that have complex conditions and require 11 or more prescriptions. The results from Flottemesch can be viewed as anecdotal in support of the results tha t are seen in this study. The Flottemesch study appears to support the view that Patient Centered Medical Homes are effective in controlling and lowering costs for specific populations with complex conditions, such as children with special health care nee ds on the case of that specific study individuals requiring 11 or more prescriptions Consistent with the results of this study, extrapolating the effects of Patient Centered Medical Homes to general populations appear to show less effectiveness in controlling health care expenditures. The Flottemesch study does have some issues in trying to infer a relationship between both studies. The first issue, consistent with other Patient Centered Medical Homes studies, is the definition of a Patient Centered Medical Home. In this particular study, the author used the Physician Practice Connections Readiness Survey (PPCRS ) to identify Patient Centered Medical Homes. The PPCRS tool similar to the NCQA tool scores a practice based upon a practice response to survey questions intended to measure the Patient Center Medical Home domains. The score is scaled from 0 100. Unlike the NCQA tool, the PPCRS does not have a threshold in which a plan is qualified as a PCMH. Nor does the PPCRS require su pported documentation to validate survey responses. The use of the PPCRS tool in this study raises the potential issue that practices included in this study as medical homes would not be considered medical homes in study done in this paper. Another
86 is sue with making inferential statements based on the results of this study, is that Flottemesch does show that PPCRS scores were associated with a change in expenditures, albeit very small. One can make an argument that despite the small size in the difference, there is a difference. Overall, it is difficult to make strong statements in regards to the results of this study and the results of current published studies. As Patient Centered Medical Homes continue to be implemented in various pilot projects across the nation, the number of studies evaluating the effectiveness on general populations will increase. At the time of this writing, the studies that have been reviewed in this section appear to loosely support the results of this analysis. Individuals that are healthy or, in the case of the reviewed studies, do not have chronic complex conditions are not positively effected by Patient Centered Medical Homes in relation to expenditures and utilization. Th e one study that also used the Medical Expenditure Panel Survey showed that individuals in Patient Centered Medical Homes (defined by their criteria) had a greater probability of utilizing health care services as well as incurring health care expenditures. While these studies support the results shown in my analysis, one can only c laim a weak association due to the inherent differences between the published studies and this study. Policy Implications The use of Patient Centered Medical Homes as a policy initiative to help improve the efficiency of the United States health care system is rapidly growing steam (Miller, Crabtree, Nutting, Stange, & Jaen, 2010; Nutting et al., 2009) Conceptually, the Pa tient Centered Medical Home model attractive, as a conceptual model because its core premise is to del iver care that is timely, coordinated and consistent with patient needs. In this manner unnecessary care and therefore excessive cost s are contained.
87 In addition, patients have higher satisfaction and overall quality of care improves. Viewed in this li ght, Patient Centered Medical Homes could be one of the most powerful tools used against the fight of skyrocketing health care expenditures. Unfortunately, the use of this tool does not come without costs. There are several challenges to the implementati on and maintenance of Patient Centered Medical Homes. These costs are both tangible (dollars) and intangible (business process changes) non monetary costs and suggests some sort of cost benefit analysis should be considered before implementation. The r esults of this study imply that the full benefits of Patient Centered Medical Homes are perhaps not worth the costs taken in context of the general population. The most glaring policy concern with Patient Centered Medical Homes related to any study is the lack of a consistent and accepted definition. Perhaps this is reflective of the fragmented health delivery system in this country. The NCQA definition has com e closest to this standardization. This is perhaps more a function of their facility accrediting role and being the early entrant into this field. The NCQA definition is not without controversy. Several studies have highlighted issues with how the NCQA defines and scores practices as Patient Centered Medical Homes. These studies draw attention to the scoring emphasis of particular domains (technology ), which may do little to actually provide the intended benefits of Patient Centered Medical Homes (Clarke, Tseng, Brook, & Brown, 2012; Reid, et al., 2009; Strickland et al., 2009) The challenge is to translate a broad, conceptual definition to a concrete working one. Studies such as this one play an important role in helping to refine and narrow what characteristics of Patient Centered Medical Homes are important. The work done
