This item has the following downloads:
1 HEALTHCARE ACCESS, UTILIZATION, AND ENROLLMENT PATTERNS OF UNITED STATES VETERANS By ROBERT PAUL FLEMMING A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIRE MENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2013
2 2013 Robert Paul Flemming
3 This dissertation is dedicated to my wife, Christine Shin Chia Lee, and my first born son, Aiden Paul Flemming, who lit the fire.
4 ACK NOWLEDGMENTS I thank my wife, Dr. Christine Shin Chia Lee, for her steadfast commitment to our dream. I am eternally proud of our son, Aiden Paul Flemming, who has already begun teaching the world the most fundamental and timeless I thank my parents, Robyn and James Flemming, as well as my mother and father in law, Henry and Mali Lee, for their patience and guidance. To my lifelong mentors, Father Robert Weber and Ron Bell I am eternally grateful of their lessons in core va lues This research would not be made possible without the support of my distinguished committee members, Dr. R. Paul Duncan Malcom and Christine Randall Professor, HSRMP Department Chair and Committee Chair, Dr. Diane Cowper Ripley Associate Director o f the Gainesville Rehabilitation Outcomes Research Center, and Co Chair, as well as Dr. Allyson Hall, Associate Professor and MHA Program Director, Dr. Jessica Schumacher Assistant Professor, and Dr. Chris McCarty Associate Professor, Director of the Bur eau of Economic and Business Research and external committee member friendship through this entire process. I also thank the wonderful people at the Center for Medicare an d Medicaid Innovation as well as friends and neighbors, Eugene Andraca and Iris Wei, I owe a special thank you to the Delores Auzen ne F oundation for awarding me their dissertation grant in 2012. Finally, and with great resp ect, let us never forget the immense sacrifices our Veterans and the families of our Veterans have made in defense of our Democratic freedoms.
5 TABLE OF CONTENTS ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF TABLES ................................ ................................ ................................ ............ 7 LIST OF FIGURES ................................ ................................ ................................ .......... 9 LIST OF ABBREVIATIONS ................................ ................................ ........................... 10 ABSTRACT ................................ ................................ ................................ ................... 13 CHAPTER 1 INTRODUCTION ................................ ................................ ................................ .... 15 2 BACKGROUND AND SIGNIFICANCE ................................ ................................ ... 28 Rationale ................................ ................................ ................................ ................. 29 Reason one: The VHA IDS is uniquely positioned to deliver care to Veterans ................................ ................................ ................................ ........ 29 Reason two: There is a large and growing diversity in the Veteran population ................................ ................................ ................................ ...... 30 Reason three: Veterans continue to experience ATC issues ........................... 30 Reason four: Prior research was not representative of the full Veteran population ................................ ................................ ................................ ...... 32 Reason five: Veteran research should address the full spectrum of access to care ................................ ................................ ................................ ........... 33 Reason six: The significance and importa nce of enrollment has changed ....... 35 Reason seven: Demand based forecasts may be refined ................................ 36 Reason eight: The value of the VHA IDS m ay be re defined ............................ 38 Reason nine: Improved access to data ................................ ............................ 39 Reason ten: The development of research methods ................................ ........ 40 Contribution to Literature and Policy Implication ................................ ..................... 44 3 LITERATURE REVIEW ................................ ................................ .......................... 55 The A ndersen Behavioral Model Framework ................................ .......................... 55 Barriers to Healthcare Access ................................ ................................ ................ 58 VHA Studies on Barriers to Healthcare Access ................................ ................ 62 Public industry studies ................................ ................................ ...................... 74 Summary ................................ ................................ ................................ ................ 86 Private Industry Studies ................................ ................................ .......................... 87 Summary ................................ ................................ ................................ ................ 93 4 THEORETICAL FRAMEWORK AND HYPOTHESES ................................ ............ 98 Overview ................................ ................................ ................................ ................. 98
6 Specific Aims ................................ ................................ ................................ ........ 101 Research questions ................................ ................................ .............................. 107 5 DATA AND METHODS ................................ ................................ ......................... 109 Source of Data and Availability ................................ ................................ ............. 110 Dependent Variables ................................ ................................ ............................ 111 Independ ent Variables ................................ ................................ .......................... 113 Study Design and Methodology ................................ ................................ ............ 126 Empirical models ................................ ................................ ................................ .. 129 6 RESULTS ................................ ................................ ................................ ............. 1 32 Descriptive Findings for Research Question One ................................ ................. 132 Modeling Veteran Characteristics on Enrollment Status ................................ ....... 133 Descriptive Findings for Research Questions 2A and 2B ................................ ..... 134 Modeling Veteran Characteristics on Utilization ................................ .................... 135 Research Question 2A Results ................................ ................................ ............. 136 Research Question 2B Results ................................ ................................ ............. 139 Limitations ................................ ................................ ................................ ............. 142 7 CONCLUSIONS ................................ ................................ ................................ ... 194 Discussion ................................ ................................ ................................ ............ 194 Implications ................................ ................................ ................................ ........... 196 Future Research ................................ ................................ ................................ ... 198 APPENDIX A ITEM MEASURES ................................ ................................ ................................ 200 B DIFFERENCES IN MISSING V ARIABLE DISTRIBUTIONS ................................ 211 LIST OF REFERENCES ................................ ................................ ............................. 213 BIOGRAPHICAL SKETCH ................................ ................................ .......................... 220
7 LIST OF TABLES Tab le page 1 1 Priority Group Rating Descriptions ................................ ................................ ..... 21 3 1 Dependent and Independent Variables ................................ .............................. 61 3 2 Independent Variables ................................ ................................ ........................ 61 6 1 Dependent Variable for Research Question One ................................ ............. 150 6 2 Predisposing Characteristics of Veterans ................................ ......................... 151 6 3 Enabling Characteristics of Veterans ................................ ................................ 152 6 4 Needs based Characteris tics of Veterans ................................ ........................ 153 6 5 Dependent Variable for Research Question 1 ................................ .................. 154 6 6 Predisposing Characteristics of Enrolled and Non enrolled Veterans .............. 155 6 7 Enabling Characteristics of Enrolled and Non enrolled Veterans ..................... 157 6 8 Needs Based Characteristic s of Enrolled and Non enrolled Veterans ............. 158 6 9 Predisposing Odds of Enrolling into the VHA IDS (N = 20,367,420) ................ 159 6 10 Ena bling Odds of Enrolling into the VHA IDS (N = 20,367,420) ....................... 160 6 11 Needs based Odds of Enrolling into the VHA IDS (N = 20,367,420) ................ 161 6 12 Dependent Variable for Research Questions 2A and 2B ................................ .. 162 6 13 Predisposing Characteristics of ER Use Among Veterans ............................... 163 6 14 Enabling Characteristics of ER Use Among Veterans ................................ ...... 165 6 15 Needs Based Characteristics of ER Use Among Veterans .............................. 166 6 16 Predisposing Characteristics of Outpatient Use Among Veterans .................... 167 6 17 Enabling Characteristics of Outpatient Use Among Veterans .......................... 169 6 18 Needs Based Characteristics of Outpatient Use Among Veterans ................... 170 6 19 Predisposing Characteristics of Inpatient Use Among Veterans ...................... 171 6 20 Enabling Characteristics of Inpatient Use Among Veterans ............................. 173
8 6 21 Needs Based Characteristics of Inpatient Use Among Veterans ...................... 174 6 22 Dependent Variable for Research Questions 2A and 2B ................................ .. 175 6 23 Predisposing Odds of ER Use for Research Question 2A ................................ 176 6 24 Enabling Odds of ER Use for Research Question 2A ................................ ....... 177 6 25 Needs Based Odds of ER Use for Research Question 2A ............................... 178 6 26 Predisposing Odds of Outpatient Use for Research Question 2A .................... 179 6 27 Enabling Odds of Outpatient Use for Research Question 2A ........................... 180 6 28 Needs based Odds of Outpatient Use for Research Question 2A .................... 181 6 29 Predisposing Odds of Inpatient Use for Research Question 2A ....................... 182 6 30 Enabling Odds of Inpatient Use for Research Question 2A .............................. 183 6 31 Odds of Inpatient Use for Research Question 2A ................................ ............. 184 6 32 Predisposing Odds of ER Use for Research Question 2B ................................ 185 6 33 Enabling Odds of ER Use for Research Question 2B ................................ ....... 186 6 34 Needs based Odds of ER Use for Research Question 2B ............................... 187 6 35 Predisposing Odds of Outpatient for Research Question 2B ............................ 188 6 36 Enabling Odds of Outpatient Use for Research Question 2B ........................... 189 6 37 Needs based Odds of Outpatient Use for Research Question 2B .................... 190 6 38 Predisposing Odds of Inpatient for Research Question 2B .............................. 191 6 39 Enabling Odds of Inpatient Use for Research Question 2B .............................. 192 6 40 Needs based Odds of Inpatient Use for Research Question 2B ...................... 193
9 LIST OF FIGURES Figure page 2 1 Enrollment Flowchart for Discharged Veteran ................................ .................... 53 2 2 ................................ ....... 54 4 1 Original ABM Theoretical Framework ................................ ............................... 100 4 2 Modified Conceptual Model ................................ ................................ .............. 100 4 3 Specific Aim One Categorical Relationships ................................ .................... 104 4 4 Specific Aim Two Categorical Relationships A) ER Use B) Outpatient Use C) Inpatient Use ................................ ................................ ................................ 106 6 1 Operationalized Enabling Variables under Research Question 2A ................... 147 6 2 Operationalized Enabling Variables under Research Question 2B ................... 148 6 3 Odds Ratios of Insurance Categories ................................ ............................... 149
10 LIST OF ABBREVIATIONS ABM Andersen Behavioral Model ABS Address based S ampling ACC Aggregated Condition Categories AD Active Duty ADLs Activities of Daily L iving ATC Access to C are CAHPS Consumer Assessment of Heath Plans Study CBO Congressional Budget Office CDC HCC Centers for Disease Control Hierarchical Condition Category CHIP Children s Health Insurance Prog ram CONUS Continenta l United S tates D HHS Department of Health and Human Services DoD Department of Defense DV Dependent Variable DWMCC Deployed Warrior Medical Management Center EDES Expedited Disability Eva luation System EHCPM Enrollee Health Care Projection Model EHR Electronic Health Record ER Emergency Room FFS Fee For Service FY Fiscal Year (FY) GAO Go vernment Accountability Office GMT G eographic Means Test HCO Healthcare Organization
11 HIE Health I nsurance E xperiment HMO Health Maintenance Organization IADLs Instrumental A ctivities of Daily L iving IDES Integrated Disability Evaluation System IDS Integrated De livery System IV Independent Variable MBP Medical Benefits Package MCO Managed Care Organization MHA Master s in Health Administration MHS Military Health System MLR Multivariate Logistic R egression MTF Military Treatment Facility NFP Not For Profit NSV National Survey of Veterans OCONUS Outside the Continenta l United States OEF Operation Enduring Freedom OIF Operation Iraqi Freedom OOP Out of Pocket PCM Primary Care Model PG Priority Group POS Point of Service POW Prisoner of War PPACA P atient Protection and Affordable Care Act PSA P rimary Service A rea PV Predictor V ariable
12 QBP Qualifie d Medicare Beneficiary Program RCT Randomized Control Trial SCDR Service Connected Disability Rating TPMRC Theater Patien t Movement Requirements Cen ter VA Veterans Affairs VERA Veterans Equitable Resource Allocation VHA Veterans Health Administration VISN Veterans Integrated Service Network VistA Veterans Health Information Systems and Technology Architecture
13 Abstract of Dissertation Presented t o the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy HEALTHCARE ACCESS, UTILIZATION, AND ENROLLMENT PATTERNS OF UNITED STATES VETERANS By Robert Paul Flemmin g August 2013 Chair: R. Paul Duncan Co chair : Diane C Cowper Ripley Major: Health Services Research Within the past few years, mi llion s of non enrolled Veterans became eligible for care under the V (VHA) fully integrate d nationwide healthcare system following the implementation of several eligibility expansion policies. In a diverging manner t hese changes have led to an unprecedented change in demand on the VHA system. On the aggregate Veterans continue to age and dema nd greater resources for care as a result of co morbidities. Augmenting this demand are a growing number of younger Veterans returning from conflicts in the Middle East and requir ing a heightened level of rehabilitative services Coupled with this social p henomena are several remarkable forces impacting enrollment and utilization to include but, not limited to ; r ising healthcare c osts, a deep recession, and sweeping changes within the healthcare industry brought on by the 2009 Patient Protection and Afforda ble Care Act (PPACA) The Dep artment of Veterans Affairs (VA) has historically presented data and research on utilization patterns within the enrolled population but, has only recently
14 begun to explore the characteristics and needs of the entire Veteran population, inclusive of eligible non enrolled Veterans By investigating plausible causes of enrollment variations more Veterans may gain better access to care (ATC) This research represents a n important first step in understanding how Veterans will l ikely use available healthcare resource options within and external to the Veterans Health Administration Integrated Delivery System (VHA IDS) and describes which Veterans are most likely to enroll into the VHA IDS based on individual level characteristics The Andersen behavioral model of healthcare utilization (ABM) is used as the theoretical basis by which to assess likely associations between independent and dependent variables. The study has found a statistically significant difference between enrolle d and non enrolled Veterans in their comprehension of VHA benefits and available community health resources after adjusting for important selected factors categorized within the ABM There were also statistically significant differences between enr9lled a nd non enrolled Veterans in the observed associations between predisposing, enabling, and needs characteristics and their health care utilization, including specifically their use of the ER, outpatient care, and inpatient care, both within and external to the VHA.
15 CHAPTER 1 INTRODUCTION The United States has a long history of commitment to its military Veterans. This is plainly evident from the days preceding hrough the Revolutionary War, to the present day. P rior to the foundi ng of the United States, English colonies provided pensions to disabled V eterans under the first Colonial pension law enacted in 1636 in Plymouth, MA. The purpose of t his legislation was to provide monetary compensation to Veteran s who defended the Colony from Native Americans and became disabled in its defense ( VA 2006) Although these pensions would not be typically th ought of as health insurance in modern times, this law became the first of its kind in the New World to commit to the general health and welfare of Veterans. Years after the US Cons titution was drafted another seminal piece of legislation was passed by the Fifth Congress in 1798; An Act for The Relief of Sick and Disabled Seamen (United States Congress, 1798) The Act effectively authorized Congress to collect a mandatory tax on the wages of Merchant Marines for two purposes; t o fund the first socialized model of health insurance in America and to fund constru ction of hospitals and other infrastructure to care for sick and disabled seamen. Throughout the 19 th Century, the Naval Home, care of all disabled Veterans. T wo of the three hospi tals were located in Washington, DC and the third, the Naval Home in Gulfport, Mississippi. President Lincoln continued to advance the national commitment towards developing Veteran resources and caring for Veterans and their families in his famous second inaugural address following the Civil War (VA, 2012) This vision prompted the creation of the National Asylum for Disabled Volunteer Soldiers and in
16 turn, the creation of large homes where r oom, board, and medical care were provided. During World War I, the Veterans Bureau was enacted to manage the agencies of three institutions; the Bureau of War Risk Insurance which p rovide d insurance policies for Veterans, the Public Health Service which delivered healthcare to Veterans and monitored public health, and the Federal Board of Vocational Education which delivered occupational health services to Hoover that Congress created the Veterans Administration by consolidating three separate bureaus under one um and the National Homes for Disabled Volunteer Soldiers. Among the multitude of services offered, medical services were considered to be one of the most valuable among Veterans who experienced combat con ditions ( VA 2006) Following WWII, the cou commitment to Veterans had extended greater than ever with the establishment of the V A hospital system and its medical compensation program for disabilities Related legislation and policy aimed at providing health care coverage and the delivery of services were primarily focused on the consolidation of efforts within Federal institutions Institutions such as the Department of Defense (DoD) military health care system (MHS) and the V eterans Health Administration (V HA ) became the two principal orga nizations from which Veterans were given choice of access to among Federal healthcare services. Remarkably the MHS and the VHA offered duplicate yet, varying degrees of coverage for Veterans. The reason for the duplication and variable accessibility is r ooted in t he MHS primary mission to provide coverage and care for the active duty (AD) personnel. An extension of this mission applies to the care of AD dependents, as well as in varying
17 degrees, military Reservi sts, National Guard personnel, military ret irees and their dependents (DoD, 2012) In essence, AD personnel are entitled to health care services through a Militar y Treatment Facility (MTF) or referral based trea tment at an alternate civilian institution and, generally, have greater priority over others in obtain ing care (Burrelli & Library of Congress Congressional Research, 2001) AD dependents, as well as military retirees, Reservists, the National Guard, and retirees and their dependents are also eligible for care at military DoD institutions but, receive care on a space available basis as determined by local policy (TRICARE, 2012) To be clear, the focus of this research is on Veterans. Since military retirees are Veterans, these individuals are by definition eligible to o btain care from the MHS based on regional capacity constraints and the discretion of the military installation Commander (DoD, 2012) When considering the variability o f eligibility determinations through localized policies access for Veterans is often limited within this area Along the same lines, and a common theme throughout this research has not always equa te d to to healthcare r esources. Furthermore, always As a result and evident throughout the history of the US ATC limitations p resent a common recurring obstacle among Veterans in need of healthcare The rationale follows that localized MHS and VA policies with restricted access provide a control mechanism by which the DoD and its military branches could refocus capacity as needed particularly during times of war or ongoing combat contingencies These types of loc al policies actually date back to the early 20 th Century in response to earlier failed attempts by the Public Health Service to provide care fo r Veterans in
18 military hospitals The practice of referring Veterans to the MHS was primarily borne out of necess ity During this time little to no VHA infrastructure hospitals, or other Veteran specific healthcare resources were built to meet the demands of this group As a result, capacity within the MHS was often filled to maximum within MTFs and Veteran s were f orced to compete with active duty military personnel for scarce healthcare resources The result was that Veterans generally lost in this proposition and were, at best, given limited care ( VA 2006) As a means of providing some relief to Veterans in need of care, local MTFs were empowered to open i ts doors when mission demands and operations tempo were considered. There is an important distinction between the accessibility of DoD M TFs and the VHA. Clearly, military retirees are Veterans who are both eligible and entitled to benefits within the VHA so long as their discharge from service was not dishonorable ( VHA 2007 ) With a lack of guaranteed coverage in the MHS, one may surmise that the VHA is the Veteran s when given the choice among the two alternatives This becomes the large r question of this research. A vast majority of payment for their care, payer. As such, how do we predict utiliz ation patterns given the multitude of options? From what we already know, the characteristics of enrolled Veteran s vary to a great er extent in relation to their decision s t o use care from others sources (Cowper, 2005). enrolled Veterans might respond given the same set of options.
19 The basis for enrollment variations is both quantifiable and contextual. Firstly, there is considerable wariness among older Veterans that the VHA represents a lesser quality alternative in comparison to other healthcare systems ( Jha, Perlin, Kizer and Dudley, 2003 ; Kizer and Dudley, 2009; Oliver, 2007) This perception likely stems from the way the VHA was positioned prior to the transformation of the delivery system by the former Undersecretary for Health, Kenneth Kizer, in 1995 ( Kizer, et al. 2009 ; Oliver, 2007) Prior to these institutional changes the VHA focused almost exclusively on inpatient and long term nursi ng home care and provided what many considered to be care ( Kizer, et al. 2009 ) Furthermore there are several known ATC barriers relating to the geographical distance to healthcare facilities, the d egree of general and mental health status an insurance status, in come level as well as predetermined individual level characteristics, and a host of other influential dynamics which were explored throughout this research ( Wooten, 2002; Cowper, 2004; DeVoe, 2007; Fontana & Rosenheck, 2006; Himmelstein, et al. 2007; Hisnanick, 2000; Hoff & Rosenheck, 1998; Hynes, et al. 2007; Liu, Maciejewski, & Sales, 2005; McCarthy, et al. 2007; Nelson, Starkbaum, & Reiber, 2007; Petersen, et al. 2010; Shen, Hendricks, Wang, Gardner, & Kazis, 2008; Yano, Washington, Goldzweig, Caffrey, & Turner, 2003a) While this list of barriers is not exhaustive the intent is twofold; firstly, to highlight some of the factors wh ich are already well researched in the area of ATC and utilization and secondly, to provide a foundation for this research on ATC barriers, enrollment and utilization within an IDS An additional barrier that was consider ed in this research is the awarene ss of health benefits within the VHA. This is important to gauge in light of changing enrollment policies over the past several years
20 possible misunderstanding of health resource options. (Hughes, 2003 ; Shen, et al. 2008; Yano, Simon, Lanto, & Rubenstein, 2007) From the 1950 s through the turn of the Century, Veterans would become eligible for VA services based on a complex set of criteria that established the level of care administered for service c onnected disabilities (United States Dept. of Veterans, 2006) Care was primarily delivered within an inpatient setting and the VHA was structured to deliver these services among Veterans with service connected disabilities only. until the late 20 th Century that Congress passed the Veteran s' Health Care Eligibility Reform Act of 1996 which empowered the VHA to create its Medical Benefits Package (MBP) ; a standardized health benefits plan which made care available to all enrolled V eterans This MBP enhanced the service delivery of the VHA wi th a focus on preventive and primary care within outpatient and inpatient settings (Congress, 1996) In line with the newly created MBP the VHA established priorities for enrollment year budget established by Congress. For reference, t he definitions of the Priority Group (PG) ratings are provided in more detail in Table 1 1 (VA, 2012 ) As of June 2009 all enrollment restrictions were lifted for any Veteran that had served honorably As with the future of anything, change is only the constant. With resources becoming limited at the time of this writing there is always the possibility that PGs run the risk of being reset as a means of control ling enrollment and protecting the capacity to deliver quality care among their current beneficiaries (Cross, 2009)
21 Table 1 1. Priority Group Rating Descriptions Priority Group Description 1 Veterans with VA Service connected disabilities rated 50% or more. Veterans assigned a total disability rating for compensation based on unemployability. 2 Veterans with VA Service connected d isabilities rated 30% or 40%. 3 Veterans who are former POWs. Veterans awarded the Purple Heart Medal. Veterans awarded the Me dal of Honor. Veterans whose discharge was for a disability incurred or aggravated in the line of duty. Veterans with VA Service connected disabilities rated 10% or 20%. Veterans awarded special eligibility classification under 4 Vete rans receiving increased compensation or pension based on their need for regular Aid and Attendance or by reason of being permanently Housebound. Veterans determined by VA to be catastrophically disabled. 5 Nonservice connected Veterans and non compensab le Service connected Veterans rated 0%, whose annual income and/or net worth are not greater than the VA financial thresholds. Veterans receiving VA Pension benefits. Veterans eligible for Medicaid benefits. 6 Compensable 0% Service connected Veterans. V eterans exposed to ionizing radiation during atmospheric testing or during the occupation of Hiroshima and Nagasaki. Project 112/SHAD participants. Veterans who served in the Republic of Vietnam between January 9, 1962 and May 7, 1975. Veterans who served in the Southwest Asia theater of operations from August 2, 1990, through November 11, 1998. Veterans who served in a theater of combat operations after November 11, 1998, as follows: Veterans discharged from active duty on or after January 28, 2003, for five years post discharge
22 Table 1 1. Continued Priority Group Description 7 Veterans with incomes below the geographic means test (GMT) income thresholds and who agree to pay the applicable copayment. Veterans with gross household incomes above the VA national income threshold and the geographically adjusted income threshold for their resident location and who agrees to pay copays 8 Veterans eligibility for enr ollment: Non compensable 0% service connected and: Subpriority a: Enrolled as of January 16, 2003, and who have remained enrolled since that date and/ or placed in this subpr iority due to changed eligibility status. Subpriority b: Enrolled on or after June 15, 2009 whose income exceeds the current VA National Income Thresholds or VA National Geographic Income Thresho lds by 10% or less Veterans eligible for enrollment: Nonservice connected and: Subpriority c: Enrolled as of January 16, 2003, and who remained enrolled since that date and/ or placed in this subpriority due to changed e ligibility status. Subpriority d: Enrolled on or after June 15, 2009 whose income exceeds the current VA National Income Thresholds or VA National Geographic Income Thresholds by 10% or less Not eligible Veterans not eligibility for enrollment: Veterans not meeting the criteria above: Subpriority e: Non compensable 0% servic e connected Subpriority f : Nonservice connected Note that PG ratings referenced at the top of the table align with a hierarchical system which establishes higher precedence of enrollment within pre defined categories ranging from one to eight. PG categories are generally awarde d to Veterans who se need for medical care derive from a standardized percentage assigned to a specific service c onnected injury or disability (SCD) or to Prisoners of War (POWs) but, may also include Veterans without a SCDR. Of note, is that Veterans with a SCDR of 10% or
23 greater account ed for approximately one third of the enrolled population of Veterans in 2009 (Percy and Elmendorf, 2009) Those in PG 5 accounted for another one third of enrolled population and referenced those that were considered very low income. Lo wer precedence is generally assigned to Veterans who are in need of care but, whose condition is not service connected and are determined by a standard f ederally quantified financial threshold as not being able to afford private sector alternatives ( VHA 2007 ) These PG categories are generally referenced from five through eight. Prior to the 1996 reforms, eligibility provisions were highly complex and inconsistent across all levels of care. Over the last fifteen years, the standardization of the MBP and enrollment policies on a national scale had a widespread effect. Following the full within the VHA expanded from 4.2 Million enrollees in Fiscal Year (FY) 1999 to 8,499,160 enrolled Veterans as reported FY 2011 ( VA Research, 2010 ; Congressional Budget Offi ce, 2007). Historically, most Veterans were largely considered eligible to obtain some form of inpatient or nursing home care but, were never guarantee d ATC (Burrelli & Library of Congress Congressional Research, 2001) Variability in ATC to services was geographic specific where the number of Veterans made it difficult to impossible for t he VHA to provide care to every eligible individual. Consequently, the priority system was used to determine which Veterans would have access to which type of care (Ashton, 1999; McCarthy, et al. 2007; Penrod, 2001) In several situations, the result was that VHA services were, in effect, not available to many Veterans.