88 in this study helps to complete the operational picture of a PCMH. The results would indicate that Patient Centered Medical Homes should emphasize characteristics that would e ffectively manage the utilization of health care resources of healthy populations. Healthy populations in PCMHs appear to seek more services because they are in an environment that perhaps encourages this behavior. The establishment of a universal operat ional definition of a Patient Centered Medical Home may never be implemented from a policy standpoint due to the fragmented nature of the US health care system (Takach, 2011) The work in this study and studies lik e this will help if and when policy makers have the ability to address issues such as these. Another policy implication related to this study involves the significant costs in establishing a practice as a Patient Centered Medical Home. Even looking past t he issue of what exactly a PCMH should look like, practices are challenged monetarily and operationally when they undertake this transf ormation. Recently, Patel, et al. evaluated Horizon of New Jerseys development of Patient Centered Medical Homes. They found that even though Horizon committed to changes in payment structures for Patient Centered Medical Homes, significant nonmonetary support was needed to ensure success (Patel, Rathjen, & Rubin, 2012) Some examples of nonmonetary support include changing the intake of patients, identifying community resources and developing new communication protocols with patients. The practices that cho o se to implement a Patient Centered Medical home must change the way they operate in many ways. The installation of an electronic medical record (EMR), while not necessary, is usually the first step for many practices. Practices though have to change the business process of entering data, how they evaluate that data, and how they share t hat data with their
89 patient populations among other things All these are significant challenges with nonmonetary costs associated with the implementation of Patient Centered Medical Homes Policy makers should fully understand these costs and what benefits are derived in implementing a Patient Centered Medical Home. The results from this study should give pause to those making policy decisions. The results would indicate a more precise imp lementation of Patient Centered Medical Homes versus a broad one. From a policy standpoint, Patient Centered Medical Ho mes could be more effective operating as Centers of Excellence for chronically ill populations. The Centers of Excellent concept could be attractive for several reasons. As a Center, it would be easier to clearly define and implement specifically what is a Patient Centered Medical Home. It would also be easier to adjust compensation to recognize the additional infrastruct ure required to support PCMHS. As research continues in this area, the picture would indicate that costs spent in transforming practices into Patient Centered Medical Homes would be better focused on specific chronically ill populations versus the practices that serve b road populations. This study and some of the other studies cited earlier in this section show that when applied to a broad general population, Patient Centered Medical Homes might have a diminished effect in regards to expenditures particularly compared to PCMH effects on chronically ill populations The expenditures of the chronically ill, such as diabetics and asthmatics might decrease, but these decreases would be offset by higher expenditures from healthy individuals. It should be noted however, that although this study may suggest a diminished impact from an expenditure standpoint, policy makers should not overlook the impact from a patient satisfaction standpoint. While this study did not evaluate patient satisfaction, there are several studies that support the
90 view of Patient Centered Medical Homes leading to increased overall patient satisfaction with their care (A. Beal, Hernandez, & Doty, 2009; Palfrey et al., 2004; Reid, et al., 2009) The use of Centers of Excellence is one of several potential policy approaches given the results of this study. One additional approach that could be taken is to stratify provider patient panels based upon chronic condition. Patients that are in the chronic condition categ ory would fall into a model of care that would incorporate most if not all of the Patient Centered Medical Home concepts. Patients that do fall into the chronic condition category would receive their primary care in a model that is less developed as a Pat ient Centered Medical Home. The stratification of the patient panel will help prevent the potential excess utilization and expenditures that is predicted in this study, while providing the cost containment and quality of care for chronically ill populations that is well documented in other studies. The challenge of this approach is that provider practices still would have to contend with many of the costs associated with the implementation of Patient Centered Medical Homes. In particular, the use of Elec tronic Medical Records to help stratify and identify chronic illness in the patient population would need to be implemented. The incorporation of this policy would have to be combined with some support for these implementation costs. Perhaps a more feas ible approach is to leverage existing public health functions with elements of the Patient Centered Medical Home. The public health providers could utilize their population viewpoint and incorporate some elements of the Patient Centered Medical Home, such as care coordination. Primary care provider practices would not need to implement all the elements of a Patient Centered Medical Home. Instead,
91 primary care providers would be tasked with identifying individuals in their patient panels that would benefi t from participating a public health Patient Centered Medical Home. Low income, chronically ill individuals would benefit the most from this public health model. In several ways, this approach dovetails with the traditional public health mission as stewa rds of population health. The potential issue with this approach, as with any public health initiative, is funding. The public health approach would need additional funding for the implementation of a national or at a minimum regional EMR. The EMR is th e only effective mechanism to monitor and effectively manage a large population. In addition, the public health approach would need to develop some compensation mechanism from insurers to recognize this important care delivery role to contain costs and im prove care. Nonetheless the public health approach is attractive since the numerous primary care provider practices would not need to transform themselves to PCMHs but rather only the few public health districts. Comparison of Patient Centered Medical Hom e Definitions The study utilized two different methods of identifying Patient Centered Medical Homes. One definition used a broad range of characteristics that Patient Centered Medical Homes may exhibit. The prevalence of 5 of these characteristics ident ified a Patient Centered Medical Home. The other definition used 5 specific characteristics of PCMHs. All five characteristics needed to be present to identify a Patient Centered Medical Home. Evaluating the model used in this study with both definitions yielded the same results. Regardless of the definition used, the model predicted that healthy individuals in Patient Centered Medical Homes would have higher expenditures and utilization than healthy individuals that are not in Patient Centered Medical Homes.
92 Study Limitations There are several limitations related to this study. The most significant limitation of this study is related to the data source and the identification of a Patient Centered Medical Home. The Medical Expenditure Panel Survey does not identify respondents as being in a Patient Centered Medical Home, nor does the survey ask specific questions to identify PCMH characteristics. As a result, this study had to infer the existence of a Patient Centered Medical Home based upon respondent self reported information. This limitation raised two issues. The first issue is that since MEPS does not specifically identify Patient Centered Medical Homes in their survey, we are limited in the Patient Centered Medical Home criteria used to identify their existence. The us e of electronic medical records is a significant component of PCMHs for example, yet there is nothing in MEPS that allows us to identify this characteristic of a provider. The utilization of the MEPS data set limits the study in evaluating the complete Patient Centered Medical Home definition. The results of this study could be inconsistent with results that might be obtained if we were able to include all the components of a Patient Centered Medical Home in the analysis. The second issue that arises in inferring the existence of a PCMH for a respondent is the self reported data. The use of self reported data introduces several potential biases. The individual reporting this data may have a recall bias in which they are unable to effectively or accurately recall the information that is being asked. There are also issues when the respondent may incorrectly answer the question due to their lack of fully understanding the question being asked to them Finally, self reported data is also subject to a potential response shift Response shift occurs when survey questions are asked at multiple points in time. At these different points, respondents can change their internal standards, values or