24 In June 2009 the VHA committed to a new enrollment model, enti tling all honorably discharged V eterans and certain qualified sp ouses to he althcare within the V H A system. The priority system remains in place but is now primarily used to make determinations on the level of out of pocket (OOP) payments expected from V eterans, rather than enrollment eligibility As mentioned before, t his could change depending on the level of available funding from year to year. By categorizing the level of services the VHA employs a standardized method to establish the level of paid services rendered at the VHA. For instance, a Veteran in PG 1 would not be responsible for any out of pocket (OOP) expenses relating to services rendered at the VHA. Conversely, a Veteran in PG 8 would be required to pay OOP for some portion of services rendered at the VHA ( VHA 2007 ) Furthermore, as the PG rating determinatio n moves from two through eight the expected level of OOP payment adjusts on a sliding scale towards more Veteran cost sharing. The immediate and long term impact of change s in enrollment policies are not fully understood and continue to be studied throug hout the organization (VHA, 2009) Perhaps the most obvious set of unanswered questions rev olve around how many and which Veterans actually en roll or otherwise take up the opportunity to identify the VHA as a source for their health care In 2010 it was estimated that 22.2 million Veterans were eligible for the VHA care under the new framework, and approximately 8.5 million of those eligible had elected to enroll in the system (NCVAS, 2011; Panangala, July 27, 2010; Westat 2010 ) The circumstances of the remaining 1 3 .7 million non enrolled Veterans, and how they differ from the enrollees, are un clear at this time
25 One may postulate that there are several reasonable possibilities which may help answer this question. For instance, i t is entirely possible that some V eterans had a full and open opportunity to reflect on the question of enrollment and made a deliberate decision to obtain their heal th care in the private sector. O thers may have deferred their decision with the thinking t hat t hey would consider the option of enrolling in to the V H A system when and if they perceived that their n eed s could b e b etter met within the VHA a t a later time. Still o thers may not have been aware that they had the option to enroll in the first place This latter category of Veterans becomes a special topic of interest given the present political, economic, and social env ironment and the continual roll out of the PPACA legislation. A more thorough understanding of which Veterans enroll and which Veterans remain non enrolled would be of great policy value to the VA as it continues to seek ways to allocate resources in an e ffective and efficient manner for a growing number of eligible Veterans I mproving the understanding of the enrollment decision is a critical first step and thus, represents one of the primary objective s of this dissertation. Furthermore, a n explicit effo rt has been undertaken by the VA to better understand why Veterans choose to enroll or choose not to enroll into the VHA This endeavor is based on the Veterans Benefits Improvement Act of 2004, P.L. 108 454, Section 805 mandating that the Secretary of Ve available under laws As a result of this Congressional mandate, the National Survey of Veterans (NSV 2010) expand e d the scope of its
26 survey design to explore the challenges Veterans may be facing in accessing their entitlements to healthcare within the VHA. P.L. 108 454 is being implemented at a corresponding time following the latest expansion of enrollment. This is important to note considering the vast changes that t ranspired throughout the VHA For more than a decade, the VHA has been implementing changes to its enrollment system by expanding eligibility in stages; from permitting the select inclusion of Veterans based on SCD and/or certain financial means test threshold s to encompassing any Veteran who s erved honorably as well as certain qualified spouses of Veterans ( VHA 2009) As mentioned previously all Veterans who h ad been honorably discharged from service were given the option to obtain care from the VHA however; the accessibility of care was limited by the capacity available to accommodate the healthcare needs of all honorably discha rged Veterans. Extensive c hanges to the enrollment system began in 1999 in accordance with provisions of the Veterans Health Care Eligibility Reform Act of 1996 (Public Law 104 262; 110 Stat. 3177). The law set in motion a series of enrollment expansion po licies which, to present include providing entitlements for Veterans of all pre defined PG s to access V H A healthcare resources. The enrollment expansion was implemented by the V H A under an amendment to P L 104 262, effectively permitting the voluntary e nrollment of all honorably discharged Veterans Veterans whose income exceeds the current means test and (Cross, 2009) Title 38, United States Code of Federal Regulation, Section 17.36, expressly empowers the Secretary of Veterans Affairs to establish this enrollment sy stem. Consequently t he PG
27 rating continues to be based on the disability rating, fina ncial status, having an honorary discharge from service, and other unique and specific inclusions of health services (Hynes, et al. 2007) Of further interest to the V H A are the effects on service demand forecasts and the associated costs of doing business. The V H A employs the use of an actuarial model known as the Enrollee Health Care Projectio n Model. This model corresponds with the Veterans Equitable Resource Allocation (VERA) system which uses a capitated payment model by which to divide fund s among twenty one geographically designated Veterans Integrated Service Network s (VISN s ) based on the volume and Priority Group ratings of enrollees (Yano, et al. 2007). The combination of these systems provides a means of forecasting financial demands within the entire VHA in every fiscal year. Veterans are interested in understanding the effects of en rollment expansion but, for very different reasons than the VHA Firstly, PG ratings affect Veterans in terms of the level of OOP expenses and past enrollment eligibility With the enrollment expansion policies, these ratings principally serve as a means o f categorizing the level of services offered at the VHA. For instance, a Veteran in PG 1 would not be responsible for any OOP expenses relating to services rendered at the VHA. Conversely, a Veteran in PG 8 would be required to pay OOP for some portion o f services rendered at the VHA ( VHA 2007 ) Veterans and Veteran advocacy groups remain principally concerned over the level of future OOP payments required to gain access to health services within the VHA and whether eligibility among the non enrolled Veterans ma y change in light of future funding limitations ( Assistant Secretary of DAV, 2009 )
28 CHAPTER 2 BACKGROUND AND SIGNIFICANCE most recent enrollment inclusion policy set forth an unprecedented level of change within its nationw ide system of facilities. These changes entitle PG 8 and any h ono rably discharged Veterans, certain qualified spouses and certain children of Veterans to healthcare within any of its 1, 7 00 points of entry ( VA, 2013; Cowper, 2011; Cross, 2009) Consequently the VHA is facing a n emerging issue in how best to serve a growing divers ity of current and future Veterans with unique ATC barriers and varying levels of aware ness of their entitlement to VHA healthcare benefits ("Veterans Benefits Improvement Act of 2004," 2004) With the exception of some VHA and affiliate organization sponsored outreach pro grams, very little attention has been given to non enrolled Veterans with persistent limitations to gaining access into the VHA. Since the second inaugural speech by President Abraham Lincoln, Veterans g for their social and health needs. reform law, the 2009 Patient Protection and Affordable Care Act (PPACA), there has been a revived interest in identifying health needs within the U. S. population. These two historical happenings intersect at a time of high national unemployment rates, high rates of uninsured, an aging population, rising healthcare costs, and a cornerstone component of the PPACA legislation mandate requiring insurance coverage among an unprecedented number of Americans. Fortunately, American Veterans have been provided with some opportunity to secure little to no cost, comprehensive health insurance protection and care within the Veterans Health
29 Administration Integrate d Delivery System (VHA IDS). Regrettably and for largely inexplicable reasons to date, Veterans around the nation rema in non enrolled and in need of medical treatment yet, continue to experience barriers with ATC through the VHA IDS or other public and pr ivate healthcare resources This dissertation lays out the foundation by which to explore probable reasons driving these issue s at a unique period of healthcare transformation To understand the significance of the issue of enrollme nt within the VHA IDS we need to address the individual level, organizational level, and environmental influences affecting enrollment and utilization. This rational e is further highlighted below and provides greater clarity to the importance of conductin g this research. Rationale Reason one: The VHA IDS is uniquely positioned to deliver care to Veterans The VHA IDS is the largest, fully integrated, closed healthcare delivery system and payer of health services in the United States ( NCVAS, 2007 ; VA, 2013 ) Although the institution retains elements of a traditional health maintenance organization (HMO), its funding mechanisms and beneficiary pop ulation are highly unique The exclusiveness of its beneficiary population permits the institution to maintain control over enrollment and coverage and with a high degree of continuity of care. Additi onally, t he system caters to a select subgroup of the American population, is fully funded by taxpayer dollars, and represents one of the leaders in industry wide, healthcare delivery and payment transformation. The VHA IDS continues to evolve its organiza action to deliver better value to US Veterans and other eligible beneficiaries as well as
30 to US taxpayers Subsequently the institution exemplifies the vanguard for comprehensiv e delivery among a nationwide network of integrated care for Veterans find value in how individuals approach the ACA individual health insurance mandate, market exchange s, participation in ACOs, and other market changes. Reason two: The re is a large and growing diversity in the Veteran population T he VHA IDS is responsible for offering care to a large group of individuals residing throughout the nation and whose needs a re highly variant. Among the 22 million living Veterans, t here are p resently over 8.5 million Veterans enrolled in the VHA IDS (NCVAS, 2011; Panangala, July 27, 2010; Westat 2010 ) There is some level of uncertai nty to the fate of the remaining 1 3.7 million non enrolled Veterans who may be experiencing limited access to the care they need through the VHA IDS or any other available healthcare resource (et., al) As dual ove rseas combat contingencies in Iraq and Afghanistan come to an end, the Veteran population is expected to grow and broaden in demographic diversity and medical needs. Consequently, a fuller assessment of what may be driving potential ATC issues prior to enr ollment and continuing throughout the full spectrum of healthcare delivery is necessary Reason three : Veterans continue to experience ATC issues Arguably, the periods before and during the enrollment phase are the most crucial to gaining realized ATC within any healthcare delivery system (Eisenberg, 2000) Without enrollment into a health plan, individuals are generally left to their own devices to pay for care out of pocket. As costs continue to rise in the delivery of healthcare, uninsured and underi nsured individuals will often forego receiving necessary care (IOM,
31 2002) In turn, there is an escalation in the volume and cost of care as untended healthcare issues worsen over time Veterans, in particular, are faced with numerous ATC barriers through out the value chain of delivery of high quality healthcare. The reasons vary from overarching challenges endemic within the system to the demographic makeup of the population. From an organizational perspective, the VHA IDS has undergone several unpreceden ted changes in its eligibility regulations within a relatively short amount of time. Previously, the VHA was largely an exclusive inpatient care institution capable of serving only the most indigent Veterans with specific service connected disabilities. In a little more than a decade the VHA transformed itself into an all encompassing national IDS which serves any honorably discharged Veteran regardless of income or service connected disability rating (SCDR) (Kizer & Dudley, 2009) As with most change s in healthcare this transformation created greater responsibility within the VHA IDS as its customer base expands and diversifies Using an individual level lens Veterans are likely experiencing varying eligibility inclusiveness. Consequently, Veterans may not be aware of their h ealthcare benefits and forego enrollment into the system. This becomes a larger issue for the vulnerable population of Veterans and any Veteran in need of healthcare who remains non enrolled and chooses to forego treatment. Since the VHA IDS mandates that enrollment precedes the delivery of care, it would stand to reason that a general lack of awareness of benefits would likely prevent enrollment in the presence of
32 need. This issu e is then exasperated by a myriad of other barriers to access being experienced by this group and should be explored further in greater detail. ATC is an important issue for all Veterans regardless of service connected disability or financial circumstances perceived declination in health status or a newly evaluated condition which demands ongoing care. Similarly speaking, the Veteran may be prevented from attaining care when the support system is misaligned wit h these needs. This study presents a case for assessing these sets of variables in a novel fashion using a sound theoretical basis supported by several decades of related literature. Reason four: Prior research was not representative of the full Veteran p opulation Veteran research has been severely limited in its assessments of the whole Veteran population as a result of its dependence on enrolled Veteran data. The fact that numerous Veterans self report a higher level of satisfaction in their patient exp erience and treatment within the VHA IDS presents only one perspective of the Veteran population (Perc y and Elmendorf, 2009) It would be easy for the reader to surmise that enrollees are likely experiencing excellent care as a result of the operational superiority over others, its long history of supporting the Veter an and its high level of interdependence with the military. While these perceptions may prove to be entirely accurate, we need to gain the perspective of the non enrolled Veteran population to truly understand possible drivers of enrollment and utilization variability within the VHA IDS The unmistakable gap in the literature on non enrolled Veteran perspectives is striking. T o date, we have limited information detailing perceptions of the VHA, their knowledge of VHA entit lements, nor what drives
33 enrollment and utilization collectively What is presented in the literature appears to be a stovepipe interpretation of these separate issues with r eference to only the enrolled population. By considering the full population of Ve terans through the inclusion of enrolled and non enrolled data, as well as their unique needs and utilization patterns, we may establish a more inclusive conclusion from which to improve upon the quality of the ongoing ATC discussion. Reason five : Veteran research should address the full spectrum of access to care To advance our understanding on why the combination of enrollment and utilization of health services is a central issue for all Veterans, we may consider the within the U S healthcare delivery system. This concept is borrowed from the fields of electrical engineering and physics. Customarily voltage drops are equated with inefficiencies found within all electronic circuitry and emerge when energy is wasted within various m odes and channels of delivery. Each voltage drop causes a reduction in power at any measured point. At the point (i.e. a light bulb, blender or any other device requiring electrical energy), there is a collective loss of power from th ese voltage drops and lower overall optimal output. Eisenberg (2000) transplanted the concept of voltage drops to the healthcare industry as a means of drawing upon important inferences between system wide efforts to deliver high quality healthcare to ind ividuals and chronological inefficiencies found at different periods of attempted ATC Analogous to voltage drops transpiring in electrical circuitry on account of inefficiencies in hcare industry value chain. The consequence of this power loss may be equated to a
34 systematic exclusion of individuals at each stage of entry into the healthcare system and an accumulative failure to deliver high quality healthcare to patients in need at t he end point If we think of the source of power residing at the outset within the availability of insurance coverage, voltage drops will occur chronologically at each stage of the delivery process. In total, Eisenberg presented seven domains where voltage drops may be occurring; prior to gaining access to available insurance, enrollment in available insurance plans, access to covered services, clinicians, and health care institutions, having a choice of plans, clinicians, and health care institutions, havi ng access to a consistent source of primary care, having access to referral services, and receiving the delivery of high quality health care services (Eisenberg, 2000). This dissertation draws heavily from this concept of voltage drops and provides a nove l approach to understanding the full spectrum of access and delivery to high quality healthcare within the VHA IDS. At the heart of this research is the intent to address barriers to enrollment along the value chain of delivery. The focus was on how the ef fects of Veteran s awareness of health benefits affect enrollment into the VHA IDS and the delivery of care within the primary outpatient care setting, the inpatient setting, and the ER. The novelty of this research stems from an emergent need to understa nd healthcare enrollment within a changed statutory environment. By including enrollment within behavioral theory, a fuller context of the healthcare delivery system is achieved and narrows an apparent gap within the plethora of applied research on healthc are utilization. Accordingly, by combining available enrollment
35 and utilization research into one area of inquiry, our general comprehension of the problem faced by Veterans is considerably enriched. Reason six: The significance and importance of enrollm ent has changed The timing of this writing presents some explanation for the presence of this gap in the literature More specifically, w e need to consider that enrollment was not always as critical an issue as what it has become today. Out of the passag e of the PPACA 2009 arose a requirement for individuals to acquire some form of insurance by the year 2014 or risk a tax penalty for non compliance. In 2012, the individual mandate element of the PPACA was being challenged by state opponents at the Supreme Court level While controversial, this center piece of the legislation was upheld as be ing constitutional under the power of the Federal government to levy taxes ( 648 F. 3d 1235 2012 ) Fortunately, the expansion of eligibility to all honorably discharged Veterans in 2009 coincides with these political changes and a unique opportunity exists for this population to comply with the law once fully implemented. According to the Department of Health and Human Services (2013) Veterans enrolled in the VHA IDS ar e considered covered by health insurance and will be exempt from paying the individual mandate penalty but, will not be eligible for a tax credit if other insurance is purchased through the health insurance exchange ( DHHS, Dec 2012; CMS Webinar April 5, 2 013 ) Furthermore, enrolled Veterans will retain access to care within the VHA nationwide network without pay ing monthly premiums to maintain coverage. Many enrolled Veterans will continue to receive care at no cost while others will have an obligatio n to share some form of costs dep endent on their level of income, SCD, period of ser vice, and several other factors.
36 Reason seven : D emand based forecasts may be refined From an organizational lens this expansion multiplies the scope of the V H mission to c are for more eligible Veterans. The VHA is now in a position where it must address how it attracts a larger and more diverse group of Veterans into its IDS This is in opposition to the previous structure of the VA healthcare institution which focused on the delivery of inpatient care which based admission into its system on a complex set of criteria dealing with SCD ratings and strict G eographic Means Test ing (GMT) based on a standardized financial threshold formula (VA, 2013) As the VHA IDS transfor med its structure it also reformed its processes D emand based forecasting was one area which fundamentally changed to meet the requirements of a new structure of care delivery In the modern VHA IDS, forecasting must account for utilization variance withi n its enrolled population as ev idence of differences between those that use and those that do not use the system (Cowper, 2005). VISNs that are successfully account ing for these differences are optimizing the globa l capitation system of paym ent (Kizer, 2009) As mentioned previously, VERA is used to forecast service demand throughout the VISN. In line with this structure, th e VHA IDS benefit s from a genera l increase in enrollment volume as well as a focused outreach to Vete rans Local VHA ins titutions which are likely to receive the most funding from the central budget office enroll Veterans with higher SCD ratings and Veterans who were deployed to Operation Enduring Freedom in Afghanistan or Operation Iraqi Freedom in Iraq (OEF and OIF, respe ctively) The VA central office justifies these different funding rates on two grounds Firstly, Veterans with higher SCD ratings are more likely to consume greater health resources and funding must offset the higher costs of care Secondly, Congress alloc ated greater funding for OEF/OIF
37 Veterans as a means of building capacity for mental health care resources ( Burnam et. Al., 2009) Consequently, VISNs which are the most efficient and effective in resource planning will secure greater resources for its re gional VHAs Through careful consideration of the diverse characteristics and needs of non enrolled Veterans, VISNs are able to attract a larger volume of enrollees while adjusting its capacity and resources in an optimal manner to serve the greatest need. The evolution of changes continues to effect the institution in other ways as the Iraq and Afghanistan conflicts rapidly come to an end Specifically, numerous military service members are being discharged from service and entering into Veteran status. A s more Veterans re integrate into the civilian world, the VHA IDS must prepare for greater demands on its rehabilitation departments on top of its other resource needs Some of these newer demands arise from what the VA and others have coined signatu re wounds OEF/OIF conflicts, post traumatic stress disorder and traumatic brain injury (VA, 2013; Tanielian, et a l., 2008). Other sources of demand which stem from these conflicts are care for amputations, vocational rehabilitation, as well as a host o f other health issues. Consequently, these demands translate into higher healthcare costs when compared to other historical military conflicts. The costs of care for wounded Veterans are perhaps far greater than any other war as a direct result of the ser vice estimated the killed to wounded ratio was 1:16 for the OEF/OIF conflicts. In comparison to WWI and WWII, that ratio was one death for every two wounded service members (VA, 2006).
38 Desp ite these rising costs, the VHA IDS has continued to function as a comprehensive safety net for Veterans presenting with a full spectrum of need s. This role includes caring for Veterans requiring everything from hospice care, to long term rehabilitation, t o the management of co morbi di ties, to the lowest levels of outpatient treatment throughout a While total cost of care is not a focus in this dissertation, there remains value in assessing enrollment and utilization data within this pop ulation and its possible impact on the VA budget Moreover we develop a critical understanding on the degree to which the VHA IDS may or may not be fulfill ing its mission to deliver high quality, comprehensive care to its beneficiaries Reason eight: T he value of the VHA IDS may be re defined Utilization studies must then evolve to meet the needs of these changing institutional, social, and political forces Earlier research provides a foundation from which to appreciate the choice of h ealthcare delivery options in the context of a shifting environment. Although changes within the healthcare environment may have fluctuated throughout time, the elements of these changes were arguably, similar and equally dramatic. At one point, Congress was very actively considering the dismantling the VA healthcare system, the only system of its kind which directly cares health needs. grown antiquated, was riddled w ith costs, and ultimately, did little to meet the needs of its beneficiaries. As a result, several studies were conducted to better understand the value of this enterprise and whether or not it should continue its operations.
39 These studies used patient an d organizational characteristics to effectively define and continuity of care patterns. As a result, this effort played a pivotal role in changing the public on the value of th e VHA. The importance of this research only grew with the transformation of the VA in under Kenneth Kizer. As the institution reformed to meet the greater needs of its customers through a comprehensive primary care model (PCM) the need fo teristics became more apparent. Likewise, these studies provided the V H A with a basis by which to justify the reorganization of its delivery system and improve the quality and ATC delivered throughout its system The relevancy of these studies remains p articularly germane to our understanding of how our healthcare delivery system continues to evolve and impact utilization patterns Building upon the application of these studies is a need to delve further into the u nderlying barriers to access to high quality healthcare. Reason nine: Improved access to data Fortunately, there are several novel reasons to expound upon past research as survey instruments continue to become refined and accessibility is improved through the widespread use of digital content. Greater context is provided through the comb ined availability of data on enr ollment, utilization, patient characteri stics, This heightened accessibility should not be consider ed coincidental. Analogous to the days prior to the transformation of the VHA into a fully functioning IDS, there is a renewed need to justify the value of the institution. As healthcare costs continue to rise in an uncertain environment t he availability of these data provide a basis by which to defend the value of the institution among taxpayers and any ensuing Congressional inquiry.
40 Reason ten: The development of research methods A s good research progresses within a similar area of interest, it is natur al for knowledge to advance as relevant issues are explored in greater depth using sound methods. P rior research forms the basis by which to draw c ontext ual inference with the present issue. With greater access to data, more sophisticated methods, and an e volving environment the quality of the discussion improves upon this foundation. To further illustrate, there are three relevant studies published within the past thirty years which employed the use of the National Survey of Veterans as a secondary data s ource Each of t hese studies inquired into Veteran s choice of the source of care but, employed the use of separate methods to explain utilization of healthcare services Pag e (1982 ) was the earlier of these studies and employed a log linear model to test the relationship among Veteran characteristics and their choices to use VA hospitalization and alternate external sources of care. Stepwise regression was used to identify the strongest correlations among the available independent and dependent variables. Based on this methodology, the author concluded that the primary factor affecting Veteran hospitalization within the VHA was the absence of other health insurance, controlling for age, income, and service connected disability status. There are obvious li mitations to this research. Firstly, stepwise regression is largely criticized for basing its conclusions on the strongest correlation of variables devoid of a theoretical model. Consequently, there is an over reliance on a single best model which is limit ed by available data and an ability to draw meaningful parameter estimations. Although useful for exploring completely new areas of
41 inquiry, the stepwise regressions methodology relies heavily on available data without genuine consideration for biased para meter estimations. This approach often leads to a misinterpretation of cause and effect in the absence of important omitted variables. For instance, if we were to ask whether smoking increases mortality, we may not find a strong correlation between these f actors because those with yellow fingers were mo re likely to die. Of course, if we think deeper into these associations, we might infer that individuals with yellow fingers may be the heaviest smokers and thus, became unhealthier as a direct effect. With r espect to study, there was a deficiency in responses from the non enrolled Veteran population As a result, the parameter estimations were likely biased by omitted variables from this population and altered the strength and direction of the conclusi ons inferred within the study Secondly, while providing an early basis for understanding utilization in the face of ATC barriers, we must refrain from overgeneralizing its results in the present day. More specifically, health insurance plans were not as multifaceted in the 1980s as they are today. The very paradigm of the health insurance payment structure only began to change after t he seminal RAND health insurance experiment (HIE) ( RAND, 1986 ) Considering that the RAND HIE was published close to the sa we could surmise that consumer cost One theory used extensively in this area of Veteran research is t he Andersen behavioral model of health services utilization (A BM) In a related study by Hibbard, et al. (1986), the authors inquired into whether predisposing, enabling,
42 and need factors were of equal importance in explaining utilization rates among younger and older HMO enrollees. Using these Veteran characteristic s as a basis for determining utilization patterns, this study expounded upon important differences in predisposing, enabling, and need characteristics between the older and younger groups. Hibbard, et al. (1986) also found that each of the comparison grou p s differed in the combination of factors driving variant utilization patterns within each population This research and others like this form a strong foundation for drawing similar inquiries in the line of Veterans utilization. Furthermore, Andersen, et al. point out that clearer contrast may be derived from how access to care is defined within a healthcare system We can think of equitable access o n one end of the spectrum as being driven by demographic characteristics and their need for heal thcare. On the other end is i nequitable access, principally defined by enabling factors driving the Veteran towards or away from the use of health resource s. The distinction in the type of access is highlighted for good reason here The contrast between th e two help s high quality healthcare in relation to the on civilian healthcare reform. H ow we define the equal treatment of individuals within either of these healthc are system s is clearly a discussion on how we value equity. Penchansky et al ., (1977) and Andersen, et al., (1978, 1995) previously health system. As the VHA strives to meet its mission to care for each Veteran in the most optimal manner, the reader must clearly understand how access ties into
43 organizational structure and processes. As a fully integrated delivery system, the VHA is distinctly interested in el iminating inequitable access and improving upon needs based access ibility within the population it serves. Moreover the IDS forms the structure by which access may be improved through meaningful processes Processes may be thought of as the vehicles by wh ich access improvements are realized. Actual advances in equitable access would thus be made through the processes Tracking back to the ten reasons used to justify this study, we need to consider that as societa l values changed, the structure and processes of the VHA IDS transformed to match a new definition of access. Cowper (2005) was arguably the choice of care using the ABM Her research advanced t he foundation of research by explaining the Veteran choice of care at an important crossroads of VHA history; its transformation into a fully integrated delivery system under Kenneth Kizer, then Under Secretary for Health (1994 1999). As stated previously, these organiza tional changes were directed at addressing health outcome s through a PCM. Since coordination of care is at the heart of the PCM, Dr. Cowper became of the VHA IDS in relation to other e xternal options Cowper furth er modified the original Anderse n model in a manner which addressed the choice of provider through utilization trends at the point of service (POS) By explaining significant differences in Veteran social behavioral characteris tics and needs Cowper was able to
44 demonstrate that coordination of care under the PCM is fundamentally hindered as Veterans access the system in diverse ways. The principal limitation a lack of available data on non enrolled Vet erans. At the time, the NSV (1999) was devoid of any data on a fully representative population of Veteran characteristics and use as a direct result of a yet, incomplete implementation of the enrollment system Respectfully, d with the initial decision to use care however, this restricted a comprehensive assessment on what drives the initial decision to use care external to or within the VHA IDS. By including non responses in the NSV 2009, we can now assess possible reasons driving the choice of care based on differences in predi sposing, enabling, and need characteristics among the non enrolled and enrolled population s Accordingly, t he theory of the ABM is expanded as more relevant individual level data is i ncorporated into t he model through the use of the NSV 2010 Contribution to Literature and Policy Implication There are several studies outside of the Veteran population that have focused on the relationships between socio economic factors, insurance cove rage, healthcare need, and other barriers and predictors of accessing healthcare resources (Borowsky & Cowper, 1999; Cowper, 2004; Hibbard, et al. 1986 ; Himmelstein, et al. 2007; Hisnanick & Gujral, 1996; Nelson, Gordon, & Reiber, 2007; Nelson, Starkbaum, et al. 2007; Shen, et al. 2008) Similarly, there a number of studies which investigate into the VHA system and have specifically focused on healthcare utilization and barriers to access among subpopulations of Veterans who have unique challenges relating to mental health, minority status, or a host of other predetermined characteristics (AF,
45 2002; Cowper, 2004; Fontana & Rosenheck, 2006; Hoge, 2004; Hynes, et al. 2007; Liu, et al. 2005; McCarthy, et al. 2007; Yano, et al. 2003a) This study contributes to the literature by examining, for the first time, the relationship between VHA enrollment and utilization patterns using predisposing, enabling, and need based predi ctors based on data comprised of en rolled and non enrolled Veterans. The significance of this research is relevant to ongoing policy reformation efforts as well as organizational structure, funding, and program restructuring endeavors within the V H A. Expli citly from law, the primary focus of the V H A is to integrate care and expand enrollment in a more meaningful manner which accounts for equity, efficiency and effective ness ("Veterans Benefits Improvement Act of 2004," 2004) In light of these mandates, the capability of th e institution is limited by its ability to provide continuity of care in the PCM when Veterans are electing to receive care outside of the V H A or are using multiple systems of care which severely impacts the coordination and continuity of care (Borowsky & Cowper, 1999; Nelson, Starkbaum, et al. 2007) It is essential for the V H A to better comprehend how it engage s eligible Veterans who may be vulnerable or who may otherwise have unmet healthcare needs. It is antic ipated that this research will inform V H A leadership as to how they can grow targeted enrollment based on an improved understanding of possible ATC barriers and the compre hension of his or her benefits. Additionally, the V H A and the Federal gover nment will gain a clearer understanding as to where they may want to devote their resources. Other stakeholders within the V H A may also find that information on enrollment variations c ould be used to assist eligible Veterans in their
46 choice of avail able ho spitals or departments. Government officials will also have information on possible gaps in needs that could be used to help promote the expansion of services through taxable funding sources. This research will expa nd upon decades of research u sing known predictors of healthcare utilization and employ a nationally representative sample of the Veteran population The study builds upon this research foundation by using a novel approach to identify, compare, and draw new associations from the enrol led and the non enrolled Veteran populations through the full value chain of healthcare delivery Significant i nferences may then be drawn from this first level of analysis and perhaps applied and adapted within the VHA to more effectively target enrollmen t efforts. In order to present the case more clearly, we will need to explore the structure of the VHA. The VHA represents both an integrated insurance and healthcare delivery system operating as one institution (Oliver, 2007) W ith the VHA functioning as both the insurer and the provider, utilization and enrollment become one and the sa me at the initial POS The VHA functions as an insurer in the sense that the institution insures its population of Veterans with h ealth coverage ranging from inpatient hospital care to outpatient services (Percy & Elmendorf, 2009) The VHA functions as the provider of care to the Veteran population employing a staff model which plac es its VHA clinicians on a market competitive salary This model of payment is financed by the VHA through appropriations designated by Congress each fiscal year and further designed to provide structural incentives for a consistent national standard cente red on high quality care ( Kizer & Dudley, 2009) Designed into this structure are processes by which to enroll the Veteran at the same time that he or she receives care within the VHA. A physician
47 encounter is thus recorded at the point of enrollment in order to process his or her entry into the VHA system of care At this point of entr y, enrollment is referenced as a healthcare visit and recorded within the VH the Veterans Health Information Systems and Technology Architecture ( VistA ) ( VHA, 2012 ) B y br idging the gap between enrolled versus non enrolled Veteran populations and utilization versus enrollment, we can begin to draw new and important inferences into why Veterans may choose to enroll or not enroll in relation to significant and important acces s barr iers W e may also be able to leverage important contextual factors used to influence the d ecision to enroll by focusing our attention on enrollment variations wi thin subpopulations of eligible Veterans This is made possible by highlighting uniquely different influences between eligible populations in comparison decision to enroll is the premise that he or she is seeking care, if for only one visit to the institution in his or her lifetime. Clearly, this first visit fulfills two functions; the enrollment process and the initial use of the healthcare system at POS By considering enrollment as a health visit within the VA system, we can bridge enrollment behaviors and pa tterns with the plethora of preceding research and theory on access barriers and utilization patterns. As healthcare coverage continues to change in light of rising costs and a changing national demographic group s of Veterans will likely be categorically affected and exhib it distinctive patterns of behavior. T he V H A will benefit in an assessment of these categorical influences and gain a more lucid understanding of how it may develop target ed enrollment campaigns. The VHA could further justify a focus on e nrollment
48 variations within sub populations and potentially achieve greater access for specific vulnerable group s of Veterans. Succinctly, the development of a study centered on enrollment variation associations will follow suit with one of President Obama national priorities and to end homelessness among V eterans by identifying these individuals through targeted prevention measures available through the VHA (VHA, 2009) Targeted programs could be developed from this research to help identify and educate groups of Veterans who may be at higher risks for encountering ATC barriers relating to their enrollment and utilization of available VHA resources These groups may consist of the u ninsured or others who have known challenges relating to mobility, finances, or other characteristics which have categorically been found to define ATC barriers. In order to further clarify the discussion on how Veterans differ in their use of healthcare resources, we should be privy to some other important changes to the process of enrollment To better understand this process, there is a need to consider the poin t of entry into the VHA IDS from two separate systems of coordination. On one hand, t he majority of Veterans experience a less coordinated decision making process as referenced in Figure 2 1 This flowchart defines the various stages and enrollment possibi lities encountered by most Veteran s following their di scharge from service. At the point in time that the service member discharges from service and assumes the status of Veteran, he or she becomes eligible for certain VA entitlements. The VA offers the Veteran population several opportunities to enroll through direct outreach marketing, open informat ion sessions and local institutional support for Vetera ns who
49 enter into VA facilities on an individual basis. At any point in which the VA is able to enga ge the Veteran, he or she may choose to enroll or not to enroll. At any point in which the Veteran chooses to not enroll, the range in time from when the choice to enroll is given and the next opportunity to gain knowledge of entitlements varies from one c period of time of VA policies which expanded Veteran eligibility The other means of entry into the VHA IDS is through ba sed on Integrated Disability Evaluation System (IDES) This flowchart is referenced in Figure 2 2 and illustrates the process of enrollment into the VHA IDS a mong severely wounded combat Veterans This joint system was recently created by the DoD in January 2009 at the height of the OEF and OIF conflict contingencies and in response to the need for expedited treatment and processing of service members who susta ined catastrophic injuries during combat operations ( DoD Guidance, 2013 ) This represents a distinct and separate program from the typical service expedite the VA benefits process by moving th e member into permanent disability status through a coordinated effort with the DoD, VA, and oth er applicable federal and state agencies. EDES functions within the VA IDES as a means of coordinating the medical evaluation system of care by bridging the Do D military and VA systems and seamless entry into the VHA IDS ( DoD Guidance, 2013 ) This latter option is based on the designated need for care of a wounded service member at the point of injury in the battlefield. In reference to Figure 2 2 the process b egins immediately following injury
50 and transport through the Aerovac system. Aerovac functions as a means of coordinating the transport of the injured from a deployed location to an Outside the Continenta (OCONUS) non combative US staging facility. At this location, the patient is stabilized for a short period prior to his or her trip continuance to the (CONUS). Throughout the entire process, the service member is receiving coordinated care and logistics support from the Theater Patient Movement Requirements Center (TPMRC) a joint service, multidisciplinary team of clinicians transportation, and administrative specialists. This team consists of service members from all branches of the military as well as certain civilians operating from several CONUS and OCONUS locations. Additionally, the Deployed Warrior Medical Management Center (DWMCC) functions as a nurse car e management team at a Germ an based US military installation which coordinates several clinical and administrative processes with TPMRC and Aerovac after the patient has been stabilized for flight to CONUS ( DoD Instruction 6000.11, 2012 ; Probus, 2004 ) Throughout this process, the Veteran may or may not be fully aware of their transition into Veteran status. The level of comprehensive global coordination of clinical is highly complex and flexible to the distinctive needs of each outgoing mission condition on the conditions of the pa tient and the environment (Probus, 2004 ) These needs are defined within the mission requirements and offer a formalized process designed to engage the combined clinical and administrative functions across the full spectrum of DoD and VA resources. The in tent of this research is to focus on the enrollment patterns of Veterans who have been discharged from service without being channeled through the EDES. The
51 underlying rationale for this emphasis is to highlight any specific issues manifesting within the g eneral process of enrollment as opposed to the DoD/VA processes in the EDES. Additionally, the literature is already conclusive in its findings that direct and periphery healthcare program take up rates would largely benefit from an automatic enrollment po licy ( Remler, Glied, 2003) Understandably, the ability to discriminate between those that processed through EDES from those that had not is limited by available data. Since we cannot be certain that Veterans have experienced the EDES without the inclusion of this item within the data set nor an ability to cross reference personally identifiable information with other external data sources, a data strategy must be employed. Fortunately, there are relatively few OEF/OIF personnel who experienced a combat inj ury and thus were not processed through the EDES. Similarly speaking, there were a limited number of overall OEF/OIF respondents in the NSV 2010. As a conservative measure, this study will match any respondents who identify as an OEF/OIF deployed member w ith responses greater than 70 % SCDR as a means of differentiating between the two types of enrollment processes. These matched respondents were removed from this study as a means of isolating the predictor variable effect from the intended assessment. Pri opportunities by which one might gain knowledge of VHA services and entitlements. This information is broadcast through several different forms, forums, and mediums in an effort to disse minate as much general information on available VA programs as widely as possible. While this process permits immediate access to information through
52 established lines of communication between the DoD and the VA, the message may be highly complex and provi ded with little to no context of the specific needs of each discharge process limits the message to generalities where specifics would be better leveraged. The root cause of t his issue lies in when the SCDR is generated by the VA. The issuance of a SCDR only follows when the newly discharged Veteran receives notification from the VA through mail delivery, which is commonly months after the individual has left the service (VA, 2 013). At this point, the Veteran either becomes knowledgeable of his or her entitlement to VHA services through a PG rating or does not become knowledgeable for any number of reasons. Further, the Veteran is presented with a choice to enroll or not enroll if he or she receives knowledge of entitlement. On the other hand, if the Veteran does not receive knowledge of entitlement, he or she may still enroll for any number of reasons. If the Veteran is determined to be eligible and does not enroll, VA outreach may influence enrollment or the Veteran may choose to enroll when presenting with a healthcare need at the VHA at a later date. Consequently, we need to explore the level of enrollment rate variance among different groups of Veterans with varying levels of awareness
53 Figure 2 1 Enrollment Flowchart for Discharged Veteran
54 Figure 2
55 CHAPTER 3 LITERATURE REVIEW The Anderse n Behavioral Model Framework The organizing frameworks used i n this research analysis are based on more than a half Century of ATC and utilization of care studies conducted with in both the private and publ ic healthcare sectors To a very substantial degree this work is built on l of health services use (Andersen, 1995) and the numerous derivative studies that have explored the utility of that framewo rk for understanding access and utilization ( Cowper, 2004) The same conceptual framework has previously been extended to questions about the use of care by enrolled Veterans (Borowsky & Cowper, 1999; Cowper, 2011; Cowper, 2004) This study will extend this research to examinations of non enroll ed V eterans and comparisons to their enrolled counterparts, effect ually treating the event of initial enrollment as utilization the outcome of primary interest. This study examines the effect s of predisposing, enabling, and need based chara cteristics of eligible Veterans on enrollment. A comprehensive review of the literature revealed several critical areas of concern by which Veterans may not be given an opportunity or choice to enroll but rather are prevented from enrolling due to ATC bar riers. One alarming concern is the number of Veterans who self report that they are uninsured when they are eligible for enrollment (Nelson, Starkbaum, et al. 2007) Additionally, the poor, the less educated, and minority Veterans are known to obtain most or all of their he alth care at the VHA as a result of m ore severe ATC limitations within private or employment sponsored health coverage (Backhus, 1999; Nelson, Starkbaum, et al. 2007) Furthermore, t he mental challenges brought on by post
56 traumatic stress or other me ntal conditions are unique barriers that could also be overcome through some form of identification and outreach process (AF, 2002; Fontana & Rosenheck, 2006; Hoge, 2004; McCarthy, et al. 2007) If the VHA is cons idered to be a safety net for these vulnerable populations of Veterans policymakers should consider outreach programs that educate and inform these groups to the benefits of care within this system (Hynes, et al. 2007; Long, Polsky, & Metlay, 2005; Oliver, 2007; Wilson & Kizer, 1997) It is in the hopes that by starting with this foundation of research, the quality of discussion on the effectiveness and efficiency of ongoing enrollment intervention programs will i mprove As a means of drawing together important and significant inferences from the NSV 2010, this research focuses on well established bivariate and multi variate analysis of barriers to healthcare access, presumed or evaluated need for care, and healthca re utilization as predictors of enrollment and use These sets of variables are further guided by the plethora of industry research on barriers to ATC and utilization stemming from the public and private sectors. Of importance is the distinction between t he VHA and most other US healthcare systems of care. There are several ways by which to differentiate the VHA from other organization s Firstly, the VHA is an organizational entity which represents the largest integrated healthcare system function ing as both an insurer and provider of care. Secondly, the VHA operates as a single payer system with several distinct advantages These advantages are principally reflected in system wide cost savings and its ability to integrate care through the use of i ts electronic health record (EHR) and nation wide infrastructure. Furthermore, c linicians are primarily paid on salary and pharmaceuticals
57 and other health products are purchased at pre negotiated bulk rates set by the institution. Quality of care is measu red on a national aggregate of Veterans using standard scales which measure clinical and operational process es and outcomes as well as patient satisfaction ratings. Consequently several peer reviewed studies have been able to show that the VHA surpasses t he quality ratings of Medicare, Medicaid, and every civilian institution on the whole (AK, et al. 2003; SM, EA, MM, RA, & P, 2004) The VHA also uses one of the leading and most comprehensive Electronic Health Records (EHRs) in existence i n the U.S (DC, WP, & JB, 2006) While the affect on quality of care has remained inconclusive at the time of this writing, one systematic review of the literature has shown positive associations in the improvement of the av ailabil ity of patient records post EHR implementation (Poissant, 2005) An examination of the VHA and other US based healthcare delivery an d financing systems is presented for a more contextual perspective. This study will draw upon the associated inferences involving ATC barriers and healthcare utilization and enrollment within public and private healthcare organizations (HCOs) As a means of understanding the commonalities an d differences between the VHA and other HCOs, a thorough review of applicable literature was used to guide the inclusion of critical variables within this study. These variables were used accordingly to assess the likelihood of enrollment and resulting uti lization of services. It should then be surmised that any discussion on utilization was made in the context of understanding the Vetera decision to enroll and use of services within a highly comprehensive and fully integrated single payer health care del ivery system.
58 T his study would naturally be less concerned with variations in organizational characteristics as a result of the uniqueness of the VHA in comparison to other US healthcare financing and delivery systems Rather, this study investigates into individual level variations as a means of discover ing likely and significant barriers to entry and use within the VHA. The use of key variables and analysis informed by theory and prior research could then be used to help prevent or ameliorate ATC process barriers The following review of the literature presents the case for further inquiry using quantitative research methods Barriers to Healthcare Access Aday, Anderse n, and Penchansky are three of the foremost leading experts on ATC and utilization res earch having written numerous studies assessing the myriad of possible reasons why ATC barriers may be occur ring (Andersen, 1995; Penchansky & Thomas, 1981) Early on, Penchansky, et al. was instrumental in defining ATC through the conceptualization of five constructs; accessibility, availability accommodation, acceptability, and affordability. As a means of better understanding how each of the five categories make up the construct of ATC, the following examples draw from the scope of this study. As such, accessibility might be limited when a low er income individual may have challenges in procuring available transportation to and from care or may not be able to communicate with available individual, family, or community resources. Similarly, the availability of VA resources may be limited by the s upply of providers and outreach representatives within a given market service area. Limitations could also arise in how the VA accommodates the population through its hours of operation, its available capacity to meet the need within the community or servi ce area of operation, or even in how it has implemented its telemedicine delivery options. I n the ever expanding world of
59 information technology, the VA must be able to bridge the gap between the acceptability of information dissemination and the preferred mode of delivery of its young and old and Veterans The acceptability of the VA would also likely vary by the health beliefs of its population raising concerns in treatment effectiveness within a model of patient centered care. Finally, affordability of h ealthcare could limit certain categories of eligible Veterans who must share some of the cost of care (Penchansky & Thomas, 1981) Anders e n and Aday provide a foundation by which to assess these barriers by addressing the predisposing, enabling, and need characteristics at the individual level of analysis (Andersen, 1995) The premise of this model is founded on the principle that preferences for care, enabling characteristics of the individual, family, and community, and need for care. These differences in utilization of health services can thereby be translated into variations in healthcare use which can best be explained through the use of the ABM Essentially, the model fra mes the use of care within the constructs of predisposing patient characteristics such as age, gender, race, and other demographic related factors; enabling characteristics such as personal, family, and community factors al structure, stand ard of living, mobility, and other similar attributes ; and need based factors whether perceived and self reported by the individual or evaluated by a professional clinician. Complementing these scholarly articles and models on barriers to access are several US Government Accountability Office (GAO) reports which offer invaluable context on the leadership and legislative actions that have transpired throughout the past several years (Backhus, 1999; GAO, 2003a, 2003b, 2006, 2011a) The GAO
60 reports provide perspective on the many challenges faced by a reformed VA in creating accessibility to a larger eligible population. Using these reports as guidance, several factors of interest were used to unders tand how the eligible Veteran population may or may not commit to enroll into the VA system of care. These issues are subsequently highlighted in the scholarly articles and provide justification for the variables of interest in this study referenced in Tab le s 3 1 and 3 2 R ese arch from the majority of scholarly articles on ATC are primarily focused on barriers relating to geography, demographics, insurance status, health status or disability rating, and options for care outside the V H A. Among the Veteran oriented studies, there are specific references to provider selection and utilization using multi variate logistical regression analysis (Cowper, 2004) Other research drew significant inferences between distance to facilities on choice and util ization of services (Ashton, 1999; Burgess & Avery DeFiore, 1994) Several other studies ass essed the effect of insurance status on utilization (Himmelstein, et al. 2007; Hisnanick & Gujral, 1996; Nelson, Starkbaum, et al. 2007; Shen, et al. 2008; Shen, Hendricks, Zhang, & Kazis, 2003) Mental health b arriers on access were assessed ( W ooten, 2002; Hoge, 2004; McCarthy, et al. 2007) Barriers faced by female Veterans with Posttraumatic Stress Disorder were assessed (Fontana & Rosenheck, 2006) Socio economic factors and environmental and legislative contextual influences on demand and utilization were assessed (Hisnanick, 2000) Patient satisfaction and use w as also assessed (Stroupe, et al. 2005) Additionally, several studies assessed the effects of dual enrollment with Medicare, Medicaid, or private insurance on utilization patterns within the VHA
61 (Borowsky & Cowper, 1999; Carey, et al. 2008; Grubaugh, Magruder, Zinzow, & Frueh, 2009; Hynes, et al. 2007; Petersen, et al. 2010; Shen, et al. 2008) Table 3 1. Dependent and Independent Variables Variable Categories DEPENDENT VARIABLE VHA Enrollment Yes; No Table 3 2. Independent Variables Independent Variables Categories Predisposin g Age Race Over 75 years of age 65 74 years of age 55 64 years of age Y ounger than 55 years of ag e White; Black; Other Races Marital Status Yes; No Education Less than HS; HS Diploma/GED; <1 Year College; 1+Years of College ( No Degree); Assoc. Deg.; Bach. Deg.; Masters Deg.; Professional Deg. (MD, DDS, PharmD, JD); Doctorate Deg. (PhD, EDD) Branch of Service Army, Navy, Air Force, Marine Deployed for Operation Enduring Freedom/Iraqi Freedom Yes; No Enabling Level of understanding of VHA entitlement A lot, Some, Little, Not at all Employment Status Yes; No; No but, Looking for Work