93 conceptualization of the questions being asked to them (Sprangers & Schwartz, 1999) The changes that can occur over time can lead to a shift or difference in how they might answer a particular question, or a response shift. The response shift can lead to a bias in the respondent answers and therefore could influence the identification of Patient Centered Medical Homes in this study. The studys reliance on self reported information is another potential limitation in this study. The study has attempted to evaluate the impact of Patient Centered Medical Homes on healthy populations. The study is a cross sectional or point in time study. The utilization of this approach is another limitation in this study. The results of this study would have a stronger impact if we could follow healthy individuals across a period of time, or a longitudinal study. The health status of individuals can and do change over time. Patient Centered Medical Homes could have a significan t effect on expenditures as these healthy individuals transition into a sick state. The effectiveness of Patient Centered Medical Homes (regardless of definition) in managing chronically ill populations has been generally established in current research. The utilization of a longitudinal study would go a long way in establishing if the benefits of a Patient Centered Medical Home accrue over time. The accrual of these benefits could make a stronger policy argument for the broad implementation of Patient C entered Medical Homes. The crosssectional analysis of this study limits the strength of the results on the effects of Patient Centered Medical Homes on healthy populations. The self reported determination of health status is not the only limitation aro und health status and this study. Further complicating the results of this study is the potential influence of chronic conditions and health status. The SF 12 form reports an
94 individuals perception of health status. An individual that has a well manage d chronic condition may feel healthy if the chronic condition does not infer with their quality of life. The classification of healthy individuals in this study may include some individuals that fall in this category; well managed chronically ill. The in clusion of chronically ill individuals in the healthy classification could have the impact of influencing the results of higher expenditures and utilization events. An alternative to determining health status could utilize diagnosis codes from claims data. The use of diagnosis codes to identify chronically ill individuals does not completely resolve this issue either, as it is dependent upon some type of claim event that captures that diagnosis. The final limitation of this study is the determination of health status. The study uses the SF 12 score to determine if the respondent is healthy or not. Several issues arise with the use of the SF 12 score to determine healthy individuals. The first issue that arises has already been discussed in regards to the identification of Patient Centered Medical Homes and self reported data. The SF 12 uses self reported data and therefore is subject to the same limitations outlined in that discussion. Another issue in using the SF 12 instrument is the instrument its elf. The SF 12 is an instrument that was developed usi ng Classic Test T heory or CTT. CTT has shortcomings in that the items that it attempts to measure ( i.e. health status ) are not sample invariant and CTT is test driven rather than item driven. The fact that the SF 12 is not sample invariant means that the sample used in the study affects the statistics of the item being measured. The SF 12 is test based versus item based, which means that different mixes of questions will provide different esti mates of a persons ability. Current methods in health outcomes research have moved away from instruments utilizing
95 Classic Test Theory to instruments using Item Response Theory (IRT). The Patient Reported Outcomes Measurement Information System (PROMIS) is the current health outcomes instrument to utilize IRT, which addresses the shortcomings of the SF 12 and other CTT based instruments. Regardless of the limitations outlined above, this study still contributes research related to Patient Centered Medi cal Homes. Perhaps more importantly, this study opens new areas of research related to PCMHs. The majority of research on Patient Centered Medical Homes has focused on the impact on chronically ill populations. The interest in this area is rightly deser ved, as the chronically ill populations are the highest consumers of health care services. Researchers however have an obligation to provide a complete picture of any research question. The picture of Patient Centered Medical Homes is not complete without evaluating its effects on the general population, including healthy individuals. The information presented in this study helps to further complete the picture of Patient Centered Medical Homes. The increased understanding of PCMH s will help policy make rs better utilize this tool in its battle against rising health care expenditures and lower quality. The work done in this study is important for these reasons despite the limitations presented here. Future Research The investigation completed in this paper points to some future research possibilities. Potential future research can help address some of the limitations highlighted in this paper. The two most pressing opportunities related to this research is conducti ng a study with a full Patient Centered Medical Home definition and a longitudinal analysis of the effects of PCMHs.