62 Table 3 2. Continued Independent Variables Categories Income <$10,000 $10,001 $19,999 $20,000 $29,999 $30,000 $59,999 >$60,0 00 Type of Health Insurance VHA Yes; No Medicare Yes; No Medicaid Yes; No Private Insurance Yes; No Uninsured Yes; No Need Perceived Health Status Excellent; Very Good; Good; Fair; Poor Evaluated Health Status Presence of need for assistance as defined by an Activity of Daily Living (ADL) or Instrumental Activity of Daily Living (IADL) In addition, an assessment of ATC from both the private and public industries is used to form a basis for further study on common challenges relating to ini tial and sustained enrollment. The following review of the scholarly literature highlights issues relating to ATC barriers in abbreviated form and provides specific statistical evidence of associations and relations among healthcare utilization and demand Studies involving VHA and Veteran research begin the review followed by other related public and private industry articles from peer reviewed scholarly journals. VHA S tudies on Barriers to Healthcare Access Distance to f acility Several scholarly article s concluded that distance was a
63 found to be a strong predictor for variations in access to outpatient care in isolation and utilization of all health services (Ashton, 1999; Burgess & Avery DeFiore, 1994; Carey, et al. 2008; McCarthy, et al. 2007; Petersen, et al. 2010) O ngoing research is being conducted in how to most effectively measure geographic access by defining spatial collections of VHA resources in relation to Vetera residences. Important variations in accessibility and availability have already been found to be highly depen dent on the geocoding methodology employed (Fortney, Rost, & Warren, 2000) Mental health These barriers were found to be particularly burdensome for the cognitively challenged disabled and the elderly who had a much more difficult time traveling to appointments in comparison with other patients (Carey, et al. 2008; McCarthy, et al. 2007) V eterans diagnosed with a mental health conditi on were also found to use fewer resources and have greater challenges accessing care the further they lived from VHA facilities (Fortney et al. 1995, 1998). Another related study assessed both geographic accessibility as being operationalized by distance t o the nearest VHA psychiatric site and decreased availability of service capacity (McCarthy, et al. 2007) Additional ly, stigma associated with seeking mental health care among military members was found to be a strongly influential barrier to access (Hoge, 2004) Another noteworthy RCT study indicated that if mental health was identified and assessed early on in Veterans health status and processes of care improved while also reducing inpatient utilization ( Grol 2001) Gender. As of 2010, the VA indic ated that female Veterans made up 8.1% of the Veteran population with growth estimates rising upwards from 10% in ten years and beyond (Actuary, 2007; Affairs, 2007) With the precedence of care being given to a
64 male population a dramatic shift in approach had to be made to understand what ba rriers may be faced by women. Specific to the recent eligibility reforms are several articles focusing on how the dramatic change s in breadth of services made available to women V eterans are effecting access within this particular group (Washington, Kleimann, Michelini, Kleimann, & Canning, 2007; Yano, et al. 2007; Yano, Washington, Goldzweig, Caffrey, & Turner, 2003b) Another article found that female Veterans were less likely than males to use the VHA and in par ticular, mental health services (Hoff & Rosenheck, 1998) Another more recent assessment of the impact of females diagnosed with PTSD found positive correlations between a female level at the VHA and her contin uity in a therapeutic PTSD care program implemented specifically within the study (Fontana & Rosenheck, 2006) Socio economic and e nvironmental Using a bivariate probit variable joint choice model, one group of researchers extrapolated interpretations of VHA demand based on two levels of analysis ; socio economic individual factors and macro level environm ental influences. By separating these levels of analysis, the autho rs were able to infer that having more abundant insurance options lower s the demand for VHA services (Hisnanick, et al. 2000) Patient satis faction Two notable studies assessed patient satisfaction. One study assessed satisfaction and odds of use of the VHA in a longitudinal study comparing VHA and non VHA utilization over time along with VHA and Medicare contracted service use over time. The investigators concluded that the odds of non VHA use decreased by 11% for each unit of increase in satisfaction among all Veterans. Medicare eligible Veterans decreased by 15% for each unit increase in satisfaction
65 (Stroupe, et al. 2005) The other study found associations between satisfaction and race and ethnicity, comparing satisfa ction ratings between white and non white Veterans. The study found that white Veterans were 1.5 to 3.4 times more satisfied than non white Veterans indicating a possible need for interventions to improve upon these ratings ( Wooten 2002) Insurance status Other research has discovered that the level of ins urance coverage among Veterans is found to be highly associated with their use of healthcare services at the VHA (Shen, 2003) In one longitudinal assessment of the NSVs in 1992 and 2000, insurance status and income had greater odds of increasing their reliance on the VHA outpatient care ( Shen, 2003; Himmelstein, et al. 2007 ; Hisnanick, 2000; Hisnanick & Gujral, 1996 ; Nelson, Starkbaum, et al. 2007 ) Other research inqui ries have centered on associations with being uninsured and use within the VHA Specifically, the VHA continues to be faced with a diverse population of otherwise uninsured eligible Veterans In light of the newly expanded entitlement system this remains a n ongoing issue that has challenged the system since its inception. The fact that many indigent Veterans are fully eligible to be enrolled into the VHA and yet, have often considered themselves uninsured only exacerbates the situation (Hisnanick, 2000; Hisnanick & Gujral, 1996) In a 2007 public health report on uninsured Veterans researchers concluded that 8.6% of Veterans younger than Medicare eligible age remained uninsured (Nelson, Starkbaum, et al. 2007) In a related VA study conducted by Himmelstein, et al. (2007), the authors concluded that Veterans were uninsured and not receiving Veterans Administration care in (Himmelstein, et al. 2007) Additionally, the 2010 National Survey of Veterans
66 highlights that 11% of eligible survey respondents indicated that they were uninsured ( VA, 2010) It is further concerting that we know very little as to what affect Veterans with insurance coverage outside of the VHA and who may be in need of care are not receiving it due to known barriers. Thus, while most Veterans are entitled to some form o f VA benefits, it is not surprising that we know very little about their barriers to access (Himmelstein, et al. 2007) Multiple coverage This variable represents an artifact of the fragmented US healthcare delivery system This term is defined in several different ways to include, multiple coverage in Medicare and Medicaid, multiple coverage in Medicare and a private i nsurer, multiple coverage in Medicaid and the VHA IDS, multiple coverage in Medicare and the VHA IDS and multiple coverage in Medicare Medicaid and the VHA IDS M ost studies on Veteran dual enrollment consider those who are concurrently enrolled in Medi care and the VHA IDS (Stroupe, et al. 2005; Nelson, Starkbaum, et al. 2007 ; Himm elstein, et al. 2007 ; Shen, et al. 2003; Borowsky, SJ, Cowper, DC, 1999; Rosen, AK, et al. 2005 ) This study expounds upon the literature by considering a fuller breadth of multiple coverage in various insurance permutations using mutually exclusive cohorts of Veterans. M ultiple coverage has become an increasingly significant consideration for the VHA IDS in terms of continuity of care between two growing Veteran demograp hics The se demographic differences are occurring between eld erly and younger Veterans, each of which represents unique ATC challenges within their respective categories The effect of these differences in multiple coverage among enrolled and non enrolled
67 Veterans is assessed here as a means of exploring crucial issues in continuity of care and access to care. In addition, a s Veterans continue to age on the aggregate, larger numbers become eligible for the Medicare entitlement N umerous VHA enrolled Vetera ns are now dual enrolled in the Medicare entitlement program and have elected to use care inside the VHA IDS and in the Medicare network of authorized providers. As a result, there are greater opportunities for issues relating to adverse health events, com munication issues between primary and specialty physicians, and the care management of chronic conditions among an aging Veteran demographic (Stroupe, et al. 2005; Nelson, Starkbaum, et al. 2007 ; Himm elstein, et al. 2007 ; Shen, et al. 2003; Borowsky, S J, Cowper, DC, 1999; Rosen, AK, et al. 2005 ) Conversely younger Veterans are fast returning from OEF, OIF and other global contingency operations (VA, 2013) Th e Secretary of the VA recognized that this younger demographic will require a different ma nner of healthcare delivery based on these unique set of healthcare needs (VA Strategy, 2011) Enabling factors may also have a unique effect on younger Veterans who hold a greater likelihood of attaining employer sponsored healthcare coverage re lative to Medicare eligible Veterans Additionally, as Medicaid eligibility expands to those meeting the 133 % means tested income the federal poverty level threshold for family incomes a larger group of younger Veterans will likely become dually eligible for the VHA IDS and Medicaid ( CMS 2012 ) The V H A IDS is rapidly changing to meet these needs by focusing on service gaps and the provision of incentives for younger Veterans to enroll (VA Strategy, 2011)
68 This builds on an established vision of the VA wh ich promotes such programs as providing free care for all Combat Veterans discharged after January 28, 2003 These Veterans represent a unique group of individuals who became eligible to receive five years of free care within the VHA IDS regardless of SCDR or financial means threshold calculations (VA, 2012) After this five year period, the VA re SCDR and financial means threshold to determine whether or not cost sharing for health services and medications may be justified (VA, 2012) The goal of this program and similar endeavors are to enhance the service offerings of the VHA IDS by align ing the mission of the VA with the needs of younger Veterans. Several sources referencing dual Medicare VHA enrollment have established that there are strong associations in demand and utilization variat ions based on contextual market factors, environmental conditions and individual level characteristics. One study concluded that with all else equal Veterans, on the aggregate, will choose care tha t is closer in proximity to the VHA and Medicare contracted services (Carey, et al. 2008) Another study found that the VHA is being tasked in an unprecedented manner to provide care for a dually eligible elderly population and a mass of younger Veterans returning from recent deployments (Grubaugh, et al. 2009) As a result, there are unique ATC issues resulting from a diverse set of needs from elderly Veterans who require greater management of co morbidities and younger Veterans who may have issues with post traumatic stress. Another study assessed the frequency that Veterans used healthcare outside of the VHA system based on individual level factors Certain Veteran characteristi cs and patient satisfaction were found to be significant predictors of dual use, where the odds of dual use increased by 2.4 times if the Veteran was
69 unsatisfied with VHA care, but decreased for those that did not have insurance, were less educated, had he art disease, or alcohol or drug dependence (Borowsky & Cowper, 1999) In another cross sectional study of VHA utilization and Medicare claims data in 1999, indivi dual level and environment a l factors were assessed in relation to their associations with Veterans reliance on V H A and Medicare services ( Hynes, et al. 2007 ) This study found that b lack s and those with a h igher VA P riority Rating were more likely to rely on the V H A. Dual use patte rns were more likely among Veterans with higher risk scores as assessed by the Hierarchical Condition Category (CDC HCC) Additionally, geographic distance and dual use Veterans living in urban regions with more healthcare reso urces were associated with having less reliance on the VHA (Hynes, et al. 2007) Another stu dy using data from 2003 2004 confirmed these results and expounded upon some other factors relating to medical conditions measured by aggreg ated condition categories used in Medicare risk adjustment, access as determined by distance to facility, and PG r at ings. In each ACC reliance on VHA or dual Medicare and VHA resources were found to be statistically significant (Petersen, et al ., 2010) I nvestigators were able to conclude that as priva te insurance coverage declined, reliance on the VHA increased over time i n one of the few studies to examine the impact of dual use with VHA and private insurance (Shen, et al. 2008) This st udy adjusted for selection affects using the NSV 1999, VA administrative data, Ame rican Hospital Association data and the Area Resource Fi le, by controlling socio demographics, health status, Priority Rating, and access to the VHA and other non VHA commun ity resources.
70 S tudies on patterns of health care service utilization have also been conducted in the private health care system where patients may choose between health care providers and plans, depending on his or her available coverage. As su ch, health plans and providers have strong incentives to target enrollment efforts into their plans or care There are several studies referencing the need for better market control mechanisms like individual risk adjustment in private plans to control for adverse selection, a consequence of setting an average price which attracts higher risk and higher costs patients (Culter & Reber, 1998; Newhouse, 2004) Private insurance sector research is somewhat in contrast to the VHA as a publicly funded system of care. This is largely due to limitations in providing market level incentives among VHA sites to attra ct new enrollees. Succinctly, the one major incentive to attract new enrollees to the VHA is the EHCPM which pr ojects enrollment, use, and associated costs of providing services within the VHA for Congressional funding (GAO, 2011b) Considering the broad nature of this incentive, there still remains little incent ive at the local market level of VHA operations. Furthermore, there is a lack of research inquiring into whether the VHA is actively competing with other publicly funded or private health systems or plans in terms of enrollment The closest there is to un derstanding market competition between the VHA and other healthcare organizations is in the ongoing discussion on dual eligibility and utilization. In the context of this dissertation, we have already established that e nrollment equa tes to utilization in t he VHA. As such, it should stand that the VHA competes with the market on enrollment since the incentive is to enroll as many Veterans into the system as possible while keeping unnecessary use and costs down.
71 Competition derives from several known source s. Succinctly, a number of Veterans have multiple systems from which to choose care Certain dual eligible Veterans with other types of health insurance predominantly consisting of Medicare, but also private insurance, provide access to more health care re sources in comparison to coverage in isolation Thus, w hen the direct or indirect costs of care from any particular source are considered to be m ore expensive to the Veteran, the alternate resource is used. From another perspective, when the dually eligibl e Veteran encounters barriers to access to one or more dimensions of access in one system, the alternate system is used. Studies referencing the effect s of various healthcare delivery models are primarily based on economic models used to interpret the eff ects of overuse and underuse on patient outcomes. The call to action to prevent adverse patient outcomes is continuing to take shape within the ongoing national discussion on improving healthcare quality. Within this context, Chassin, and Galvin (1998) and Grol (2001) c lassified healthcare delivery quality problems into three broad categories; overuse, underuse, and misuse. Overuse of health services is apparent when too much care is delivered to a patient, when the risks of treatment are greater than the p ossible reward, and when the service causes harm or the potential for harm to the patient Conversely, underuse occurs when there is a deficiency in healthcare service when need is most apparent and a favorable patient outcome is likely. In addition, m isus e may occur as an example, if an incorrect treatment such as penicillin is administered to a patient when known allergies to this medication ( Chassin et al. 1998; Grol, 2001 ).
72 An evaluation of these claims was conducted by Baumgardner (1991) using econom ic models to depict differences in marginal value of healthcare delivery wi thin different insurance models. In comparison to a conventional insurance plan with fragment ed payment and delivery systems, HMO's are thought to be better positioned than conventi onal insurance to control diseases through medical delivery inn ovations. The reasoning follows that moral hazard may be reduced by the internal controls found within the contractual design of the HMO and the integration of payment and healthcare delivery m echanisms. These controls relate to the inherent design of the HMO to control costs of a patient population and reduce unnecessary visits. Conversely, conventional insurance provides incentives to increase the volume of care. At a simplistic level, as volu me of care rises within the conventional insurance scenario, the potential for overuse grows in direct effect. The insurance effect is also evident on the other side of the scale, with underuse disproportionately effe cting the elderly and uninsured popula tions Ash, et al. (2000) found that underuse resulted in negative outcomes in these vulnerable populations u sing a claims based method to determine need and resulting use. In another related study, it was found that positive health outcomes were evident i n the uninsured population after gaining access to insurance When the uninsured populati on was given coverage and compared with those already insured who gained better coverage, the effect on patient outcomes among the uninsured was found to be highly sta tistically significant while the insured populations showed little to no statistical difference (Ware, et.al., 1980)
73 Health status Health status is represented in this study as a need based factor which impacts utilization in some form With all else c onsidered, it is also considered the only categorical factor which determines the equity of a system of healthcare delivery ( Andersen, 1995) The relative equitability of the VHA IDS is being considered he re to shed light on how well it s systematic processes are designed to ameliorate barriers to ATC among different categories of Veterans. The logic follows that equitable healt hcare access can only be realized if individuals base their care seeking decisions exclusively on their perceived or evaluated need. E valuated need is defined by professional clinician Activities of daily liv ing (ADLs) and Instrumental activities of daily living (IADLs) are used as means of assessing evaluated need using self reported data. Conversely, perceived need is completely his or her own health status. Di sability ratings are determined through a VA led system of standardized evaluative protocols designed to determine the level of care needed as well as the level of structure support afforded to eligible Veterans (VA, 2011) Fo r the purposes of this research, disability ratings of eligible Veterans were categorized as an enabling factor given the level of focused outreach and structural support offered to groups with higher SCDRs. Based on previous research by Hisnanick, et al. ( 1996, 2000), there is strong evidence that as evaluated and perceived needs within the population rises, utilization is uniformly affected in a direct manner (Hisnanick et al. 1996 2000 ) Drawing from several other literature sources, Veterans were found to yield much higher rates of chronic disease and co morbidities relative to the civilian p opulation. As a result of these higher rates, Veterans, on the aggregate, have a relatively higher need for care (Asch et
74 al. 20 04; Kazis et al. 1998 ; Kazis et al. 2004b; Selim, et al., 2004; Selim, et al., 2006 ). Using the context of perceived and eval uated need conjointly within this study, we may further refine these findings in the context of realized ATC within enrolled and non enrolled Veteran populations. Public industry s tudies In principle, the health care delivery and payer systems within the pu blic sector operate similarly to the private industry The public health financing and delivery system s are comprised of federal and state HCOs and payers Dep artment of Defense (DoD ) military treatment facilities (MTFs) and contracted payers and the Vete ra ns Health Administration (VHA) as an integrated provider and payer system. Operating independently from each payer system, with the exception of the VHA, are public delivery systems which come in the form of community free access clinics, state, and fed eral hospitals, and MTFs. T he payer systems consist of the VHA, Medicare, Medicaid, TRICARE, and the State Children's Health Insurance Program ( CHIP ) and function as financers of healthcare claims operating within and sometimes outside of their respective systems or networks. On the other hand, public HCOs function differently than the private industry in terms of the breadth of their delivery systems, the degree of coordination in care, the beneficiaries deemed eligible to enroll for care, and the manner of payment This is a primarily a result of their missions to improve public health on the aggregate through tax payer money Upon a review of the literature, an abundance of references relating to public HCO system s were devoted to barriers and challeng es to enrollment as a means of assessing ATC; and continuity of care, a means of assessing use These included
75 articles devoted to understanding why Medicare, Medicaid, and CHIP had varying degrees of success with insurance uptake rates (Blustein & Hoy, 2000; Brown & Langwell, 1988; Clement, Retchin, Brown, & Stegall, 1994; Morgan, 1997; Moscovice, Casey, & Krein, 1998; Tai Seale, Freund, & LoSasso, 2001) There were no peer reviewed scholarly articles discussing A TC within state, federal, or DoD hospitals. There are three likely reasons for this: Firstly, that state and federal hospitals are devoted to the care of a very specific and small population of beneficiaries, many of which are inpatient. This limit s the ne ed to study ATC barriers as this issue is primarily experienced in the outpatient setting Secondly, the DoD system is largely engaged with the VHA as a means of improving ATC for military members that have been injured and are being medically discharged from service or have engaged in a memorandum of agreement between and MTF and a VHA hospital or clinic Since the VHA has a research division which assesses these issues and publishes in peer reviewed journals, it becomes a natural proxy for some DoD relat ed ATC challenges. T hirdly, DoD hospitals are primarily reporting directly to Congress on ATC issues. This reporting structure is inherently presented within a government bureaucracy which generates reports through government watchdog organizations such a s the Congressional Budget Office (CBO) and the Governm ent Accounting Office (GAO). As a condition of this academic study, these government reports were only included in a limited fashion as a means of providing guidance on central trends and policy initia tives. T he reasoning is central to the analysis of quantitative studies which rely on the statistical significance of variable associations to draw inference. Since these reports
76 were written for other purposes, only data relating to descriptive demographi cs, trends, and policy were used. In contrast to other state and fed eral healthcare delivery system studies the VHA has been at the center of numerous studies on access and utilization patterns. The reasoning may be attributed to the fact that the VHA fu nctions as both payer and provider of care for over 8.5 million Veteran beneficiaries operating as the largest healthcare system in the U.S. (NCVAS, 2011; Pananga la, July 27, 2010; Westat, 2010 ) These long term ob ligations also create a need for a large budget to finance its service delivery, payment system, operations, research and development. To put into perspective, Congress passed legislation in fiscal year (FY) 2012 mandating $51.1 billion in appropriations t o be used by the VHA (Panangala, Aug. 1, 2011) Finally, it should be noted that the VHA is the only full y integrated single payer and national healthcare system in the US. with a comprehensive electronic health record database in existence since 1977 (Kolodner & Douglas, 1997; Oliver, 2007) Upon extensive review, t he re was an apparent gap in how ATC is af fecting enrollment i nto the VHA, a payer and provider of care. Thus, the following review on public HCOs begins with an examination of Medicare, Medicaid, and CHIP payer programs to provide perspective on enduring ATC barriers faced by individ uals within an entitled system su ch as Medicare and social welfare systems such as Medicaid and CHIP (CMS, 2012 ) Studies reflecting ATC and use within the VHA delivery system will then be used to inform further study into ATC barriers relating to enrollment. CHIP studies The State Children's Health Insurance Program ( aka SCHIP and CHIP ) is a social welfare program created under Title XXI of the Social Security Act
77 intended to help states finance the insurance of low income children who are ineligible for Medicaid (CMS, 2012) Through the latter part of the 1990s, several studies focused on enrollment var iations and ATC an d use barriers within CHIP. This was in response to vuln erable children within eligible subpopulations. Throughout the remaining part of the last decade and a half, research became more focused on associations between policy changes and program and family characteristics. Earlier studies used odds ratios to assess the likelihood of enrollment using dichotomous and categorical predictors within a multivariate logistic model. In a 1996 study, researchers found predisposing, demographic predictors coupled with basic program characteristics were significant determinants of enrollment and use. As a preliminary study, these conclusions were important informants for further research into more refined predictors of different levels of premium subsidies and other possible demographic predictors outside of age, sex, race, and ethnicity (Shenkman, Pendergast, Reiss, Walther, & et al. 1996) Subsequent research built upon these results by using a simulation design to evaluate the ratio of eligible to enrolled children in relation to predisposing and enabling characteristics This design used the standard formula for eligibility se t by the state and combined the results with 1996 Medical Expenditure Panel Survey to determine the ratio. It was determined that children from birth to twelve years of age were more likely to enroll than other eligible children. This was attributed to pub lic policy which affects certain children who remain covered by Medicaid through the age of twelve, by law.
78 Additionally, younger children were more likely to require more healthcare resources and likely drew up Medicaid take up rates (Selden, Banthin, & Cohen, 1998) In response to policy recommendations to expand CHIP coverage to parents, healthcare administrators warned of possible adverse selection of high risk adults resulting in rising costs. One study investi gated into these concerns by assessing the characteristics of eligible sample of survey data. The results yielded large variations in adult enrollee characteristics in terms of age, health status and utilization ( Fries, Kullgren, & McLaughlin, 2003) Program characteristics were also used to explore possible reasons for State variations in enrollment. Using data from the 2001 Current Population Survey, th e authors conclude d that state s with integrated Medicaid expansion programs were more successful at enrollment than states which had separate outreach services, benefits, and coverage services. The authors attributed this result to the possible excessive a dministrative burden associated with managing separate processes relative to an integrated system (Kronebusch & Elbel, 2004) In a more recent study, researchers assessed both the characteristics of eligible, non statewide survey adminis reality of enrollment through an administrative check of enrollment. The study found that greater than 20% of parents believed their children were enrolled when, in fact, they
79 were not. The study also reported that the characteristics of eligible children who tended to be non enrolled were older, came from employed families with higher incomes and had unin sured parents (DeVoe, Ray, & Graham, 2011) Medicaid studies Medicaid is a social welfare health program created under Title XIX of the Social Security Act as a financial means based and disability based State run program for U.S. citizens and legal residents (Medicaid, 2012) The program is monitored by CMS and funded through both the state and federal governments and covers low income adults, their children, and certain disabled individuals (Medicaid, 2012) Several policy revisions have been made inception in 1965. Most notably, these changes are evident in the move from the traditional f ee for service (FFS) coverage towards managed care, a method of payment which pays private HCOs a fixed premium for the provision of all or part of a healthcare services. In theory, e nrollment into managed care programs promise to reduce costs to taxpayers by reducing the demand for unnecessary medical services and transferring some of the risk of coverage from the State to the private insurer. There are several studies which highlight some of the inherent issues with enrolling individuals into Medicaid and Medicare managed care programs. One study assessed the issues rooted in the provision of care in rural areas and found persistent concerns with lower rates of enrollment in comparison to other geographic areas (Moscovice, et al. 1998) Another study investigated into utilization trends and racial disparities resulting from the mandatory enrollment of Medicaid recipients into managed care plans. The researchers used several count data models which were risk adjusted for nonrandom selection effects and a difference in differences analysis. The authors
80 concluded that Blacks were disproportionately affected by a reduction in services which pointed to unequal ATC and probable dissatisfaction in service (Tai Seale, et al. 2001) Further research argued that managed care plan characteristics may be limiting ATC and coordination of care for disabled beneficiaries within Tennessee's Medicaid managed care program (Hill & Wooldridge, 2002) The design employed the use of a statewide survey of this subpopulation conducted between 1998 and 1999. Analysis compared self reported ATC and perceived qu ality across all TennCare managed care organizations ( MCOs ) and found little overall variation across plans. The authors attributed this indifference to the use of case utilization management Additionally, n o inference s could be made on possible statewide policy variations, an area for furthe r investigation (Hill & Wooldridge, 2002) Community program s tudies In response to the growing number of uninsured individuals, the Robert Wood Johnson Foundation funded the development of several community based programs designed to insure those without healthcare coverage. This program formed the basis for studies designed to assess several areas of impact. One of these studies assessed enrollment and sust ained utilization issues within three of the fourteen sep arately funded program sites (Erin Fries, McLaughlin, Warren, & Song, 2006) All sites were chosen for their variability in demographics, location, and health status of par ticipants. The study found that, controlling for health st atus the primary reason for ini tial and sustained enrollment in two of the three sites was to gain access to preventative services and a usual source of care. Consistent u tilization of preventative services was also found to be statistically important in all three sites ( Fries, et al ., 2006) Therefore, c onsideration must be given to the type of service offered within the HCO or
81 program as being an i mportant determinant of the perceived need for enrollment and sustained use. Medicare studies Medicare was created in 1965 under Title XVIII of the Social Security Act as a healthcare insurance program funded and operated by the Federal government through Federal taxpayer money. In contrast to Medicaid, Medicare is an age based entitlement program offered to US citizens and permanent residents who are 65 years of age and older. Similar to Medicaid, Medicare has moved towards a managed care system as a means of curbing some of the rising costs of healthcare. As a result, Medicare policymakers and planners alike are interested in how it may increase enrollment into Medicare MCOs and how it might justify its continued use through cost reductions and decreased use of unnecessary services (Medicare, 2012) One study assessed the impact of Medicare managed care on symptomatic improvement, access, and use of services in enrollees experiencing joint and chest pain using random digit dial of a stratified random sample of managed care enrollees in 1990 (Clement, et al ., 1994) Risk adjustments were considered in the a nalysis of the outcome variable. These included the use of vari able controls for demographics, health/functional status, and health behaviors over a twelve month period. The results indicated that in comparison to FFS enrollees, managed care reduced self r eported improvements in symptomatic pain as well as access and use of services (Clement, et al. 1994) This study offers further support for the use of predetermining variables relating to demographics and enablin g variables relating to the restrictiveness of the plan. In addition, this study provides support for the use of self reported health status as a predictor of need.
82 Selection bias was evident in managed care versus FFS enrollees. This study assessed the d ifferences in use over time among managed care inpatient enrollees before enrollment (under FFS) and managed care enrollees after disenrollment into the FFS plan, controlling for demographic characteristics (Morgan, 1997) The results confirmed their hypothesis, showing 66% use prior to enrollment and a tripling percentage of use, post disenrollment (Morgan, 1997) In consideration of perceived need, one may infer that enrollees chose to dis enroll to gain better ATC. Another article challenges the assumption that th e Federal community benefit requirement of non profit (NFP) organizations influences the greater provision of care for the less healthy, indigent populations in comparison to for profit (FP) organizations who are not bound by this federal mandate This is an important consideration in light of the rise in FP Medicare MCOs. The research inquired into this matter by assessing the characteristics of Managed care enrollees in both NFP and FP MCOs through a nationally representative cross sectional analysis of t o the 1996 Medicare C urrent Beneficiary Survey The results showed that while the demographics of enrollees in MCOs were generally identical, FP MCOs enrolled the poorer and less educated (Blustein & Hoy, 2000) Ad ditionally, while these same enrollees had little to no knowledge of the tax status of their MCO plan, they self reported that they joined these pla ns to reduce their OOP expenses (Blustein & Hoy, 2000) Accordingl y, this study supports the use of predisposing and enabling variables by assessing the perceived costs of healthcare enrollment and use with relative income. Medicare prescription c overage In 2006, Medicare implemented Part D, a new benefit which covers beneficiaries for outpatient prescriptions under the Medicare
83 Prescription Drug, Improvement, and Modernization Act (MMA) of 2003. Eligibility requirements for Part D remained identical to enrollment into Medicare hospital or medical insurance but, require d the payment of a monthly premium by the enrollee. The CMS set a goal of 90% coverage of beneficiaries prior to the deadline for open enrollment on 15 May 2006 (Medicare, 2012) A group of researchers inquired into whether or not beneficiaries were inten ding to enroll before this deadline and who decided to enroll into this new program. The demographics of the beneficiaries were used in conjunction with survey data conducted pre and post enrollment deadline. Responses from the first wave assessed Medicare follow up survey included items which referenced the decision making process of enrollment as well as an assessment of attitudes relating to satisfaction with the plan (Heiss, 2006) Further a nalysis found that CMS was able to attain its goal of 90% or greater enrollment. One of the primary reasons for their initial success was the automatic enrollment of employees of the fed eral government, military, VHA, and certain union plans. Additional analysis was conducted on the remaining population presented with a choice to enroll or not enroll. W idows, unmarried women, and less educated individuals were found to be more likely to h old minimally comprehensive prescription coverage and were the least likely to enroll. Others without pre existing prescription insurance were less likely to enroll. Y ounger individuals with zero or one prescription and little to no immediate need for care were also less likely to enroll. These groups were all faced with tradeoffs between the immediate OOP costs of medication with and without Part D. It was unclear whether future costs were considered by individuals who elected to not enroll and coverage wa s needed post open enrollment. Of those that chose not to enroll, the
84 majority claimed that they had trouble understanding the benefits of processes of enrollment of Part D (Heiss, 2006) Subsequently, the association between b eneficiary awareness of ent as an enabling factor within this proposal. Public healthcare and social program s There are several articl es referencing uptake rates between social programs ranging from Health Insurance Program (CHIP), Woman, infants, and children (WIC), food stamps, unemployment insurance, and many other social programs designed to help the el derly, poor, disabled, homeless, and other vulnerable populations receive public assistance. Research on uptake rates involving healthcare entitlement and social welfare program participation is crucial to understanding which barriers are being faced by di fferent eligible populations. By looking at the issue from within and peripheral to healthcare delivery and health insurance, we may gain a better perspective on the degree of difficulty faced by individuals who are quali fied for services but, remain non e nrolled for various reasons. Remler and Glied (2003) offer a well rounded literature review of uptake rates using this strategy. This review offers an interesting evaluation of the effects of commonly believed reasons for enrollment issues ranging from a l ack of information or awareness about different programs and eligibility, the negative stigma tied to the acceptance of public programs, the time required to become enrolled, the degree of administrative processes required to complete the application proce ss, or to a general lack of interest in the particular program. The study was quick to point out the limited research on the effect of information and awareness of benefits on enrollment using quantitative data. Curiously, the review
85 was written at a simi lar time in history when new legislation greatly expanded healthcare eligibility within government programs. In 1997, the DHHS expanded Medicaid eligibility to American children through CHIP (2013) While the impact of this policy development was considere d one of the largest since the inception of Medicare and Medicaid, it was not as far reaching as the PPACA. Nevertheless, its findings on barriers to enrollment and access to care are highly germane to the issues faced by Veterans today. One of the article s it reviewed by Neumann et al. (1995) specifically addressed the effect of benefits on enrollment into the Qualified Medicare Beneficiary Program (QBP) using a nationally representative sample from a cross sectional survey. The QBP is an entitlement program which reduces premiums and the level of cost sharing for low income Medicare beneficiaries using a means based threshold test (CMS, 2013). While there was limited evidence that awareness was correlated with a 20 percentage po int increase in beneficiary enrollment into the QMP, it also found that many enrollees remained unaware of the program. Outside of its original hypothesis, this research uncovered the effect of the provider in the process of enrolling eligible beneficiarie s into the QMP given a built in financial incentive to avoid risks of losses and bad debt. Yelowitz (2000) offered some confidence to the effect of information over time on enrollment in his cross sectional study on QMP. Yellowitz found that greater enrol lment occurred in beneficiaries who were eligible in the period(s) prior to their first opportunity to enroll. In another cross sectional study researchers pointed to a negative relationship between the complexities of eligibility rules in the Medicaid pro gram and over all take up rates (Stuber et al ., 2005 ).
86 The authors were able to conclude from their literature review that many programs had commonalities existed in how uptake rates could be improved. The preponderance of evidence suggests that the size o f the benefit measured over greater amounts of time was the most important predictor of enrollment. As such, Veterans with higher SCDRs and lower incomes would be the most likely to find the greatest benefit to nal IDS and over an entire lifetime. awareness to their entitlements on enrollment. While Remler and Glied (2003) provided a thorough literature review, this section provides f urther evidence that several other factors are, in fact, influencing uptake rates. In consideration of the lack of data on the literature, this study will follow through with assessing this factor using available quantifiable dat a in relation to the size and duration of the benefit Summary The articles summarized within this section focused on access barriers in terms of geography, demographics, perceived and evaluated health status, and options for care within several public he alth organizations Distance to facility was common ly used as a proxy for access. ictor of variations in access. When applied to dual use between public systems, distance was also found to be a strong pr edictor of choosin g care. Consequently, this becomes a particularly important consideration for the predominantly less mobile; the disabled and the elderly. Other barriers to access included the socio economic status of individuals. It was not uncommon to learn that the indigent remained uninsured even though they were often eligible for care within the VHA and welfare programs such as Medicaid and CHIP
87 Medicare and Medicaid research was primarily focused on ATC and use within ma naged care plans. The rese arch investigated into associations between enrollment and utilization as related to adverse selection based on health status, beneficiary demographics, geographic area of residence, and expected OOP cost s and risk sharing assumed by the beneficiary. CHIP research primarily focused on the characteristics of eligible children in relation to enabling factors linked to the family support mechanism, State policy, and the degree of program integration as be ing strong predictors of enrollment and use. Need was no t a strong determinant of use unless the child was very young. The one community program article addressed the need to consider type of service within enrollment and use analysis. Studies which presented some analysis on the possible effect of a wareness o f benefits on participation and enrollment into public health programs provide some level approached with caution, primarily due to their focus on alleviating the prem ium effect. Since the VHA IDS does not charge a premium for continuous coverage, this specific effect is moot. What may yield the greatest effect is how awareness is impacted by the perceived size of the benefit and expected length of time the benefit is t o be received. Private Industry Studies Several articles discussed choice in plans and satisfaction with access to quality care and cost using organizational, behavioral, and economic theories These theories set the groundwork and provided the means by which to assess management issues dealing with the development of institutional capacity focused on improving ATC controlling utilization rates, and improving enrollment take up rates within different programs The summation of this research has revealed mixed evidence for several
88 decades, most likely as a result of the complexities of the various private industry offerings and the uniqueness of market conditions. One of these articles provides a good platform from which to understand the inherent difficu lties o f institutional associations and the ongoing assessment of cause and effect from which to draw meaningful conclusions Specifically, we may consider one case study on Kaiser Permanente in Northern California which highlights the undesired effects of possible omitted variables in research design. The research question attempted to answer how management s endeavors to reverse declining enrollment rates in the face of reimbursement cuts may be affected by improving patient satisfaction T he study explor ed this inquiry using two different organizational theories, resource dependence theory and institutional theory. The resource dependence theory postulates power (Pfeffer & Alison, 1987) In reference to institutional theory, structural change in organizations is driven less by the ne ed for efficiency or as a competitive response to market conditions, than as the result of processes that make organizations more similar without necessarily making them more efficient (DiMaggio & Powell, 1983) The authors tested the idea that organizations with the greatest power to improve patient satisfaction would be better incli ned to improve enrollment rates. Further, the research used a time series analysis using qualitative and multivariate regression methods then tested for error correlation using ordinary least squares. The results found inconclusive evidence claim that improved patient satisfaction or strategic management caused an improvement in enrollment (Barr, 1998) This article considers the
89 impo rtance of exercising caution in inferring cause and effect as there may be a myriad of factors operating influence over the end result. In another study on patient satisfaction and enrollment across 3,000 adult enrollees in fee for service (FFS) and manag ed care insurance plans the results were found to vary by type of plan and socioeconomic factors (Davis, Collins, Schoen, & Morris, 1995) This study concluded that take up rates could be improved by focusing on w hich types of individuals it wanted to attract into its different plan offerings. FFS enrollees were more satisfied with access while managed care enrollees were more satisfied with cost. L ow income managed care enrollees were the least satisfied overall i n terms of cost, access, and quality. Consumer OOP payments were also shown to have an effect on enrollment, use, and satisfaction in a case study on University of Michigan Medical Center Using a time series multivariate analysis over a one year time period, the investigators were able to surmise that self selection was occurring a mong young, healthy, working individuals who were offered lowe r premiums and preventable coverage (Billi, Wise, Sher, Duran Arenas, & Shapiro, 1993) In another seminal paper, several invest igators explored the relationship between employed decision to enroll in an open or closed network HMO based on risk and economic vulnerability to premium and OOP costs (Berki, Ashcraft, Penchansky, & F ortus, 1977) It is w orth mentioning that this pap er was written on the heels of nization Act of 1973 activated the now well established practice of government subsidizing for employers offering HMO health insurance. The authors used an economic model of maximum likelihood of enrollment
90 into one open and two closed HMO network models as well as a FFS Blue Cross/Blue Shiel d model existing in the 1970 s. In theory, enrollees would self select into a closed HMO based on their lower relative income To test this hypothesis, the research consider ed multivariate effects of per capita income effects as well as the effect of having pre existing patient provider relationship among other individual and plan characteristics. The resul ting analysis indicated that choice of plan considered these effects in two separate ways. In a closed plan, the enrollee elects to simultaneously choose the plan and its network of available providers. In an open insurance plan, the enrollee primarily sel ects the plan based on its cost and coverage. Additionally, by adjusting for income effect on the per capita basis as opposed to family income, the effect of coverage for larger than average families becomes more refined as higher living costs are risk adj usted Additional adjustments to the age of beneficiaries their health status, and whether or not there was prior use, uncovered a lack of adverse selection occurring in plan selection based on economic vulnerability The researchers noted that if differe nt plans would become more competi ti ve on price in the future non financial factors would likely have more effect on access and choice (Berki, et al. 1977) In addition, this research set up future research in it s consideration of health status as a likely variable affecting a dverse selection. Another more contemporary assessments of care and plan selection by considering the associations within insurance and in dividual characteristics and the different aspects of product types offered by insurance plans; indemnity, PPO, point of service (POS), and HMOs. The characteristics of these plans were evaluated using commonly employed cost control
91 mechanisms of health pl ans ; the use of networks, gatekeeping, capitation, and group/staff model delivery systems. Analysis was based on data from the Community Tracking Study (CTS) 1996 97 Household and Insurance Followback Surveys and the use of multivariate models. The authors found a continuum of satisfaction and dissatisfaction as consumers traded off cost and access. As enrollees moved towards more restrictive managed care models, dissatisfaction became more evident. In contrast, enrollees who were attracted to less restrict ive models were more satisfied (Kemper, Tu, Reschovsky, & Schaefer, 2002) The effects of non financial barriers were further made apparent in a study on ATC within vulnerable populations enrolled in private HMOs. This study used Consumer Assessment of Heath Plans Study (CAHPS) survey data from 1997 to 1998 within New Jersey and among 13,952 enrollees participating in 13 plans, ten of which were part of a national consortium. CAHPS has been modified several times si nce the date of this publication however; the primary intent remains the same, to compare self reported individual assessments of health plan quality and ATC. This study used a cross sectional design and multivariate logistic regression to assess differenc es across individual characteristics such as age, education, income, race, and health status. Overall, nonfinancial barriers were found to be more statistically significant than financial barriers in predicting ATC challenges. This was particularly apparen t within categorically defined vulnerable populations with lower income, education, poorer health, and among minorities (Carlson & Blustein, 2003) In consideration of market service area research articles were a lmost exclusively limited to the insurance perspective. In an effort to broaden the search to health delivery
92 systems, I relaxed the limit of peer reviewed articles and found a handful of case studies with little to no scientific rigor. Given the limits of this literature review, these were summarily excluded. As such, the inclusion of studies relating to the insurance industry and ATC and utilization were used primarily as a result of their ability to draw power from their sample size and representativenes s but, also as a result of their inter relatedness to the private healthcare deliver y system. On this latter point and in its present form, our modern health delivery system has become highly reliant on its ability to attract volume from third party payer systems. Since insurers have taken on the role as gatekeepers to many of its beneficiaries, any issues relating to ATC are also inter related. Much like insurers, strategic healthcare managers use measures to assess service areas. O ne study provided an ar gument for the inclusion of a standardized method of identifying ATC barriers through the use of a primary care service area tool, or PCSA (Goodman, 2003) The PCSA has often been used by strategic hospital and insurance executives as a means of identifying and developing markets that offered low competition and higher or lower utilization patterns (note that a typical insurance plan would be looking for lowe r utilization patterns while a typical hospital would be seeking higher utilization patterns) From the vantage of the HCO, this tool serves as a competitive marketing tool but, also serves as a means of assessing ATC. This study used several sources of se condary data drawing from Medicare, Medicaid, and Blue Cross/Blue Shield, all of which have the potential to link several sources of care and improve upon ATC through better coordinated care delivery These data were a ll found to be generalizeable across b oth public and private plans and are now used extensively
93 throughout the industry The findings pointed out that the tool could successfully iden tify varying leve primary service area s (PSA) dependent on the size of the population. More concisely, this study found that patients who were highly mobilized outside of their PSA typically resided in smaller populations while t he inverse was shown for larger populations. The use of this tool could then be used in place o f the standard methodology which defines physician competition through an aggregation of county/state market data to one which refines the measure to show greater variability in market areas (Goodman, 2003) Furthermore r esearchers and health planners could use this tool in conjunction with the Dartmouth Atlas of Health Care (2013) as a complementary resource as means of assessing both public and private mar ket capacity in an effort to improve upon the availability of its scarce resources 1 Healthcare delivery would thus be altered by refining the focus of delivery on the needs of the population and less on the effects of limitations associated with enabling factors. Summary There are several important conclusions that can be drawn from this literature review on private industry ATC and utilization. Firstly, these studies span several decades of research on access and enrollment within private managed care health plans. At the outset of the enactment of the HMO Act of 1973, the diversity of managed care plans has expanded in various innovative ways with the primary goal of thwarting rising healthcare costs assumed by the insurer I nstitutions and providers h ave 1 The Dartmouth Atlas use s Medicare data to identify market inefficiencies and opportunities for improvement as well as possible fraud, waste, and abuse of Federal tax dollars used to finance healthcare across the nation.
94 increasingly been held at the mercy of the insurance industry and its control over how care is delivered This is particularly germane to how healthcare d elivery has been m odified by the restrictiveness of care within plan networks the level of OOP ex penses assumed by the consumer, and overall benefit coverage and plan carve outs within various markets As a result of these peculiarities the generalizeability of results from any study should be taken with caution with heightened skepticism reserved fo r comparative research focused on the private industry Furthermore, we need to consider that each of these aforementioned studies used different conceptua l designs, data sources, and populations The use of quantitative qualitative, or mixed methods are all represented with varying levels of power and population representativeness and should also be measured with caution when drawing inferences across different populations As a fact, employing even the most rigorous m ethods in these methodological desig ns limits any inference in drawing causation from any specific independent variable. Rather, these studies provide further support for the use of a collection of plausible individual characteristics as a means of assess ing enrollment and utilization. These may also be used in a more meaning ful manner by private industry leadership as a means of approach ing strategic interventions designed to improve ATC through the full value chain of service delivery and within the context of different service areas With the preponderance of private industry research point ing to adverse selection as a primary cause for concern there is convincing justification that it should be tested within the context of this research. PG ratings and geographic service areas would like ly hold the greatest promise for future research as a means of controlling for
95 adverse selection. By employing a crossover analysis within a subset of PSA analysis using these two variables we may better understand the possible associations tied to how Vet erans may have been selected into the VHA IDS or some other form of healthcare resource. As a res ult of data limitations in the de identified set of Veteran data used in this dissertation any PSA analysis would be infeasible at this time. If future itera tions of NSVs employ these data or permit linkage to other data sets we may discover more detail which explains Veteran enrollment and utilization patterns In the absence of PG ratings and more geographic detail, adverse selection was controlled for at a less detailed level using predisposing, enabling, and need data within the enrolled and non enrolled Veteran populations. Public policy and s t rategic management i mplications This research had significant relevance to ongoing policy and organizational re formations within private and public systems. Explicitly, the primary focus of public systems is to inte grate care and expand enrollment in a more meaningful equitable, ef ficient, and effective manner. In light of this, there are limitations to how healthc are institution s and public program s provide quality care to individuals who face barriers to enrollment or are unable or unwilling to receive c are when care is needed In reference to private organizations, the effects of adverse selection, patient satisf action, expectations of the level of cost sharing and risk sharing, along with socio demographic and need factors rela t ing to age, education, income, race, and health status, were shown to have statistical relevance to predicting enrollment and use. The l iterature gap on ATC is most apparent in the VHA public health system in
96 terms of barriers to enrollment. By studying ATC barriers which are considered endemic to private payer HCOs and public health programs, further research can be guided through program comparisons. Utilization of services and enrollment were at the center of this review as a result of the recurrence of references drawn from ongoing challenges faced by HCOs and public health programs. These references build upon each other from the vanta ge of each separate sector and from a historical perspective of changes to policy, consideration of demographics and the advancement of research rigor. Private plan studies uncovered comparable issues facing the VHA IDS but, were considered in light of d ifferences in the populations measured, plan characteristics, and the level of fragmentation between the ir delivery and payer systems. Of interest in their analysis was the use of service area analysis as a means of improving enrollment into their systems. While the VHA may use service area analysis in some form, the research does not conclude whether this has been effectively translated into improving ATC and enrollment in any scale able fashion. Since most private HCOs are primarily concerned with how to improve ATC and take up rates through a model of localized competition there remains a degree of uncertainty as to how this may translate to the larger mission of the VHA IDS. The Medicare entitlement program may be the most comparable to the VHA IDS in t erms of its governance, breadth of coverage, and demographic makeup. This program co vers individuals 65 and over and permits dual enrollment with other insurance or welfare programs The VHA coverage functions much the same way in terms of payment and cove rage, and the majority of its enrollees are Medicare eligible. The main difference between the VHA IDS and Medicare is the lack of comprehensive
97 integration between the payer and delivery mechanisms. As a result the literature diverged from comparisons wi th the VHA as a result of its heavy focus on enrollment into managed care plans. MCOs are also, in effect, private organizations which operate separately from the payer system. While differences exist in regards to the VHA being a fully integrated nationw ide payer and provider of care and an entitlement program, this literature review remains critical to understanding the myriad of factors which may be affecting ATC. Further research will attempt to coalesce each of the ATC issues found within the public a nd private healthcare industries with ongoing VA research using a novel perspective; gauging these issues within the full spectrum of the healthcare delivery system from initial enrollment through realized use of services within and external to the VHA.
98 CHAPTER 4 THEORETICAL FRAMEWORK AND HYPOTHESES Overview determinants of enrollment into the VHA and service utilization within and outside the VHA, respectively. The study population is a nationally representative sample of all non enrolled and enrolled Veterans based in the year 2009. The theoretical model is derive d from the original version of the ABM and modified to include enrollment and utilization of services as outc ome variables The Anders e n model postulates predisposing characteristics, enabling cha racteristics and need for care (Andersen, 1995) Within this context, predisposing characteristics are broken into demographics (age, gender, biological traits), social structure (race, education, occupation and other factors that de and health beliefs (attitudes, knowledge and values, how people perceive providers, the hea lth system, their own health). Other than health beliefs, which are only moderately flexible to change, predisposing characte ristics are largely fixed. Next, enabling characteristics are those that facilitate people to access or utilization health care and are highly flexible to change. These include community factors ( i.e. the number of health providers available in a service a rea the number of facilities operating within a service area distance to available healthcare delivery resources and wait time s ) which must exist to enable utilization of he alth services as well as family/ individual level factors ( i.e. income levels in surance status degree of family support and structural capacity to include having transporta tion resources and a r egular source of care). Need s based factors are defined
99 care services through the health system Perceived should thcare resource or home remedy. E valuated need is more concretely defined than perceived need, yet both cont ain a major social component which necessitates assistance beyond what individuals may be able to provide on their own The goal of this research wa s to assess the magnitude effect of these factors on the decision to enroll and use care within and outside the VHA IDS The conceptual model is derived from this original theory but has been modified in an im portant manner. As stated previously the VHA IDS functions as a fully integrated payer and delivery system of health services This system requires enrollment and h ealthcare utilization to occur simultaneous or prior to the initial point of entry into the VHA IDS. Subsequent to th e first use of service, enrollment into the nation By integrating enrollment with healthcare utilization as dependent variables and operationalizing the predisposing, enabling, and need categories with relevant independent variables, the modified model evolves from the basis of the original theory. Figures 4 1 and 4 2 provide a graphical representation of the se modification s from the original ABM framework to the revised conceptual model While the ABM has been highly successfully in predicting utilization of health care resources it is not without limitation. Specifically, its usefulness falls short in predicting the volum e of care used, since frequency of use is commonly
100 Figure 4 1: Original ABM Theoretical Framework Figure 4 2: Modified Conceptual Model Healthcare Utilization Predisposing Factors Enabling Factors Need Factors Enrollment and Utilization within the VHA Predisposing Factors: Veteran Demographics Enabling Factors Veteran Undertanding of VHA entitlement Insurance Status Need Factors: Self Reported Health Status ADL/IADL Evaluated Need
101 referral and prescription powers. In other words, physicians have the final authorit y on healthcare use any time professional care is sought. Given the unique process of enrollment within the VHA IDS, predisposing, enabling, and need factors are both predictive of enrollment and th e decision to use care in the VHA IDS, regardless of wheth The ABM model was chosen precisely for this reason and provides a basis from which to draw inferences upon the strengths of the model. As such, any associations described within this context will remain logically indifferent to the need to measure more than one use. The intent of this research is to further develop the use of the ABM theory by employing enrollment into the VHA IDS and healthcare utilization within and external to the VHA IDS as de pendent variables By advancing the ABM theory we expound upon the impact of the theory both academically and practically. With respect and consideration of A customer satisfaction as a healthcare outcome, the potential remains for greater understanding of the impact individual level factors have on healthcare use and patient outcomes (Aday, 1974) As the ABM model continues to evolve we gain a more comprehensive appreciation of the complexities within the full spectrum of the healthcare delivery system Specific Aims While VA policies have been instrumental in expanding eligibility within the V eteran population, these actions have also created novel issues in accessibility, availability, accommodation, afford ability, an d the acceptability of services; factors w hich collectively effect ATC (Penchansky & Thomas, 1981) As a result the V A has bec ome highly active in its search for new and innovative ways of caring for a more
102 diverse population of eligible Veterans The diversity in needs has spawned the expansion of service lines previously unthought of within the VHA IDS. To meet these needs, mam mography, pre natal services, and PTSD/TBI therapy have been created along side traditional lines of treatments for cancer, heart disease, and stroke (Jha, et. al., 2003). While not all inclusive, these services provide evidence that the institution is str iving to meet these new challenges while also developing innovative strategies for reducing barriers to access through a more comprehensive care delivery model. Emerging from this new primary care model are several cutting edge programs designed to cut thr ough some of the most prevalent ATC barriers affecting the industry. Telemedicine has quickly become the hallmark of the VHA IDS and continues to show great promise in managing the needs of Veterans that are less mobile or geographically isolated ( Girard, 2007) In addition, reintegration programs were designed to Veterans back into the folds of civilian society using healthcare resources alongside other VA resources such as educational, job training, or home buying programs. These programs have become high ly integrated within the DoD and the community. In reference to resolving the healthcare needs of Veterans, the presence of reintegration programs have been highly effective in alleviating ATC barriers ( Sayer, 2010) Nevertheless, there remain ongoing issu es with how the VHA IDS identifies and engages with non enrolled Veterans in a more focused manner ( Longman, 2010) Studies on the V H A IDS have almost entirely presented data and research on utilization patterns and variability analysis within the enrolle d population To date, it has only recently begun to explore how Veteran characteristics, structural support, needs and ATC barriers of eligible non enrollees tie into the full spectrum of healthcare
103 delivery. Tantamount to the extensive research on utili zation variability, this analysis presents the case that enrollment variations and patterns of behavior represent fundamental opportunities to advance the discussion in a meaningful manner With the expansion of enrollment policies and outreach programs, the Secretary of the VA has stated that it will be able to meet its mission more effectively (Shinseki, 2010) Working in tandem with the Secretary of the VA, Congress enacted Section 805 of the Veterans Benefits Improvement Act of 2004 effectively mandating that the Secretary of the V A assess beneficiary awarenes s and understanding of its benefits and services ("Veterans Benefits Improvement Act of 2004," 2004) T he intent of these laws and policies are primarily designed to institutionalize incentives within the V H A to more efficiently and effectively allocate its resources for th e care of all eligible Veterans. This includes focusing enrollment efforts on higher rated PGs ("Veterans Benefits Improvement Act of 2004," 2004) Hence, it can be presumed that any ongoing VHA efforts would certainly benefit from a clearer understanding of Veteran enrollm ent variations and behaviors. To address the gaps in the literature and expand upon the ABM theory in a meaningful manner the following specific aims were employed to assess important differences with enrollment and utilization of healthcare resources wi thin the Veteran population: Specific Aim 1 The first aim is to determine whether enrolled and non enrolled Veterans display important differences in predisposing, enabling, and need characteristics as a means of e xplain ing their decision to seek enrollme nt and use care in the VA. The dependent variable ( DV ) reported
104 dichotomous response to e nrollment into the VHA IDS. The independent variables (IVs) include individual level Veteran characteristics categorized within the A BM. These variables are reference d as predisposing, e nabling, and n eed factors of which the predictor v ariable (PV) is represented within the enabling category a wareness of their entitlement of care within the VHA IDS. The PV was used to tes t the specific relationship mandated by Congress and found in the literature as having a plausible effect on enrollment and utilization along with other referenced predisposing, enabling, and needs based factors referenced in Appendix A This aim t est ed t he strength of the as sociation between enrollment into the VHA IDS and categorical differences in predisposing, enabling, and need factors. The approach employ s the use of a multivariable logistic regression model. C orrelates of likelihood were used to det ermine categorical differences between enrolled and non enrolled Veterans. These categorical relationships are highlighted in Figure 4 3 below. Figure 4 3 : Specific Aim One Categorical Relationships Specific Aim 2 The second aim demonstr ate s how utilization varies among the enrolled and non enrolled Veteran populations controlling for individual predisposing, enabling, and need s based factors Enrollment into the VHA IDS Predisposing Characteristics Enabling resources Need
105 The reported dichotomous response to utilization of health care resources. There are three DV s constructed by using three models reported use of ; the ER outpatient physician care and inpa tient care at any time within the past six months. The IVs reference predisposing, enabling, and needs base d factors as individual level Veteran characteristics categorized within the ABM. E nrollment status is considered in this model and constructed by combining four sets of insurance status options within the enabling characteristics of Veterans. These option sets are; Veteran enrolled in VHA IDS with other health insurance, Veteran enrolled in VHA IDS with no other health insurance, non enrolled Veteran with other health insurance, and non enrolled Veteran with no other health insurance Thi s assess ed whether Veteran enrollment and insurance status have a combined effect on healthcare utilization controlling for other differences in predisposing, enabling, and need factors. The approach employ ed the use of a multivariable logistic regression model. Following suit with the statistical strategies described in the first aim aim two contro l l ed for differences in predisposing, enabling, and need factors among enrolled and non enrolled Veterans. Aim two expounds upon the relationship of enrollment status and utilization by including enrolled Veterans with or without other insurance, non enrolled Veterans with other insurance and the uninsured as PVs within the enabling factor category Predicted probabilities were used for ease of interpretation A s reference, these categorical relationships are highlighted below in Figure 4 4
106 Figure 4 4. Specific Ai m Two Categorical Relationships A) ER Use B) Outpatient Use C) In patient Use ER Use Predisposing Characteristics Enabling resources Need Outpatient Use Predisposing Characteristics Enabling resources N eed Inpatient Use Predisposing Characteristics Enabling resources Need
107 Researc h questions In order to delineate between a ability to seek care based on his or her predisposing, enabling, need and preference for services, we first need to understand what factors influence the decision to enroll and what risk factors may sys tematically limit ATC within different groups Veterans throughout the full spectrum of healthcare delivery. The principal objective is to better understand the differences between Veterans that choose to enroll into the VHA IDS in comparison to other Veter ans who do not enroll. Further analysis is conducted on understanding of entitlements to care and available local resources within the context of other known access barriers. This research uses data gathered in the latest National Survey of Veter ans 2010 as a means of exploring the enrollment decision by examining the degree and manner in which enrollees can be distinguished from their non enrolled counterparts. Such analysis include s a thorough assessment of the following inquiries: Research Q ues tion 1 : W hich Veteran predisposing, enabling, and need characteristics are associated with enrollment into the VHA IDS? Research Question 2A : Do enrolled and non enrolled Veterans differ in their utilization of the ER, outpatient care, and inpatient ca re controlling for differences in predisposing, enabling, and need factors? Research Question 2B : How does the combined effect of Veteran enrollment status in the VHA IDS and having other health insurance affect utilization of the ER, outpatient care, and inpatient care controlling for differences in predispos ing, enabling, and need factors? Hypothesis 1.0 : T here are sign ificant differences between Veterans who enroll into the VA health care system versus non enrolled Veterans in terms of p redisposing, enabling, and need factors. Hypothesis 2.1 Veterans enrolled in the VHA IDS and who have other health insurance will categorically have a higher probability of ER use than privately insured Veterans controlling for differences in predisposing, enabling and need s based factors.