96 The lack of a consistent definition of a Patient Centered Medical Home has been frequently highlighted in this paper. A comprehensive st udy incorporating all the domains of PCMHs will be critical for future research. A study in this area will help identify and distinguish the domains of Patient Centered Medical Homes. The research completed to date only evaluates the effects of PCMHs wit hout specifically looking at these domains. It is important to understand the effects of each individual domain in addition to the interaction effects between the domains. Research in this area will help policy makers and providers better understand wher e they should focus their efforts. The longitudinal analysis of populations in Patient Centered Medical Homes is a nother area of future research. An analysis of this type would help to improve our understanding of the effects of Patient Centered Medical Homes on healthy populations. Additional research such as this would help us understand the long term effects of PCMHs. Healthy individuals in this environment may experience improved outcomes and reduced expenditures as they transition from healthy to sick. If research were to show results such as this, it would make a compelling argument that Patient Centered Medical Homes are an effective tool to improve outcomes and lower health care expenditures. Similar to many financial investments that return dividends over time, Patient Centered Medical Homes could be viewed in this light if an appropriate longitudinal analysis was completed. The further specification of utilization types, such as facility, ER, office visits, etc and the impact of Patient Cen tered Medical Homes on those categories is another area of potential future research. Research on this topic will provide clarification on the full effect of Patient Centered Medical Homes on healthy populations. Patient Centered
97 Medical Homes may be mor e effective at controlling certain categories of expense or utilization. Understanding which categories that may be positively influenced by Patient Centered Medical Homes will refine potential policy choices of this model of care. The story is not complete on Patient Centered Medical Homes. The ability to improve patient experiences, control utilization and expenditures and manage the well being of the whole person is an intriguing concept. It is obvious that in particular situations Patient Centered Me dical Homes can deliver some or all of these things. Further research is needed to better understand under what care delivery characteristics and populations this concept can be fully leveraged to our benefit. It is the authors hope that future research can shed additional light on these questions.
98 LIST OF REFERENCES Assurance, N. C. f. Q. (2008a). Standards and Guidelines for Physician Practice Connections Patient Centered Medical Homes (PPC PCMH) National Committee for Quality Assurance. Washington DC. Assurance, N. C. f. Q. (2008b). Standards and Guidelines for Physician Practice Connections Patient Centered Medical Homes (PPC PCMH) CMS Version National Committee for Quality Assurance. Washingt on DC. Beal, A. (2007). Closing the divide: How Medical Homes promote equity in health care: Results from the Commonwealth Fund 2006 Health Care Quality Survey. Commonwealth Fund Beal, A., Hernandez, S., & Doty, M. (2009). Latino access to the patient centered medical home. J Gen Intern Med, 24 Suppl 3, 514520. doi: 10.1007/s1160600911191 Berenson, R. A., Hammons, T., Gans, D. N., Zuckerman, S., Merrell, K., Underwood, W. S., et al. (2008). A house is not a home: keeping patients at the center of p ractice redesign. Health Aff (Millwood), 27(5), 12191230. doi: 27/5/1219 [pii] 10.1377/hlthaff.27.5.1219 [doi] Bethell, C. D., Read, D., & Brockwood, K. (2004). Using existing populationbased data sets to measure the American Academy of Pediatrics definition of medical home for all children and children