108 Hypothesis 2 2 Veterans enrolled in the VHA IDS and who have other health insurance will categorically have a higher probability of outpatient use than privately insured Veterans controlling for differences in predisposing, ena bling, and need s based factors. Hypothesis 2 3 Veterans enrolled in the VHA IDS and who have other health insurance will categorically have a higher probability of inpatient use than privately insured Veterans controlling for differences in predisposing, enabling, and need s based factors Hypothesis 2 4 V eterans enrolled in the VHA IDS and who do not have other health insurance will categorically have a lower probability of ER visits than privately insured Veterans controlling for differences in predis pos ing, enabling, and need s based factors. Hypothesis 2 5 Veterans enrolled in the VHA IDS and who do not have other health insurance will categorically have a lower probability of outpatient use than privately insured Veterans controlling for categoric al differences within predisposing, enabling, and need s based factors Hypothesis 2 6 Veterans enrolled in the VHA IDS and who do not have other health insurance will categorically have a lower probability of inpatient use than privately insured Veterans controlling for categorical differences within predisposing, enabling, and needs based factors. Hypothesis 2 7 Veterans not enrolled in the VHA IDS and who do not have other health insurance will categorically have a higher probability of ER use than pr ivately insured Veterans controlling for categorical differences within predisposing, enabling, and need s based factors. Hypothesis 2 8 Veterans not enrolled in the VHA IDS and who do not have other health insurance will categorically have a lower proba bility of outpatient use than privately insured Veterans controlling for categorical differences within predisposing, enabling, and need s based factors. Hypothesis 2 9 Veterans not enrolled in the VHA IDS and who do not have other health insurance will categorically have a lower probability of inpatient use than privately insured Veterans controlling for categorical differences within predisposing, enabling, and need s based factors.
109 CHAPTER 5 DATA AND METHODS This section describes the data sourc es, variables, and methodology used to conduct research on Veteran enrollment and utilization of health services. This is a cross sectional study based on publicly available data from the Nati onal Survey of Veterans, 2010. All data being employed within th is study are descriptive in nature and used to assess possible reasons for variations in Veteran enrollment and utilization. SAS Enterprise Guide was used to conduct descriptive analysis as well as multivariate surveylogisti c These analyses used weighted covariates drawn from the survey design of the NSV 2010. The NSV 2010 represents the sixth iteration of a nationwide survey of Veterans by the Department of the VA. Data from this instrument and subsequent versions of the NSV have been made available to the general public and widely used by health services researchers studying Veteran healthcare patterns of utilization. While some of the content changed between each NSV instrument, inquiries into Veteran healthcare utilizat ion and other Veteran issues remain comparable (NSV 2010; NSV 2001) Prior to the first NSV and for several decades leading up to its publication, the VA Reports and Statistics service published Veteran trend data as a means of assessing Veteran needs. Sin ce the 1970s, and approximately every decade since, the NSV has been used to assess the needs of Veterans and provide information for planners and policymakers on how they may develop future programs around these needs. These data are comprised of a nation wide snapshot of the Veteran population using the most sophisticated surveying techniques available at the time (VA, 2013) This research uses data exclusively derived from the NSV, 2010. The rationale for the exclusive use of the NSV
110 2010 as a data source enrolled and enrolled Veterans and the inclusion of key variables which may help explain enrollment into the VHA IDS, healthcare utilization patterns, and Veteran needs. The NSV 2010 Veteran response was completely voluntary at the time of inquiry and did not provide any financial incentive for participation. The survey was conducted using a mailed, self administered questionnaire and responses were completely anonymous. Address based sampling (ABS) and li st based sampling were employed to gain the highest possible response rate. Veterans were surveyed on a national level with a response rate of 66.7 percent. Individual survey items were weighted using the most recent Census and address data to permit compa risons among a nationally representative sample. For the purposes of this research, it is important to note that while the estimates are weighted, the confidence intervals for all regressions are not calculated with the weighted population data. Source of Data and Availability The data source is based entirely on the National Survey of Veterans, 2010 ( NSV 2010), a publicly available survey available through t he National Center for Veterans Analysis and Statistics (NCVAS 2013 ) The re are several iterations of the NSV spanning several decades with most published approximately every decade. The NSV has been used extensively by academics and VA operations as a means of improving its service to Veterans under the direction of U.S. Code Title 38, Section 527. U nique to NSV 2010 is the additional requirement from P.L. 108 454, Section services communication preferences (to improve outreach to VA beneficiaries), and future plans f or u se of benefits and services (VA, 2010) This iteration expanded the
111 surveyed population to include non enrolled Veterans as a means of assessing beneficiary awareness in a more comprehensive fashion. Previously, the customary focus was on enrolled Veterans and their demograp hic and utilization statistics. Subsequently, analysis was then able to deliver a more complete in depth perspective on important dissimilarities between both populations. The instrument was administered by mailed survey wit h the option to respond online. D ata collection efforts ran from October 16, 2009 to March 19, 2010 and conducted as a nationally representative, retrospective, survey offering a cross sectional view of Veteran data at one point in time Consequent to the changes in the focus of the sur vey, t he 2010 NSV different sampling frame and data collection modality than the prior surve (VA, 2010) The NSV 2010 employed the use of mai led surveys along side the option to respond online affording greater flexibility in how respondent s cho se to reply. Previous NSVs used random digit dialing (RDD) with telephone interviewing and mailed surveys As landline usage declined in beginning of this Century response rates among RDD administered surveys fell in kind M ailed surveys in the NSV 2010 w ere conducted in two phases. The first simply inquired as to whether or not the respondent was a Veteran and whether he or she would like to respond to a second, extended survey via mail or through a web based source. If the mailing was returned, another m ailing was sent out with information pertaining to the website hosting the instrument or with the entire instrument and a postage paid envelop for returning (NSV, 2010) Dependent Variable s The dependent variable s in this study are enrollment into the nat ional Veterans Health Administration and utilization of health services
112 full integration of its health information system, Veterans are identified as enrollees of the VHA and not as enrollees of a particular VHA hospital. There are several reasons to regard this is an important point of consideration in the definition of the dependent variable in this study. Specifically enrollees are viewed at the aggregate level as opposed to the institutional level. V ariability researc h is highly sensitive to the population of interest because it defines the measurement which, in turn, enables the researcher to draw infe renc es for policymakers to create, amend, or repeal policy. Any conclusions drawn from this analysis was then used as a means of improving the larger system of care as opposed to any other distinctively defined institution within the system. As a result, t he data used in this study are s imilar to the structure and process which was designed to enroll V eterans at the aggregate level Furthermore, it is certainly possible to draw enrollment and utilization variations from the institutional l evel using other supplementary data resources. If these were used, t he research objectives would then be mis aligned with the measurement. T hus, an aggregate level dependent variable is used to quantify these variations by assessing patient behaviors within a system of care rather than one based on geographic defined Veterans Integrated Service Networks (VISNs) at the i nstitutional level Finally, enrollment and utilization of health services matter because different groups of Veterans may be affected in different ways on the aggregate Previous studies on enrollment and utilization variations in public and private HCOs have focused on a plethora of factors ranging from patient characteristics to organizational characteristics to changes in state and federal policy
113 The revised conceptual ABM model references utilization and enrollment as dependent variables. Veteran en rollment is referenced as a dependent variable due to the simultaneous and concerted process of healthcare utilization and enrollment at the point of entry into the VHA IDS Utilization is broken down into three elements; ER utilization, outpatient utiliz ation, and inpatient utilization. Further detai ls on are outlined below and in Appendix A : Enrollment This variable is characterized as a dichotomous variable ( either ) referenced in item E1: ER u tilization This variable is characterized as a dichotomous variable ( either ) referenced in item Outpatient u tilization This variable is characterized as a dichotomous variable ( either ) referenced in item s E 5 and E7. E5 In the last six months, have you had outpatient care for doctor visits, urgent care, routine exams, medical tests, or shots? E7 inquire In the last six months, have you had outpatient visits for psychological counseling, therapy or mental health, or substance abuse treatment Inpatient u tilization This variable is characterized as a dichotomous variable ( either ) referen ced in item s E4 and E 6 E4 inquires: In the last six months, have you stayed in a hospital for medical or surgical care? E6. In the last six months, have you stayed in a hospital for mental health or substance abuse treatment? Independent Variables The i ndependent v ariables include predisposing, enabling, and need
114 characteristics which have been highly researched and found to influence utilization of health care services (Andersen, 1995) The explanatory variables which were used in the conceptual model are referenced in Appendix A and described as followed : Predisposin g c haracteristics. These f actors reference the least mutable among the three categories defined in the ABM. This revised conceptual model references all the subcategories of the predisposing category within the original ABM. Demographics include age and gender, s ocial structure inc ludes level of education, occupation ethnicity (identified as Hispanic) social networks (identified as period of service) social interactions (identified within combat exposure items) and culture beliefs (to included perceived quality of healthcare). Further details are outlined below and in Appendix A : Age This variable is a continuous variable characterized within the demographics element Age is calculated from the birth year of the Veteran and corresponds to item O2: This variable is shown to be a strong predictor of co morbidities and mortality as well as utilization of care. This variable has been re opera tionalized into four mutually exclusive categories of approximate equal distributions to include; Veterans who are 75 years of age and o lder, 65 75 years of age 55 64 years of age, and 55 years of age and younger The comparison category for this research is 65 75 years of age Veteran research due to their enrollment in Medicare and higher use of health services. Race This categorical variable is characterized in one of eleven separat e categories within the social structure category. Race has been shown to be an
115 important issue for gauging healthcare disparities and is a strong predictor of utilization. The VHA also functions as a primary healthcare resource for a large proportion of V eteran minorities. I tem mark all that apply. In essence, each corresponding race category represents a dichotomous response. This research uses Veterans who self operationalized variable representing any other races outside of being white or black. The decision to re operationalize this variable was made to generate greater power within this group as a whole. By sampling design, each of these three races represents non exclusive dichotomous categories and each race compares to all other races as comparison groups. Education This ordinal variable is characterized within the social structure element and consists of one of nine separate ca tegories ranging from less than high school to doctoral degree. Education has been shown to be a strong predictor of type of service utilization and is referenced in item The responses represent mutually exclusive categories referenced as having less than a High School education, having a High School Diploma/GED, having less than one year of College, having one or more years of DDS, DVM, LLB, JD ), and having a Doctorate degree (i.e PhD, EdD). Having less than an Associate degree was used as a comparison group for all categories.
116 Dates of Service There are three separate ways by which to measure this question; by deployment, by a pre defined period of service and by the date of discharge from serv ice The date of discharge from service is a continuous variable referenced in A6: In what year were you last released from active dut The period of service i s an ordinal variable consisting of one of nine separate options referenced in item A3 I tem A4 references a dicho Did you deploy in support of Operation Enduring This is a special interest item included within this survey as a result of mandates referenced in P.L. 108 454. Age and period of service are highly collinear and cause issues with over inflating standard errors in regression models. Deployment status has fewer issues with collinearity since it uses a smaller group of the total sample size of Veterans. In addition, the use of this item satisfi es the special interest of the aforementioned Congressional inquiry. As such, deployment status is used as a means of categorizing dates of service. This variable is characterized as a dichotomous variable (either ). Branch of Service This c ategorical variable is characterized in one of six separate elements within the social structure category. These consist of Air Force, Army, Navy, Marine Corps, Coast Guard, and Other (e.g., the Public Health Service, the Environmental Services Administrat ion, the National Oceanic and Atmospheric Administration, U.S. Merchant Marine) and r eferenced in item In which branch or branches did you serve on active duty? Due to power issues among Veterans classified as C oast Guard and Other this category onl y uses Air Force, Army, Navy,
117 and the Marine Corps Each mutually exclusive group represents a dichotomous variable such that each branch compares to all other branches as comparison groups. The rationale behind the inclusion of this variable can be attri buted to the different histories, traditions, and service member characteristics categorized within each branch of service. In line with the lack of mutability, or changeability, of this h of service much akin to an occupation Further, these factors accrete within each branch and form a culture and structure which likely influences processes and outcomes. For instance, the processes may vary by service in respect to policies relating to e ncouragement to seek information and services from the VA. As a result, these policies are likely influencing the outcome of the newly discharged Veteran; namely, to seek VA services and entitlements to receive care in the VHA IDS. Combat zone exposure T his variable is characterized within the social structure category as a dichotomous variable (either ) referenced in item A7: Did you ever serve in a combat or war zone? This variable represents an important inquiry into the use of health c are as Veterans present to physicians with health conditions relating to mental and physical conditions possibly sustained while being exposed to combat. Since these conditions are not likely to be considered a service connected disability unless the Veter an was seen and diagnosed by a VA physicia n, Veterans that had experienced some form of combat may be using more care or are in need of care but, do not have insurance coverage. Cowper (2005) found that over half of VA only users and dual users experienced some form of combat.
118 Exposure to mortality This variable is characterized within the social structure category as a dichotomous variable (either ) referenced in item your military service, were you ever exposed to dead, dying, or This variable represents an important inquiry into the use of health care as Veterans present to physicians with health conditions relating to mental and physical conditions possibly sustained while being exposed to combat. Since these c onditions are not likely to be considered a service connected disability unless the Veteran was seen and diagnosed by a VA physicia n, Veterans that had experienced some form of combat may be using more care or are in need of care but, do not have insurance coverage. Prisoner of War (POW) This variable is characterized within the social structure category as a dichotomous variable (either ) referenced in item Were you ever a prisoner of war? nto the use of health care as Veterans present to physicians with health conditions relating to mental and physical conditions possibly sustained while being exposed to combat. Since these conditions are not likely to be considered a service connected disa bility unless the Veteran was seen and diagnosed by a VA physicia n, Veterans that had experienced some form of combat may be using more care or are in need of care but, do not have insurance coverage. Environmental hazards exposure This ordinal variable i s characterized in one of five separate elements within the social structure category. These categories range referenced in item During your military service, were you ever exposed to environmental hazards s uch as Agent Orange,
119 chemical warfare agents, ionizing radiation, or other potentially toxic substances? to create a dichotomous variable was made on the exposed. This variable represents an important inquiry in to the use of health care as Veterans present to physicians with health conditions relating to mental and physical conditions possibly sustained while being exposed to environmental hazards. Since these conditions are not likely to be considered a service connected disability unless the Veteran was seen and diagnosed by a VA physicia n, Veterans that had been exposed to some form of e nvironmental hazard may be using more care. Enabling c haracteristics These factors reference the most mutable among the three categories defined in the ABM and provide guidance on possible areas of targeted assistance to the individual or system wide improvement. These reference person level or family level means income, health insurance, a regular source of care, ability to travel, and the extent and quality of social relationships. Further details are outlined below and in in Appendix A : Insurance status There are four mutually exclusive c ategories insurance status outlined in research Question 2A and referenced CURRENTLY covered by any of the following types of health insurance or health operationalized fro m nine available options as Veterans enrolled in the VHA IDS with other health insurance
120 Veterans enrolled in the VHA IDS only Veterans with Private Insurance Only, and U ninsured Veterans. Research Question 2B identifies eight mutually exclusive categori es in F1 Responses are re operationalized into eight mutually exclusive categories referenced a s either VHA IDS enrolled with Medicare, VHA IDS enrolled with Medicaid, VHA IDS enrolled with Medicare and M edicaid, VHA IDS enrolled with Private Insurance, Veterans with Medicare Only, Veterans with Private Insurance Only, Veterans enrolled in the VHA IDS only and U ninsured Veterans. Veterans with Private Insurance Only are the reference category for all com parisons. Marital status This is a categorical variable consisting of one of six options. Marital status has been shown to predict health status, mortality, and utilization and thus included in this analysis using item your current marital Responses included; now Married, Widowed, Divorced, Separated, Never Married, and Civil Commitment or Union. To generate greater power among the non married population, Veterans that were Widowed, Divorced, Separated, or Never Married were re cate u nderstanding of VA health b enefits This is an ordinal variable consisting of one of four options. I n re sponse to P.L. 108 454, Section 805 the NSV 2010 included items relating to VA benefits. Item B1, sub VHA entitl ement and available benefits as followed Please indicat e how much you understand about the following statements regarding the Veterans benefits provided by the Department of
121 Veterans Affairs (VA) : b. The Veterans healt are For ease of interpretation, A lot some healt h care benefits. Employment s tatus This is a categorical variable characterized in on e of four elements and assessed in item E3a i s a known enabling variable which identifies probable healthcare coverage and subsequent utilization of healthcare services. T his is Respondents were provided the following options; Working, or on paid v acation or sick leave from work; Not working, but looking for work; Not working and not looking for work Income This is an interval variable characterized in one of sixteen elements and is a known enabler of he althcare utilization. Item category represents the total combined income of all members of this family during the past 12 months? or more. S ervi ce connected Disability Rating (SCDR) These variables are referenced i n items C2 and C2a connected disability ratin es o option. he or she is asked to answer sub item C2a. This is an ordinal variable characterized in one of six elements and inquires: What is your current VA servi ce connected disability 0 percent 10 or 20 percent 30 to 40 percent 50 to 60 percent 70 percent or higher Both of these items are used in this dissertation as a means of assessing Veterans with different levels of disability ratings
122 rating whereas his distinction in semantics is tied to how disability payments are disbursed. Veterans with 0% SCDR are unable to obtain disability disbursements but, are consider ed eligible for a higher PG rating. At this point, the reader is likely wondering why t he SCDR is employed within the ABM as an enabler Primarily, the SCDR is thought to be an enabling factor as a result of variant access to care policies which have affe history of the VA Changes within VA enrollment policies present a peculiar dilemma when considering possible reasons for healthcare utilization and enrollment within the VHA IDS. Information variance is likely greater among different generations of Veterans as a direct result. Further, many non institutionalized elderly Veterans discharged well before the transformation of the VHA into a fully integrated health system. As the aggregate Veteran population continues to ag e greater than the overall US population it becomes paramount to understand how many Veterans remain unaware of their ability to enroll regardless of SCDR. Furtherm ore, there is a likely connection between different levels of SCDRs and the relative cost of care both within and external to the VHA IDS. It is expected that Veteran self selection is occurring among available health services options relative to the level of cost sharing within other plans, to include the VHA IDS. Although the effect of health p lan cost sharing variations on utilization is not being tested directly within this model, there is logic in the use of the SCDR as an enabling factor.
123 Health i nsurance This categorical variable represents the various types of multiple insurance coverag e options Veteran s may secure within and external to the VHA IDS. The effect of health insurance on the utilization of healthcare is well documented enabler within the ABM. This variable is characterized in one of eleven elements and assessed in item statu s is referenced in item F1; Are you CURRENTLY covered by any of the following types of health insurance or health coverage plans? Veteran respondents are given several options referenced in the following categories; Insurance throug h a current or former employer or union (of yours or another family member) Insurance purchased directly from an insurance company (by you or another family member) Medicare, for people 65 and older, or people with certain disabilities Medicaid, Medical Assistance, or any kind of government assistance plan for those with low incomes or a disability VA (including those who have ever used or enrolled for VA health care) TRICARE, TRICARE for Life or other military health care Indian Health Service Any o ther type of health insurance or health coverage plan Need s based characteristics Need is divided into two sub categories and assessed through several different items relating to perceived need and evaluated need As with all other ABM characteristics data used within this study, needs based factor s use self reported data H ealth status is measured through se veral categorical and dichotomous variables quantified responses to general health status and function al status Central to t he concept of need is the perception and evaluation of what it constitutes. Both are necessary in gaining the full perspective of
124 the modern defini tion of evidence based medi cine and form the foundation of the primary care model. Evaluated need provides the Veteran with the necessary context by which to assess the severity of his or her health status. Perceived need provides the clinician with the crucial context by which to u nderstand his or her patients. Likewise, by accounting for patient preferences and clinician judgment, the VHA IDS is better informed in how it addresses the needs of its Veteran population. General health status This is an ordinal variable characterized in one of five elements and assessed in item status In general, would you Excellent, perceived need for healthcare. Functional status Further consideration is given to the limited focus on e valuated need within this study for two reasons. As previously described the SCDR is already employed as an enabling factor and cannot be used withi n more than one ABM factor. Opportunely evaluated need is not entirely excluded from this study. To compensate for a lack of data on specific clinician evaluated conditions self reported functional stat us provides us with a practical proxy in the absence of these data. A ctivities of daily living (ADL) and instrumental activities of daily living (IADL) may be operationalized Section D These are referenced in items D 3 and D4. D3 inquires; In the past week, how much assistance did you require in the following activities due to a health condition? and
125 assistance, I can do with some assistance, I am completely depende nt on assistance, D4 inquires; Are you currently in need of the aid and According ly, t here are six basic self care ADL tasks ; eating, bathing, dressing, toileting, ambulation (i.e. walking), and continence ( Bookman, et. al., 2007) IADLs are not necessary used to define fundamental functioning, but permit the individual to live independently in a community (Bookman, et. al ., 2007) IADLs include activities such as telephone, and taking medications properly, all of which are represented within the NSV 2010 data. While these self reported items do n ot imply that need is evaluated, we can infer in a conservative manner that responses to D3 such as pletely dependent on assistance hig hly predictive of evaluated need. Furthermore, Veterans claiming that they are dependent on the aid of another person or completely dependent on assistance affect ed approximately 33 % of the nationally represented sample of Veterans surveyed within the NSV 2001 (NSV 2010) With an aging Veteran demographic, all indicators poi nt to a likely increase in the percentage of Veterans needing some form of assistance in the NSV 2010 While functional status is not all inclusive to the entire sample, evaluated need is represented among many Veterans In combination with perceived need, the use of functional status as an instrumental variable for evaluated need provides a more comprehensive pers pective on needs based factors.
126 Study Design and Methodology Descriptive analyses of variable variations were conducted to answer research Q ues tion 1 and H ypothesis 1 0 The selection of the variables of interest was based on a theoretical and empirical foundation described in detail throughout this document. Each of the categorical variable s wa s fitted within the ABM by regressing the DV on eac h of the IVs previously outlined in this research. Statistical significance is set at p < .05. Each IV is measured within each of the ABM predisposing, enabling, and need individual variable categorized under the enabling characteristics of the Veteran. The same method is used to assess significance of each of the remaining IVs within each of the previously prescribed ABM factors. The estimated logistic regression coefficie nts are used to determine the significance of the variables at p < .05 Once statistical significance was determined for each individual variable, a complete model was tested using predisposing, enabling, and need factors. Each factor was then tested indivi dually and collectively to determine their effect on enrollment. The predictor of interest highlighted in this endeavor to determine the strength of its association in the model. The sign of each coefficient wa s assessed to d etermine its effect on enrollment. In addition, a likelihood ratio test was used to judge the overall p redictability of the full model by assessing the area under the curve of the receiver operating characteristic The results of this test found that 91 pe rcent of enrollment could be explained through the use of each IV. Tests for correlation were conducted between predictor v ariable 1 and predictor variable 2 referenced in Appendix A T ests for correlation were used to det ermine the level of dependence between each variable. Both predictor variables were used as
1 27 enabler factors within the first model. Further analysis using tests for correlation are used to determine if any other IVs are highly correlated and remain in the model. Secondary analysis is used to test goodness of fit for hypothesis 1.0. This is conducted using squared test as a means of assessing possible variations in enrolled versus non enrolled Veteran characteristics at multiple levels of anal ysis. Predisposing and need factors are assessed using the aforementioned individual level variables categorized within each respective factor. T he primary predictor of intere is assessed as an individual variable categorized un der the enabling characteristics of the Veteran. Likewise, every other independent variable is assessed individually within the predisposing enabling, and need categories of the theoretical framework. A full multivariate logistic model (MLR ) is tested wi thin each predisposing, enabling, and need category based on the ABM theory Several iterations of these analyses are used to answer hypothe sis 1.0 and draw important inferences on magnitude of effect. This analysis form s the basis by which to justify the development of succeeding model iterations and examinations of possible associations between predisposing, enabling, and need categories and the likelihood of enrollment. MLRs are conducted on the dependent variable (enrollment) to assess whether the sign awareness to their healthcare benefit along with other predisposing enabling, and needs based factors. In addition to each of the control factors, only those variables tha t are foun d to be significant in the MLR analyses using coefficients of determination (R 2 )
128 are entered into the multivariate logistic regression equation for the dependent variable. Statistical significance is determined at a representative level of 5%. F urther analysis using tests for correlation are used to determine if any other IVs are highly correlated with each other and remain in the models. Any IVs found to be highly related or insignificant at the .05 threshold are eliminated The remaining covari ates form a reduced model used to determine probability of enrollment Research q uestion 2 em ploy s the use of predictive analysis. MLR is used to address hypotheses 2 .1 2. 9 using a similar methodological strategy outlined for research question 1 Correlat es of likelihood (i.e. R 2 ) are used to uncover how each IV affects emergency utilization, primary care utilization and inpatient utilization respectively Utilization effects were determined among enrolled and non enrolled Veterans and other predisposing enabling, and need factors Utilization effects were determined among four mutually Sensitivity analysis was conducted on effects within the multiple coverage categories to capture which groups are mo re likely to use care. S equential conditional MLRs are conducted on each of the three dependent variables ( i.e. ER utilization, outpatient utilization, and inpatient utilization) to assess whether the significant bivariate correlations remained significan t after controlling for predisposing, enabling, and needs based factors. In addition to each of the control factors, only those variables that were foun d to be significant in the MLR analyses using coefficients of determination (R 2 ) are entered into a redu ced MLR equation. Statistical significance is determined at p < .05 Likelihood ratio tests were used to judge the overall p redictability of each full model by assessing the area under the curve of the receiver
129 operating characteristic. The results of this t est found that 70 percent of ER model could be explained through the use of each selected IV, 71percent of the outpatient model could be explained through the use of each selected IV, and 67% of the inpatient model could be explained through the use of eac h selected IV. S equential conditional MLRs are conducted on each of the three dependent variables ( i.e. ER utilization, outpatient utilization, and inpatient utilization) to assess whether the significant bivariate correlations remained significant after controlling for predisposing, enabling, and needs based factors. In addition to each of the control factors, only those variables that were foun d to be significant in the MLR analyses using coefficients of determination (R 2 ) are entered into a reduced MLR equation for each of the DVs Statistical significance is determined at p < .05 Empirical models Based on Hypothesis 1 .0 in Chapter 3, the first empirical model is: Enrollment into the VHA =f (predisposing, enabling, and need characteristics of the Vetera n). There are significant differences between Veterans who enroll into the VA health care system versus non enrolled Veterans in terms of predisposing, enabling, and need factors Based on H ypothesis 2 .1 the next empirical model is : Emergency room use ( ER use) = f ( enrolled Veterans + other insurance (Medicare, or Medicaid, or Private, or Medicare and Medicaid), controlling for predisposing, enabling, and need s based factors )) Enrolled Veterans with other insurance will yield statistically significant hig her odds of ER use than privately insured Veterans. Based on H ypothesis 2 .2 the next empirical model is:
130 Outpatient care service use (OP use) = f ( enrolled Veterans + other insurance (Medicare, or Medicaid, or Private, or Medicare and Medicaid), controlli ng for predisposing, enabling, and need s based factors )) Enrolled Veterans with other insurance will yield statistically significant higher odds outpatient care than privately insured Veterans Based on hypothesis 2 .3 the next empirical model is: Inpatie nt care service use (IP use) = f ( enrolled Veterans + other insurance (Medicare, or Medicaid, or Private, or Medicare and Medicaid), controlling for predisposing, enabling, and need s based factors )) Enrolled Veterans with other insurance will yield statis tically significant higher odds of inpatient care than privately insured Veterans Based on Hypothesis 2 .4 the next empirical model is: Emergency room use (ER use) = f ( enrolled Veterans + no other insurance controlling for predisposing, enabling, and nee d s based factors ). Enrolled Veterans with no other insurance will yield statistically significant higher odds of emergency room use than privately insured Veterans Based on Hypothesis 2 .5 the next empirical model is: Outpatient care service use (OP use) = f ( enrolled Veterans + no other insurance controlling for predisposing, enabling, and need s based factors ). Enrolled Veterans with no other insurance will yield statistically significant higher odds of outpatient care use than privately insured Veterans Based on Hypothesis 2 .6 the next empirical model is:
131 Inpatient care service use (IP use) = f ( enrolled Veterans + no other insurance controlling for predisposing, enabling, and need s based factors ). Enrolled Veterans with no other insurance will yield st atistically significant lower odds of inpatient care use than privately insured Veterans Based on Hypothesis 2 .7 the next empirical model is: Emergency room use (ER use) = f ( uninsured Veterans controlling for predisposing, enabling, and need s based fact ors ). Uninsured Veterans with no other insurance will yield statistically significant higher odds of emergency room use than privately insured Veterans Based on Hypothesis 2 8 the next empirical model is: Outpatient care service use (OP use) = f ( uninsur ed Veterans controlling for predisposing, enabling, and need s based factors ). Uninsured Veterans with no other insurance will yield statistically significant lower odds of outpatient care use than privately insured Veterans Based on Hypothesis 2 9 the nex t empirical model is: Inpatient care service use (OP use) = f ( uninsured Veterans controlling for predisposing, enabling, and need s based factors ). Uninsured Veterans with no other insurance will yield statistically significant lower odds of inpatient car e use than privately insured Veterans
132 CHAPTER 6 RESULTS Descriptive Findings for Research Question One Research question one inquired into which Veteran predisposing, enabling, and need characteristics are associated with enrollment into the VHA IDS. To fully answer this question, descriptive statistics were first used to show the distributions in the sample and the weighted population measures within each respective independent and dependent variable category. Next the model was assessed for sig nificant differences between Veterans who enroll into the VA health care system versus non enrolled Veterans in terms of predisposing, enabling, and need factors. These differences were assessed using p values and chi square within the 95% confidence inter val. Finally, a fully adjusted model was used to measure the likelihood of enrollment across the ABM Tables 6 1 through 6 5 reference the distributions of the DV and each of the IVs. enly generate power in further analysis and so, whites and blacks were used in multivariate analysis going forward. Enabling characteristics are approximately evenly dis tributed throughout each representing IV. The exceptions within this category are enrolled Veterans with Medicare and Medicaid, Veterans with Medicaid, and non enrolled Veterans with Medicare and Medicaid. Further analysis employ ed these IVs but, any stati stically significant findings were interpreted with caution. Need characteristics are approximately evenly distributed throughout each representing IVs and thus, all were used for further analysis. The DV is also approximately evenly distributed and repres ents two meaningful comparison groups.
133 Secondary descriptive analysis was used to assess differences between enrolled and non enrolled Veterans across all previously described ABM factors These data are provided in tables 6 6 through 6 8 below. The pre were found to have significant differences between enrolled and non enrolled Veterans with a p value of .4163 and service categories with varying p values. The Air Force category yielded a p value of 002; the Army, a p value of .0003; the Navy, a p value of .0021; and the Marines, a p value of .0008. Given that race, period of service and othe r culturally identifiable experiences such as being deployed are aptly represented, service was not be used in subsequent analyse s. Modeling Veteran Characteristics on Enrollment Status Hypothesis 1 was used to assess the facto rial associations between predisposing, enabling, and need based characteristics on enrollment Predisposing, enabling, and need factors were all found to be associated with enrollment into the VHA IDS. fits were 9.6 times more understanding of their entitlement is a likely driver of enrollment into the VHA IDS. In addition, Veterans who are younger than 5 5 years of age, have lower incomes, have a greater understanding of VHA benefits, and have low self reported health status, are most likely to enroll, controlling for other predisposing, enabling, and need factors. For enabling characteristics Veterans with that were financially impoverished were highly likely to enroll. And those that were
134 married or employed were less like ly to enroll. For needs based factors, Veterans with good, fair, and poor health were more likely to enroll as health status diminished in comparison to those in excellent health. Specifically, for predisposing factors, the youngest group of Vet erans was 1.4 times more likely to enroll than those 65 74 years old. Additionally, if Vets were ever exposed to hazardous materials, they were 1.3 times more likely to enroll. For enabling factors, Veterans enrolled in Medicare were 2.3 times more to enroll than Ve terans enrolled in private insurance Veterans with incomes les s than $10K were 1.7 times more likely to enroll than Veterans with incomes over $60K. Veterans with incomes ranging from $10,001 $19,999 were 2.7 times more likely to enroll than Veterans with incomes over $60K. Veterans with incomes ranging from $20,000 $ 29,999 were 2.6 times more likely to enroll than Veterans with incomes over $60K. Veterans with incomes ranging from $30,000 $59,999 were 1.6 times more likely to enroll than Veterans with inc omes over $60K. Veterans receiving payment for a SCD were 6.2 times more likely to enroll than Veterans who did not receive payment. Veterans with good health status were 1.7 times more likely to enroll than Veterans than Veterans who reported excellent h ealth. Veterans with fair health status were 2.8 times more likely to enroll than Veterans who reported excellent health. Veterans with poor health status were 4 times more likely to enroll than Veterans who reported excellent health. Table s 6 9 through 6 11 referenced below offer a greater detail of this analysis. Descriptive Findings for Research Question s 2A and 2B Research question s 2A and 2B inquired into whether enrolled and non enrolled Veterans differ in their utilization of the ER, outpatient care and inpatient care
135 controlling for differences in predisposing, enabling, and need factors To fully answer this question, descriptive statistics were first used to show the distributions in the sample and the weighted population measures within each res pective independent and dependent variable category. These weighted data indicate that 75.26 percent of all Veterans had some form of outpatient care use, 16.52 percent had some form of ER visit, and 12.74 percent used some form of inpatient care. Each of the DVs was approximately evenly distributed and represents meanin gful comparison groups. Table 6 12 reference s the unweighted and weighted distributions of each of the se DV s. Next, the model used bivariate analysis to assess significant differences betwe en each utilization category in terms of predisposing, enabling, and need factors. These differences were assessed using p values and chi square within the 95% confidence interval. Secondary descriptive analysis was used to assess differences across all th ree previously described utilization categories and each ABM factor. The complete bivariate descriptive analysis references the unweighted and weighted distributions of each of the IVs as related to each type of use in Tables 6 13 to 6 21 below. Table 6 12 references the bivariate analysis of DVs to enrollment status. Finally, a fully adjusted model was used to measure the likelihood of utilization across a fully adjusted ABM. Modeling Veteran Characteristics on Utilization Aim 2 tests the Veteran enrollme nt effect on emergency room, outpatient, and inpatient care utilization controlling for other predisposing, enabling, and need factors. Research Question 2A inquires further into the full spectrum of healthcare delivery using Veterans insurance status alon g with other predisposing, enabling, and need factors to describe associations with higher or lower likelihood of utilization. The basis for this inquiry is to ascertain where ATC challenges may be occurring in the delivery of care
136 within a primary care me dical model. To do this, insurance status was operationalized under the enabling factor of the ABM as such: Four mutually exclusive predictor variable definitions were created for Veterans with multiple coverage, Veterans enrolled in the VHA IDS only, Vet erans with private insurance only, and uninsured Veterans. Veterans with multiple coverage were comprised of Veterans enrolled within the VHA IDS and had Medicare and/or Medicaid and/or TRICARE coverage. Veterans enrolled within the VHA IDS consisted of Ve terans who use the VHA IDS only and were not covered by any other insurance. Veterans with other health insurance were comprised of Veterans with either individually purchased insurance (to include TRICARE) or insurance purchased through their employer or COBRA. Finally, uninsured Veterans consisted of those who responded as being `uninsured` and cross referenced with any reported coverage under any category of insurance. When there was other reported coverage, these Veterans were then considered insured wi thin the appropriate category and the remaining Veterans remained as `uninsured`. These associations ar e further referenced i n Figure 6 1 below: Research Question 2A Results Odds of ER Use Under Research Question 2A : Controlling for other predisposing, ena bling, and need factors, enrolled Veterans with multiple coverage were 1.4 times more likely to use the ER then Veterans with private insurance. Uninsured Veterans were 1.6 times more likely to use the ER then Veterans with private insurance. When compare d to Veterans 65 74 years of age the youngest cohort of Veterans were 1.4 times more likely to use the ER and Veterans over 75 years of age were 1.3 times more likely to use the ER. As expected for needs based factors,
137 Veterans with a perception of declin ing health were more likely to use the ER. Specifically, Veterans who rated their health as good were 1.8 times more likely to use the ER than those with excellent health. Veterans with fair health were 2.9 times more likely to use the ER than those with e xcellent health. Veterans with poor health were 6.2 times more likely to use the ER than those with excellent health. Additionally, Veterans who had the presence of an ADL were 1.7 times more likely to use the ER than those without. Table s 6 23 through 6 2 5 referenced below offers a greater detail of this analysis. Table 6 22 offers a greater detail of the DV analysis for Questions 2A and 2B Odds of Outpatient Care Use Under Research Question 2A : Controlling for other predisposing, enabling, and need fact ors, enrolled Veterans with multiple coverage were 1. 8 times more likely to use outpatient then Veterans with private insurance. There were no statistically significant findings in the VA only cohort nor, among the uninsured When compared to Veterans 65 7 4 years of age the youngest cohort of Veterans were 23 percent less likely to use outpatient care and Veterans 55 to 64 years of age were 52 percent less likely to use outpatient care When compared to Veterans without a degree, Veterans with an Associate Degree or higher were 1. 7 times more likely to use outpatien t care. For enabling factors, Veterans with incomes less than $10K were 37 percent less likely to use outpatient care than Veterans with incomes over $60K. Across both sets of measured incomes, Veterans wit h incomes between $10K to $60K, were 36% less likely to use outpatient care than Veterans with incomes over $60K. Veterans receiving
138 a SCD payment were 1.8 times more likely to use outpatient care than Veterans without. For needs based factors Veterans who rated their health as very good were 1.4 times more likely to use outpatient care than Veterans in excellent health. Veterans who rated their health as good were 1.7 times more likely to use the ER than those with excellent health. Veterans with fair health were 3.9 times more likely to use the ER than those with excellent health. Veterans with poor health were 9.1 times more likely to use the ER than those with excellent health. Detailed analysis of these effects is referenced in Table s 6 26 through 6 28 Odds of Inpatient Care Use Under Research Question 2A : There were no statistically significant enabling factors however, predisposing and needs based factors yielded significant odds of use for inpatient care. Specifically, w hen compared to Veterans 65 74 years of age the youngest cohort of Veterans was 36 percent less likely to use inpatient care. The oldest cohort of Veterans was 1.7 times more likely to use inpatient care than Veterans 65 74 years of age For needs based factors, Veteran s who rated their health as very good were 1.7 times more likely to use inpatient care than Veterans in excellent health. Veterans who rated their health as good were 3.5 times more likely to use inpatient care than those with excellent health. Veterans wi th fair health were 5.4 times more likely to use inpatient care than those with excellent health. Veterans with poor health were 12.4 times more likely to use inpatient care than those with excellent health. Additionally, Veterans with an ADL were 1.6 time s more likely to use inpatient care than those without. Detailed analyses of these effects are referenced in Table s 6 29 through 6 31
139 Research Question 2B Results Further analysis on the insurance a ffect was conducted to ascertain whether specific categor ies of Veterans were showing higher odds of use, particularly among those with multiple coverage. Given the evidence of highly significant associations of utilization among Veterans who had multiple coverage, insurance variables were further specified and re operationalized from four to nine mutually exclusive categories. These were created as a means of uncovering other meaningful associations within the broader insurance category of enabling factors Veterans were divided into nine mutually exclusive cate gories consisting of the uninsured, Veterans enrolled in Medicare only, those enrolled in Medicare and Medicaid only those enrolled in private care only, and for Veterans enrolled within the VHA IDS and Medicare, Veterans enrolled within the VHA IDS and M edicaid, Veterans enrolled within the VHA IDS, Medicare, and Medicaid, and Veterans enrolled within the VHA IDS and TRICARE. These insurance categories are highlighted below in Figure 6 2 along with other predisposing and need factors within the ABM. Odds of ER Use Under Research Question 2B : Based on this analysis, several important associations became apparent within these separate permutations of multi covered Veterans and Veterans with Medicare and/or Medicaid only. As expected, uninsured Veterans were 1.6 times more likely to use the ER then Veterans with private insurance after controlling for other predisposing, enabling, and need factors. What is interesting to note is that the rest of the significant findings for odds of ER use were very similar to uninsured Veterans. Specifically, enrolled Veterans with private coverage were 1.7 times more likely to use the ER then Veterans with private insurance only. In addition, enrolled Veterans with Medicare
140 coverage were 1.5 times more likely to use the ER th en Veterans with private insurance. When compared to those 65 74 years old, the youngest cohort of Veterans was 1.6 times more likely to use the ER. Veterans that were in poor health were 5.6 times more likely to use the ER than those in excellent health, Veterans in fair health were 2.7 times more likely to use the ER than those in excellent health, Veterans in good health were 1.6 times more likely to use the ER than those in excellent health, and Veterans in very good health were 1.2 times more likely t o use the ER than those in excellent health. Additionally, Veterans that had the presence of an ADL were 1.8 times more likely to use the ER than those without an ADL. Detailed analysis of these effects is referenced in Table s 6 32 through 6 34 Odds of Ou tpatient Care Use Under Research Question 2B : Outpatient care use was found to be highly significant among several groups of Veterans pointing to the possibility of overuse or underuse within the primary care setting. Specifically, enrolled Veterans with M edicare coverage were 1.5 times more likely to use outpatient care then Veterans with private insurance. Enrolled Veterans with private coverage were 1.6 times more likely to use outpatient care then Veterans with private insurance. Enrolled Veterans with MEDICARE coverage were 3.2 times more likely to use outpatient care then Veterans with private insurance. Enrolled Veterans with Medicare and Medicaid coverage were 7.2 times more likely to use outpatient care then Veterans with private insurance. When co mpared to those 65 74 years old, the oldest cohort of Veterans was 1.3 times more likely to use outpatient care. When compared to Veterans without a degree,
141 Veterans with an Associate Degree or higher were 1. 5 times more likely to use outpatient care. Vet erans exposed to dead, dying, or wounded people were 1.2 times more likely to use outpatient care than those that had not been exposed. Veterans exposed to hazardous chemicals were 1.3 times more likely to use outpatient care than those that had not been e xposed. Veterans with incomes les s than $10K were .6 times less likely to use outpatient care than Veterans with incomes over $60K. Veterans with incomes ranging from $10,001 $19,999 were less likely to use outpatient care than Veterans with incomes over $ 60K. Veterans with incomes ranging from $30,000 $59,999 were .75 times less likely to use outpatient care than Veterans with incomes over $60K. Veterans receiving payment for a SCD were 1.7 times more likely to use outpatient care than Veterans who did not receive payment. Veterans with very good health status were 1.5 times more likely to use outpatient care than Veterans who reported excellent health. Veterans with good health status were 1.9 times more likely to use outpatient care than Veterans who rep orted excellent health. Veterans with very fair health status were 4.4 times more likely to use outpatient care than Veterans who reported excellent health. Veterans with poor health status were 9.8 times more likely to use outpatient care than Veterans wh o reported excellent health. Additionally, Veterans with an IADL were .53 times less likely to use outpatient care than those without an IADL. Detailed analyses of these effects are referenced in Table s 6 35 through 6 37 Odds of In patient Care Use Under R esearch Question 2B : Predisposing and needs based factors were the only statistically significant factors for inpatient care use. The oldest cohort of Veterans was 1.4 times more likely to