with special health care needs. Pediatrics, 113(5 Suppl), 15291537. Blough, D. K., Madden, C. W., & Hornbrook, M. C. (1999). Modeling risk using generalized linear models. J Health Econ, 18(2), 153171 doi: S01676296(98)000320 [pii] Boyd, C. M., Shadmi, E., Conwell, L. J., Griswold, M., Leff, B., Brager, R., et al. (2008). A pilot test of the effect of guided care on the quality of primary care experiences for multimorbid older adults. J Gen Intern Med, 23 (5), 536 542. doi: 10.1007/s1160600805299 Brotherton, S. E., Rockey, P. H., & Etzel, S. I. (2005). US graduate medical education, 20042005: trends in primary care specialties. JAMA, 294 (9), 10751082. doi: 294/9/1075 [pii] 10.1001/jama.294.9.10 75 Chandran, L., Gusic, M., Baldwin, C., Turner, T., Zenni, E., Lane, J. L., et al. (2009). Evaluating the performance of medical educators: a novel analysis tool to demonstrate the quality and impact of educational activities. Acad Med, 84(1), 5866. doi : 10.1097/ACM.0b013e31819045e2 0000188820090100000022 [pii]
99 Clarke, R. M., Tseng, C. H., Brook, R. H., & Brown, A. F. (2012). Tool used to assess how well community health centers function as medical homes may be flawed. Health Aff (Millwood), 31(3), 627 635. doi: hlthaff.2011.0908 [pii] 10.1377/hlthaff.2011.0908 CMS. (2007, 4/14/2011). Details for Medicare Medical Home Demonstration, from https://www.cms.gov/demoprojectsevalrpts/md/itemdetail.asp?itemid=cms11992 47 Conrad, D., Fishman, P., Grembowski, D., Ralston, J., Reid, R., Martin, D., et al. (2008). Access intervention in an integrated, prepaid group practice: effects on primary care physician productivity. Health Serv Res, 43(5 Pt 2), 18881905. doi: HESR880 [pii] 10.1111/j.14756773.2008.00880.x Cooley, W. C., McAllister, J. W., Sherrieb, K., & Clark, R. E. (2003). The Medical Home Index: development and validation of a new practicelevel me asure of implementation of the Medical Home model. [Research Support, NonU.S. Gov't Validation Studies]. Ambul Pediatr, 3(4), 173 180. Devries, A., Li, C. H., Sridhar, G., Hummel, J. R., Breidbart, S., & Barron, J. J. (2012). Impact of medical homes on quality, healthcare utilization, and costs. Am J Manag Care, 18(9), 534 544. doi: 78615 [pii] Donabedian, A. (1966). Evaluating the quality of medical care. Milbank Mem Fund Q, 44(3), Suppl:166206. Duan, N. (1982). A Comparison of alternative models for the demand for medical care. Santa Monica, CA: Rand Corporation. Farmer, J. E., Marien, W. E., Clark, M. J., Sherman, A., & Selva, T. J. (2004). Primary care supports for children with chronic health conditions: identifying and predicting unmet family needs. J Pediatr Psychol, 29(5), 355 367. Fields, D., Leshen, E., & Patel, K. (2010). Analysis & commentary. Driving quality gains and cost savings through adoption of medical homes. Health Aff (Millwood), 29(5), 819 826. doi: 29/5/819 [pii] 10.1377/hlthaff.2010.0009 Flottemesch, T. J., Fontaine, P., Asche, S. E., & Solberg, L. I. (2011). Relationship of clinic medical home scores to health care costs. J Ambul Care Manage, 34(1), 7889. doi: 10.1097/JAC.0b013e3181ff6fd2 00004479201101000000 10 [pii] Fuller, R. L., Clinton, S., Goldfield, N. I., & Kelly, W. P. (2010). Building the affordable medical home. J Ambul Care Manage, 33(1), 71 80. doi: 10.1097/JAC.0b013e3181c9fca9 0000447920100100000014 [pii]
100 Grossman, M. (1972). On the concept of h ealth capital and the demand for health. The Journal of Political Economy, 80(2), 223 255 %@ 00223808. Hall, M. N., & Kruse, J. (2010). Allies in family medicine advocacy: the patient centered primary care collaborative. [News]. Ann Fam Med, 8(1), 89 90 doi: 10.1370/afm.1085 Harman, J. S., Scholle, S. H., Ng, J. H., Pawlson, L. G., Mardon, R. E., Haffer, S. C., et al. (2010). Association of Health Plans' Healthcare Effectiveness Data and Information Set (HEDIS) performance with outcomes of enrollees with diabetes. Med Care, 48(3), 217223. doi: 10.1097/MLR.0b013e3181ca3fe6 Homer, C. J., Klatka, K., Romm, D., Kuhlthau, K., Bloom, S., Newacheck, P., et al. (2008). A review of the evidence for the medical home for children with special health care needs. [Research Support, U.S. G ov't, P.H.S. Review]. Pediatrics, 122(4), e922937. doi: 10.1542/peds.20073762 Institute of Medicine (U.S.). Committee on Quality of Health Care in America. (2001). Crossing the quality chasm : a new health system for the 21st century Washington, D.C.: National Academy Press. Jeffe, D. B., Whelan, A. J., & Andriole, D. A. (2010). Primary care specialty choices of United States medical graduates, 19972006. [Comparative Study Research Support, NonU.S. Gov't]. Acad Med, 85(6), 947 958. doi: 10.1097/ACM.0b013e3181dbe77d Joint principles of the Patient Centered Medical Home. (2008). Del Med J, 80(1), 21 22. Jones, A. (2000). Health Econometrics In: Culyer, A., Newhouse, J. Handbook of Health Economics Kane, D. (2007). Access & coverage. Hospitals help patients find a medical home. Hosp Health Netw, 81(3), 28. Kaye, N. T. (2009). Building Medical Homes in State Medicaid and CHIP Programs. 114. Retrieved from http://www.commonwealthfund.org/Publications/FundReports/2009/Jun/Building Medical Homes In State Medicaidand CHIP Programs.aspx#citation Manning, W. G. (1998). The logged dependent variable, heteroscedasticity, and the retransformation problem. J Health Econ, 17(3), 283295. doi: S01676296(98)000253 [pii] Manning, W. G., Basu, A., & Mullahy, J. (2005). Generalized modeling approaches to risk adjustment of skewed outcomes data. J Health Econ, 24(3), 465 488. doi: S01676296(05)00005 6 [pii] 10.1016/j.jhealeco.2004.09.011
101 Manning, W. G., & Mullahy, J. (2001). Estimating log models: to transform or not to transform? J Health Econ, 20(4), 461494. doi: S0167629601000868 [pii] Maxfield, M., Centers for, M., Medicaid Services Office of Research, D., Information, & Mathematica Policy Research, I. U. h. b. g. c. b. i. o. z. (2008). Design of the CMS Medical Home Demonstration: Draft : Mathematica Policy Research, Incorporated. McCarthy, D. Geisinger Health System: Achieving the Potential of System Integration Through Innovation, Leadership, Measurement, and Incentives (Vol. 9). McCarthy, D (2009). Group Health Cooperative: Reinventing Primary Care by Connecting Patients with a Medical Home (Vol. 18). Miller, W. L., Crabtree, B. F., Nutting, P. A., Stange, K. C., & Jaen, C. R. (2010). Primary care practice development: a relationshipcente red approach. Ann Fam Med, 8 Suppl 1 S68 79; S92. doi: 8/Suppl_1/S68 [pii] 10.1370/afm.1089 Mullahy, J. (1998). Much ado about two: reconsidering retransformation and the twopart model in health econometrics. J Health Econ, 17(3), 247281. doi: S01676296(98)000307 [pii] NCQA (2008). Standard and guidelines for Physician Practice Connections Patient centered medical home. Nutting, P. A., Miller, W. L., Crabtree, B. F., Jaen, C. R., Stewart, E. E., & Stange, K. C. (2009). Initial lessons from the f irst national demonstration project on practice transformation to a patient centered medical home. Ann Fam Med, 7(3), 254 260. doi: 7/3/254 [pii] 10.1370/afm.1002 Palfrey, J. S., Sofis, L. A., Davidson, E. J., Liu, J., Freeman, L., & Ganz, M. L. (2004). T he Pediatric Alliance for Coordinated Care: evaluation of a medical home model. Pediatrics, 113(5 Suppl), 15071516. Patel, U. B., Rathjen, C., & Rubin, E. (2012). Horizon's Patient Centered Medical Home Program Shows Practices Need Much More Than Paymen t Changes To Transform. Health Aff (Millwood), 31(9), 2018 2027. doi: 31/9/2018 [pii] 10.1377/hlthaff.2012.0392 Paulus, R. A., Davis, K., & Steele, G. D. (2008). Continuous innovation in health care: implications of the Geisinger experience. Health Aff (Millwood), 27 (5), 12351245. doi: 27/5/1235 [pii] 10.1377/hlthaff.27.5.1235
102 Pless, I. B., Satterwhite, B., & Van Vechten, D. (1978). Division, duplication and neglect: patterns of care for children with chronic disorders. Child Care Health Dev, 4 (1), 9 1 9. Pugno, P. A., McGaha, A. L., Schmittling, G. T., DeVilbiss, A., & Kahn, N. B., Jr. (2007). Results of the 2007 National Resident Matching Program: family medicine. Fam Med, 39(8), 562 571. Ralston, J. D., Carrell, D., Reid, R., Anderson, M., Moran, M., & Hereford, J. (2007). Patient web services integrated with a shared medical record: patient use and satisfaction. J Am Med Inform Assoc, 14(6), 798 806. doi: M2302 [pii] 10.1197/jamia.M2302 Reid, R. J., Coleman, K., Johnson, E. A., Fishman, P. A., Hsu, C., Soman, M. P., et al. (2010). The group health medical home at year two: cost savings, higher patient satisfaction, and less burnout for providers. Health Aff (Millwood), 29 (5), 835843. doi : 29/5/835 [pii] 10.1377/hlthaff.2010.0158 Reid, R. J., Fishman, P. A., Yu, O., Ross, T. R., Tufano, J. T., Soman, M. P., et al. (2009). Patient centered medical home demonstration: a prospective, quasi experimental, before and after evaluation. Am J Manag Care, 15(9), e71 87. doi: 11745 [pii] Rittenhouse, D. R., & Shortell, S. M. (2009). The patient centered medical home: will it stand the test of health reform? JAMA, 301 (19), 20382040. doi: 301/19/2038 [pii] 10.1001/jama.2009.691 Rittenhouse, D. R., T hom, D. H., & Schmittdiel, J. A. (2010). Developing a policy relevant research agenda for the patient centered medical home: a focus on outcomes. J Gen Intern Med, 25(6), 593 600. doi: 10.1007/s11606010 1289x Romaire, M. A., Bell, J. F., & Grossman, D. C. (2012). Health care use and expenditures associated with access to the medical home for children and youth. Med Care, 50(3), 262269. doi: 10.1097/MLR.0b013e318244d345 Sia, C., Tonniges, T. F., Osterhus, E., & Taba, S. (2004). History of the medical ho me concept. Pediatrics, 113(5 Suppl), 14731478. Sprangers, M. A., & Schwartz, C. E. (1999). Integrating response shift into healthrelated quality of life research: a theoretical model. Soc Sci Med, 48(11), 15071515. doi: S0277953699000453 [pii] Stang e, K. C., Nutting, P. A., Miller, W. L., Jaen, C. R., Crabtree, B. F., Flocke, S. A., et al. (2010). Defining and measuring the patient centered medical home. J Gen Intern Med, 25(6), 601612. doi: 10.1007/s11606010 12913
103 Starfield, B., & Shi, L. (2004). The medical home, access to care, and insurance: a review of evidence. Pediatrics, 113(5 Suppl), 14931498. Strickland, B. B., Singh, G. K., Kogan, M. D., Mann, M. Y., van Dyck, P. C., & Newacheck, P. W. (2009). Access to the medical hom e: new findings from the 20052006 National Survey of Children with Special Health Care Needs. Pediatrics, 123(6), e996 1004. doi: 123/6/e996 [pii] 10.1542/peds.20082504 Takach, M. (2011). Reinventing Medicaid: state innovations to qualify and pay for patient centered medical homes show promising results. Health Aff (Millwood), 30(7), 13251334. doi: 30/7/1325 [pii] 10.1377/hlthaff.2011.0170
104 BIOGRAPHICAL SKETCH Daniel J. Estrada was born and raised in Tampa, FL. with an older brother and sister. Daniel completed his undergraduate studies at the University of Florida in industrial e ngineering in 1992. After a brief period of time, he returned to the University of Florida. Influenced by his exposure to the healthcare field from his father Daniel completed his joint master program to obtain his Masters in Business Administration and Masters in Health Science at the University of Florida in 1998. He worked several years in industry as a healthcare consultant. His experience in this career provided him with broad exposure to regional variations in healthcare and the financing of healthcare services through the insurance industry. Family considerations brought Daniel off the road and led to his pursuit of a PhD in Health Services Research at his alma mater He is particularly interested in econometric analysis and the financial impact of healthcare policy decisions on disadvantaged populations.