142 use inpatient care than those aged 65 74 years of age. The youngest cohort of Veterans was .64 times less likely to use inpatient care than those aged 65 74 years of age. Veterans that were in poor health were 11.6 times more likely to use inpatient care than those in excellent health, Veterans in fair health were 5.3 tim es more likely to use inpatient care than those in excellent health, Veterans in good health were 3.3 times more likely to use inpatient care than those in excellent health, and Veterans in very good health were 1.7 times more likely to use inpatient care than those in excellent health. Additionally, Veterans that had the presence of an ADL were 1.7 times more likely to use inpatient care than those without an ADL. Detailed analyses of these effects are referenced in Table s 6 38 through 6 40 These findings suggest that insurance status is strongly associated with utilization. Veterans with more insurance coverage are more likely to use ER and outpatient care. Enrolled Veterans with private insurance and enrolled Veterans with Medicare have similar odds of u sing the ER as the uninsured. Furthermore, other predisposing and need characteristics are predictive of ER, outpatient, and inpatient care use. Figure 6 3 referenced below offers a greater detail of the full analysis of the associations between insurance and utilization with statistically significant findings highlighted with an asterisk. Limitations As a result of the cross sectional data collection design, the outcome and predictor variables were measured simultaneously and, as such, causation cannot be derived from these results. Secondly, although data was collected from a nationally representative sample of the veteran population, generalizability is limited to within the VA population only. There is the potential for recall bias on the dependent varia ble since
143 T his is a weak assumption given the difficulty in know ing whether or not the Veteran is enrolled and had visited the VA at least once. Of the insurance variables ope rationalized with the analysis for the second aim, there were two categories which yielded low samples in comparison; VA + Medicare + Medicaid and VA + Medicaid These variables were kept within the analysis as a means of expanding the concept of what is c ommonly refer r e d dually individuals. This term is being phased out within the VA precisely due to the changing nature of the insurance variable (VHA, 2013). For example, dually enrolled may be referred to as individuals enrolled in the M ed icare and Medicaid or enrolled in the V HA IDS and Medicare. The inclusion of these variables shows that there are more interpretations of how Vetera ns may be enrolled in different types of insurance options. In the case of Veterans simultaneously enrolled in the VHA, Medicare, and Medicaid, there is consideration of categories of Veterans with more than two insurance options. Though utilization analysis did not yield any statistically significant findings on V eterans dually enrolled in the VHA IDS and Medic aid there were statistically significant odds of outpatient use for Veterans enrolled in the VHA IDS, Medicare, and Medicaid category. Due to a low sample size in comparison to the Private insurance reference category, the interpretation of the results in the VHA IDS, Medicare, and Medicaid category must be exercise d with some caution All data is de identified to protect the anonymity of respondents. As a result, we cannot verify whether enrollment actually occurred nor geographic identifiers which could have been used to determine unique market conditions such as employment,
144 mean income, a nd organizational competition. There is a lack of data on several independent variables which limits interpretability of findings. Finally, homeless non enrollees, who a ccount for one of the most vulnerable sub populations, may not have been surveyed due to their transient behaviors, limiting the effectiveness of a mailing strategy. The NSV 2010 Final Report has corrected for non response bias by adjusting survey response s using sampling weights cross and Census region. In turn, the respondents to the NSV 2010 would be representative of the full population of Veterans. Sampling error was also adjusted within these weights by creating con fidence intervals for single survey items and determining statistical significance across each subpopulation. There is a l ack of data on certain IVs to include c o morbidities, diagnosis codes, and disease states Furthermore, there is no manner of linking data with other available data in medical and VA Priority Group records, geographic residence, co payment structure, or market conditions to validate Veteran responses or draw clearer associations to enrollment and utiliz ation. If these data were available it may permit stronger inferences into why Veterans cho o se multiple coverage for various reasons ranging from greater financial security, better access to care, better quality of care, or other plausible explanations In effect, the use of more specific health and geographic IVs may help answer some of the reasons underlying the self selection effect. As an example, sicker VHA enrollees may be more likely to purchase or enroll in other available insurance coverage for added financial security resulting in greater odds of utilization. On the other hand, there may be a preference selection effect driving certain
145 V H A enrollees who prefer non VA care to purchase other available insurance and use less VHA care Also, if Priority Group rating data were available the payment structure would be clearly d efine d within each Veteran group and could be used to determine the level of accessibility to care. As a result ER and outpatient care may change in relation to co payments associated with each Priority Group and adjusted for other factors. There is some recall bias on the dependent variable s used in this study since respondents were asked to recall enrollment status into the VHA IDS and utilization of care within the primary care setting as well as inpatient ca re and the ER within the past six months Since these responses were dichotomous, the risk or data integrity is only To compensate for missing data prior to analysis, multiple imputation was used within the survey design and survey weights were used to reduce the risk of biased parameter estimates within the Veteran population. Given these precautionary steps in the survey design, there remained some missing data within the responses of Vetera ns who partook in the survey. D escriptive analysis was used to determine the degree of missing independent variables within each categorical ABM factor. As a result, Veterans who were deployed to OEF or OIF were POWs, and those that were categorized as missing data and an overall low level of self identification. Of the remaining missing data, t here were no more than 10% missing within each IV and o f these data. Initial MVR analysis using SAS Enterprise yielded a lower sample size as a result of missing data within some observations. R espondents that did not answer every survey item were subsequently dropped from the multivariate analysis. To assess
146 whether this had an ef fect on the outcome within each model, a chi squared analysis was used The data were examined for the pattern of missingness within responses following the method developed by Allison (2002). Briefly, this approach involves defining an indicator variable for the missing patterns with those who complete all responses as the reference category. Veteran non respondents were then compared to Veteran respondents to determine if these two groups differed in any statistically significant manner within each depen dent variable; enrollment, ER use, outpatient use, and inpatient use. More specifically, if the non missing covariates of the responders and non responders were similar, we can deduce that the non responders would have likely answered in a similar way to t he responders and therefore any a nalyses based on the responders only is likely to be unbiased and is not influencing the DV responses A separate analysis using a sub set of responders and non responders within each dependent variable was used to compare the characteristics within each of these groups as the basis for subsequent decisions of validity In terms of responders versus non responders, the distribution of each covariate was not different in any statistically significantly manner. T he results of these missing data analyses determined that non responses within each categorical DV response were missing at random Subsequent to this assessment, there is some validation to the original results of the multivariate analysis. For more detail on the dist ributions and chi squared analyses, please r eference Appendix B
147 Figure 6 1 : Operationalized Enabling Variables under Research Question 2A Enabling Predisposing Factors Need Factors
148 Figure 6 2 : Operationalized Enabling Vari ables under Research Question 2B Utilization Enabling Need Factors Predisposing Factors
149 F igure 6 3 : O dds Ratios of Insurance Categories 1.6* 1.7* 1.5* 1.7* 1.6* 3.3* 7.9* 0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 ER OUTPATIENT INPATIENT Represents statistically significant odds ratios of utilization by insurance status
150 Descriptive Measures Table 6 1. Dependent Variable for Research Question One Variable Unweighted ( n =7996) Unweighted Percent Weighted ( N =~20MIL) Weighted Percent Enrolled in the VHA* Yes No 2 319 5 716 28.86 71.14 5466679 14900742 26.84 73.15 * 3 9 missing after adjusting for enroll ment and VHA insurance variables
151 Table 6 2. Predisposing Characteristics of Veterans Variable Unweight ed ( n = 799 6 ) Unweighted Percent Weighted ( N =~20 MIL ) Weighted Percent Age Categories 75 years of age and older 65 75 year s of age 55 64 years of age Y ounger than 55 years of ag e Race Categories White Black Education Categories No College Degree College Degree and Beyond Service Categories Air Force Army Navy Marine Deployed to Operat ion Enduring Freedom (OEF) or Operation Iraqi Freedom (OIF) Yes No Served in Combat or War Zone Yes No Exposed to Dead/Dying/Wounded Yes No Exposed to Hazardous Chemicals Yes No 2015 1932 2339 1749 7108 543 5935 2605 1570 3851 1823 721 478 71 84 2842 5026 2834 5049 2070 5965 25.08 24.04 29.11 21.77 91.4 0 6.98 68.2 29.9 19.87 48.73 23.07 9.12 6.24 93.76 36.12 63.88 35.95 64.05 25.76 74.2 0 4553819 4184891 4679381 6949330 17156101 2192240 12246908 6969743 2557 01 4054844 9471058 4682095 1984352 1563345 1796623 6653537 1333070 6720802 4893107 1547431 22.35 20.55 22.97 34.12 87.19 11.14 61.17 34.81 20.19 47.15 23.31 9.89 8.01 91.99 33.29 66.71 33.57 66.43 24.02 75.98
152 Table 6 3 Enabling Characteristics of Veterans Variable Unweighted ( n = 799 6 ) Unweighted Percent Weighted ( N =~20MIL) Weighted Percent Insurance Categories VA Only VA + Medicare VA + Medicaid VA + Medicare + Medicaid VA + Private Private Medicare + Medicaid Medi care Uninsured 136 963 54 51 988 2 320 84 2587 744 1.72 12.15 0.68 0.64 12.46 29.27 1.06 32.64 9.39 341125 1979815 131135 136501 1995212 7 018818 20577 6 5946432 2314629 1.7 9.85 0.65 0. 68 12.89 32.1 1.02 29.59 11.52 Income Categories <$10,000 $10,001 $19,999 $20,000 $29,999 $30,000 $59,999 >$60,000 871 704 992 2532 2936 10.84 8.76 12.35 31.51 36.54 2287296 1882074 2444804 6304848 7448398 11.23 9.24 12.00 30.96 36.57 Receiving Service Connected Disability Payments Yes No Married Yes No 1 146 6889 5847 2188 14.26 85.74 72.77 27.23 2872039 17495382 14197214 6170206 14.1 85.9 69.71 30.29 Employed Yes No 3527 4508 43.9 56.1 9713639 10653781 47.7 52.31
153 Table 6 4. Needs based Characteristics of Veterans Variable Unweighted ( n = 799 6 ) Unweighted Percent Weighted ( N =~20MIL) Weighted Percent PERCEIVED HEALTH Excellent Very Good Good Fair Poor EVALUATED HEALTH Presence of any ADL Yes No Presence of any IADL Yes No 648 2176 2821 1639 634 910 7125 1177 6858 8.18 27.48 35.63 20.7 8.01 11.33 88.67 14.65 85.35 1691027 5601214 7290381 3939838 1563515 2153480 18213941 2872039 17495382 8.41 27.89 36.3 19.61 7.78 10.57 89.43 14.1 85.9
154 Table 6 5 Dependent Variable for Research Question 1 Variable Enrolled (n= 2280 ) Non Enrolled (n =5716) Chi Square p value % Enrolled in the VHA 28.87 71.14 1436.1679 <.0001
155 Table 6 6 Predisposing Characteristics of Enrolled and Non enrolled Veterans Variable Enrolled (n= 2280 ) Non Enrol led (n =5716) Chi Square p value % Age Categories O ver 75 years of age 19.58 27.31 66.1005 <.0001 65 74 years of age 23.24 24.37 55 64 years of age 32.3 27.82 Y ounger than 55 years of ag e 24.88 20.5 % Education Categories No Colle ge Degree College Degree and Beyond Race Categories % White Yes No 76.61 23.39 87.9 12.1 66.28 33.71 92.8 7.2 94.0266 48.3366 <.0001 <.0001 % Black 58.5643 <.0001 Yes No 10.48 89.52 5.58 94.42 % Service Categories % Air Force Yes N o % Army Yes No % Navy Yes No % Marine Yes No 17.66 82.34 51.88 48.12 21.43 78.57 10.82 89.18 20.76 79.24 47.45 52.55 23.73 76.27 8.43 91.57 9.8073 12.791 4.8545 11.1934 .002 .0003 .0021 .0008 % Deployed to OEF/OIF Yes No 9.62 90.38 4.86 95.14 61.0048 <.0001
156 Table 6 6. Continued Variable Enrolled (n=2280) Non Enrolled (n =5716) Chi Square p value % Served in Combat or War Zone Yes No % Exposed to Dead/Dying/Wounded Yes No % Exposed to Hazardous Chemicals Yes No 48.39 51.61 51.65 14.88 38.42 61.58 31.14 68.86 30.93 69.07 20.63 79.37 208.3644 212.9461 273.1584 <.0001 <.0001 <.0001
157 Table 6 7 Enabling Characteristics of Enrolled and Non enrolled Veterans Variable Enrolled (n=2 280 ) Non Enr olled (n =565 6) Chi Square p value % Understand VA Healthcare Entitlement Yes No % Insurance Categories Medicare 76.37 23.63 22.99 77.01 1987.432 <.0001 Yes No 42.13 57 .87 47.01 52.99 15.6373 <.0001 Medicaid Yes No 4.4 95.6 2.73 97.27 14.7319 0.0001 Private Yes No 18.80 81.20 28.30 71.70 609.1588 <.0001 Uninsured Yes No Other Yes No 23.68 76.32 10.99 89.01 5.63 94.37 16.33 83.67 549.0865 388.1780 <.0001 < .0001 % Income Categories 420.6005 <.0001 <$10,000 13.54 9.74 $10,001 $19,999 14.66 6.37 $20,000 $29,999 17.34 10.32 $30,000 $59,999 32.95 30.93 >$60,000 21.52 42.63 % Receiving Service Connected Disability Payments Yes No 35.4 64.6 5.69 94.31 1191.4054 <.0001 %Married Yes No 64.86 35.14 75.98 24.02 103.0228 <.0001 %Employed Yes No 37.73 62.27 46.4 53.6 50.2871 <.0001
158 Table 6 8 Needs Based Characteristics of Enrolled and Non enrolled Veterans V ariable Enrolled (n= 2280 ) Non Enrolled (n =5716) Chi Square p value % PERCEIVED HEALTH 397.6441 <.0001 Excellent 3.99 9.88 Very Good 19.58 30.68 Good 33.25 36.59 Fair 29.43 17.16 Poor 13.75 5.68 EVALUATED HEALTH %P resence of any ADL Yes No 13.8 86.2 10.32 89.68 19.8609 <.0001 % Presence of any IADL Yes No 18.54 81.46 13.07 86.93 39.5365 <.0001
159 Table 6 9 Predisposing Odds of Enrolling into the VHA IDS (N = 20 367 420 ) Effect Estimate 95% Confide nce Limits Chi Square p value Age Categories Over 75 years of age 6 5 74 years of age 0.837 REFERENCE 0.664 1.054 1.2146 0.2704 55 64 years of age 1.222 0.922 1.621 0.2482 0.6183 Younger than 55 years of age 1.375 1.016 1.861 5.7685 0. 011 3 Race Categories White ( No ) 1.265 0.613 2.61 0.0999 0.752 Black ( No ) 1.521 0.659 3.509 0.6534 0.4189 Education Categories No College Degree College Degree and Beyond REFERENCE 0.927 0.654 1.31 4 2.6186 0.1056 Military Status Deployed to OEF/OIF ( No ) 1.376 0.916 2.067 1.1389 0.2859 Served in Combat or War Zone ( No ) 1.117 0.901 1.384 0.5392 0.4628 Exposed to Dead/Dying/Wounded ( REFERE No ) 1.181 0.955 1.46 2.9448 0.0862 Exposed to Hazardous Chemicals ( No ) 1.331 1.099 1.613 0.5163 0.4724
160 Table 6 10 Enabling Odds of Enrolling into the VHA IDS (N = 20,367,420) Effect Estimate 95 % Confidence Limits Chi Square p value Understand VA Healthcare Entitlement ( No ) 9.578 8.008 11.456 2653.22 <.0001 Insurance Categories Medicare Medicaid Other Uninsured Private 2.296 1.751 0.786 0.016 REFE RENCE 1.172 0.966 0.522 0.008 8.629 3.174 1.184 1 .03 19.0932 13.5895 0.7582 72.7312 <.0001 0.3839 0.7791 0.3627 Income Categories <$10,000 1.742 1.232 2.462 3.4299 0.043 4 $10,001 $19,999 2.674 1.918 3.729 4.7548 0.031 2 $20,000 $29,999 2 .582 1.939 3.44 10.06 0.0292 $30,000 $59,999 >$60,000 1.584 REFERENCE 1.253 2.003 3.4088 0.0015 Receiving Service Connected Disability Payments ( No ) 6.238 4.705 8.271 24.5804 <.0001 Married ( No ) 0 .775 0.648 0.927 1.3227 0.2501 Employed ( No ) 0.87 0.723 1.047 2.0757 0.1497
161 Table 6 11 Needs based Odds of Enrolling into the VHA IDS (N = 20,367,420) Effect Estimate 95% Confidence Limits Chi Square p value PERCEIVED HEALTH Excellent Very Good REFERENCE 1.463 0.961 2.226 10.0478 0.1834 Good 1.736 1.158 2.603 1.7701 0.0015 Fair 2.76 1.787 4.262 6.7922 0.0092 Poor EVALUATED HEALTH 3.974 2.477 6.376 13.2795 0.0003 Presence of any ADL ( REFERENCE = No ) 1.012 0.73 1.402 0.05 0.8231 Presence of any IADL ( REFERENCE = No ) 0.854 0.617 1.18 0.0455 0.8311
162 Table 6 12 Dependent Variable f or Research Questions 2A and 2 B Variable Unweighted ( n = 7996 ) Unweighted Percent Wei ghted ( N =~20MIL) Weighted Percent Emergency Room Visit* Yes No 1249 6626 15.86 84.14 3301417 16686730 16.52 83.48 Outpatient Visit** Yes No 5906 1736 77.28 22.72 14549053 4781111 75.26 24.73 Inpatient Visit*** Yes No 1011 6536 13.4 86.6 2442014 16713967 12.74 87.25 160 missing ** 393 missing *** 488 missing
163 Table 6 1 3 Predisposing Characteristics of ER Use Among Veterans Variable ER Use (n=1249) No ER Use (n =6626) Chi Square p value % Age Categories O ver 75 ye ars of age 34.12 30.11 29.4064 <.0001 65 74 years of age 25.23 25.52 55 64 years of age 16.72 26.33 Y ounger than 55 years of age 23.93 18.04 % Education Categories No College Degree College Degree and Beyond Race Categories % Whit e Yes No 73.81 26.19 87.68 12.32 68.32 31.68 92.23 7.77 32.6144 27.1923 <.0001 <.0001 % Black Yes No 10.67 89.33 6.21 93.79 31.4183 <.0001 % Service Categories Air Force Yes No Army Yes No Navy Yes No Marine Yes No 17.32 82.68 52.52 47.89 21.95 78.05 9.84 90.16 20.32 79.68 47.48 52.11 23.36 76.64 9.02 90.98 5.85 8.87 1.16 0.82 0.016 0.003 0.280 0.360
164 Table 6 13. Continued Variable ER Use (n=1249) No ER Use (n =6626) Chi Square p value % Se rved in Combat or War Zone Yes No 40.98 59.02 35.39 64.61 13.903 .0002 % Exposed to Dead/Dying/Wounded Yes No % Exposed to Hazardous Chemicals Yes No 42.97 57.03 29.06 70.94 34.58 65.42 25.13 74.87 31.5606 8.5103 <.0001 0.0035
165 Table 6 14 Enabling Characteristics of ER Use Among Veterans Variable ER Use (n=1249) No ER Use (n =6626) Chi Square p value % Insurance Categories VA Only VA + Medicare VA + Medicaid 2.27 1.06 12.84 1.53 0.58 8.75 116.4895 <.0001 VA + Medicare + Medicaid 22.18 34.51 VA + Private 16.98 11.18 Private 0.97 0.6 Medicare + Medicaid Medicare Uninsured % Income Categories 1.62 30.71 11.37 0.98 32.84 9.04 <$10,000 13.53 10.01 64.276 <.0001 $10,001 $19,999 11.77 8.13 $ 20,000 $29,999 15.61 11.68 $30,000 $59,999 29.38 32 >$60,000 29.7 38.18 % Receiving Service Connected Disability Payments Yes No 80.22 19.78 86.72 13.28 36.1491 <.0001 %Married Yes No 68.86 31.14 73.72 26.28 12.6296 0.0004 %Employed Yes No 38.43 61.57 44.84 55.16 17.5274 <.0001
166 Table 6 15 Needs Based Characteristics of ER Use Among Veterans Variable ER Use (n= 1249 ) No ER Use (n = 6626 ) Chi Square p value % Perceived Health Status Categories 358. 2732 <.0001 Excellent 3.83 9.04 Very Good 17.35 29.47 Good 31.43 36.48 Fair 28.66 19.13 Poor 18.73 5.88 Evaluated Health Status Categories %Presence of any ADL Yes No 19.14 80.86 9.36 90.64 103.3956 <.0001 % Presence of any IADL Yes No 22.74 77.26 12.65 87.35 87.5885 <.0001
167 Table 6 16 Predisposing Characteristics of Outpatient Use Among Veterans Variable Outpatient Use (n= 5906 ) No Outpatient Use (n = 1736 ) Chi Square p value % Age Categories O ver 75 years of age 25.62 19.24 129.5545 <.0001 65 74 years of age 19.34 20.22 55 64 years of age 29.77 28.86 Younger than 55 years of age 25.28 31.68 %Education Categories No College Degree College Degree and Beyond Race Categories % White Yes No 41.26 58.74 91.86 8.14 33.87 66.13 91.12 8.88 30.6554 0.9373 <.0001 0.333 0 % Black Yes No 6.56 93.44 7.16 92.84 0.7466 0.3875 % Service Categories Air Force Yes No Army Yes No Navy Yes No Marine Yes No 20.12 79.88 48.99 51.01 23.11 76.89 8.71 91.29 19.4 80.6 45.94 54.06 23.67 76.33 11.05 88.95 0.4289 4.9471 0.2295 8.5815 0.5125 0.0261 0.6319 0.0034 % Deployed to OEF/OIF Yes No 5.51 94.49 9.17 90.83 28.8118 <.0001
168 Table 6 16. Continued Variabl e Outpatient Use (n= 5906 ) No Outpatient Use (n = 1736 ) Chi Square p value % Served in Combat or War Zone Yes No % Exposed to Dead/Dying/Wounded Yes No % Exposed to Hazardous Chemicals Yes No 36.7 63.3 37.49 62.51 27.4 72.6 34.1 65.9 30.67 69.33 21.31 78.69 3.8398 26.6198 25.7895 .005 <.0001 <.0001
169 Table 6 17 Enabling Characteristics of Outpatient Use Among Veterans Variable Outpatient Use (n= 5906 ) No Outpatient Use (n = 1736 ) Chi Square p val ue % Insurance Categories VA Only VA + Medicare VA + Medicaid 1.75 14.26 0.7 1.46 4.75 0.41 241.1835 <.0001 VA + Medicare + Medicaid 0.75 0.18 VA + Private 9.91 7.38 Private 30.33 42.41 Medicare + Medicaid Medicare Uninsured % Income C ategories 1.06 33.34 7.89 1 28.3 14.12 11.9539 0.0177 <$10,000 9.96 11.87 $10,001 $19,999 8.23 9.27 $20,000 $29,999 12.29 11.52 $30,000 $59,999 31.31 32.55 >$60,000 38.22 34.79 % Receiving Service Connected Disability Paymen ts Yes No 16.15 83.85 8.7 91.3 60.2829 <.0001 %Married Yes No 74.59 25.41 68.61 31.39 24.4653 <.0001 %Employed Yes No 41.75 58.25 51.5 48.5 51.7009 <.0001
170 Table 6 18 Needs Based Characteristics of Outpatient Use Among Vete rans Variable Outpatient Use (n= 5906 ) No Outpatient Use (n = 1736 ) Chi Square p value % PERCEIVED HEALTH 237.4166 <.0001 Excellent 6.8 13.2 Very Good 26.09 34.11 Good 35.22 37.44 Fair 22.55 12.5 Poor 9.34 2.75 EVA LUATED HEALTH %Presence of any ADL Yes No 10.4 89.6 9.91 90.09 0.3468 0.5559 % Presence of any IADL Yes No 13.61 86.39 14.29 85.71 0.511 0.4747
171 Table 6 19 Predisposing Characteristics of Inpatient Use Among Veterans Variable I npatient Use (n= 1011 ) No Inpatient Use (n=6536 ) Chi Square p value % Age Categories O ver 75 years of age 35.91 22.87 104.7654 <.0001 65 74 years of age 25.52 23.76 55 64 years of age 24.53 30.03 Y ounger than 55 years of age 14.05 23.33 % Education Categories No College Degree College Degree and Beyond Race Categories % White Yes No 66.37 33.63 91.65 8.35 59.64 40.36 91.45 8.55 16.6055 0.0425 <.0001 0.8367 % Black Yes No 6.8 93.2 6.97 93.03 0.0365 0. 8485 % Service Categories Air Force Yes No Army Yes No Navy Yes No Marine Yes No % Deployed to OEF/OIF Yes No 18.85 81.15 51.41 48.59 22.68 77.32 8.97 91.03 4.49 95.51 20.14 79.86 48 52 23.2 76.8 9.28 90.72 6.63 93.37 0.8937 3.9973 0. 1306 0.0958 6.3949 0.3445 0.0456 0.7178 0.757 0 .0114
172 Table 6 19. Continued Variable % Served in Combat or War Zone Yes No % Exposed to Dead/Dying/Wounded Yes No % Exposed to Hazardous Chemicals Yes No Inpatient Use (n=1011) 37.99 62.01 40.16 59.84 27.4 72.6 No Inpatient Use (n=6536) 35.92 64.08 35.41 64.59 25.84 74.16 Chi Square 1.5869 8.3879 1.102 p value 0.2078 0.0038 0.2938
173 Table 6 20 Enabling Characteristics of Inp atient Use Among Veterans Variable Inpatient Use (n= 1011 ) No Inpatient Use (n=6536 ) Chi Square p value % Insurance Categories VA Only VA + Medicare VA + Medicaid 1.61 16.5 0.7 1.73 11.62 0.6 102.0422 <.0001 VA + Medicare + Medicaid 1. 21 0.57 VA + Private 8.15 9.72 Private 22.03 33.96 Medicare + Medicaid Medicare Uninsured Income Categories 2.31 39.64 7.85 0.85 31.22 9.73 34.2885 <.0001 % <$10,000 12.56 10.16 % $10,001 $19,999 9.89 8.43 % $20,000 $29,999 15. 92 11.75 % $30,000 $59,999 31.36 31.44 % >$60,000 30.27 38.22 % Receiving Service Connected Disability Payments Yes No 16.32 83.68 14.01 85.99 3.7933 0.0515 %Married Yes No 71.51 28.49 73.19 26.81 1.2548 0.2626 %Emplo yed Yes No 33.53 66.47 45.43 54.57 50.3127 <.0001
174 Table 6 21 Needs Based Characteristics of Inpatient Use Among Veterans Variable Inpatient Use (n= 1011 ) No Inpatient Use (n=6536 ) Chi Square p value % PERCEIVED HEALTH 4 15.0436 <.0001 Excellent 2.41 9.15 Very Good 14.56 29.65 Good 32.73 36.07 Fair 29.42 19.33 Poor 20.88 5.8 EVALUATED HEALTH %Presence of any ADL Yes No 18 82 9.49 90.51 66.8587 <.0001 % Presence of any IADL Yes No 20.08 79.9 2 13 87 36.4969 <.0001
175 Table 6 22. Dependent Va riable for Research Questions 2A and 2 B Variable Enr olled (n=2280 ) Non Enrolled (n =5716) Chi Square p value % Emergency Room Visit Yes No 21.65 78.35 13.52 86.48 79.9861 <.0001 % Out patient Visit Yes No 86.35 13.65 73.64 26.36 143.927 <.0001 % Inpatient Visit Yes No 14.69 85.31 12.86 87.14 4.4793 0.0343
176 Table 6 23 Predisposing Odds of ER Use for Research Question 2A Age Categories O ver 75 years of age 65 74 years of age 1.293 REFERENCE 1.068 1.565 6.9445 0.0084 55 64 years of age Y ounger than 55 years of age 0.856 1.37 0.684 1.079 1.073 1.74 1.8148 6.6633 0.1779 0.0098 Education Ca tegories No College Degree College Degree and Beyond Race Categories White (REFERENCE Black ( REFERENCE No ) REFERENCE 0.965 0.911 1.184 0.784 0.577 0.731 1.188 1.438 1.916 0.0923 0.0708 0.0355 0.7355 0.6885 0.492 Served in Combat or War Zone ( No ) 1.038 0.872 1.236 0. 0589 0.6741 Exposed to Dead/Dying/Wounded ( No ) Exposed to Hazardous Chemicals ( No ) 1.175 1.007 0.97 0.792 1.422 1. 281 1 1256 1 2513 0.0991 0.9559 Effect Estimate 95% Confidence Limits Chi Square p value
177 Table 6 24 Enabling Odds of ER Use for Research Question 2A Effect Estimate 95% Confidence Limits Chi Square p value Insurance Categories VA Only 1.023 0.594 1.761 0.9345 0.3337 Multiple 1.379 1 .12 1.698 1.1545 0. 035 6 Uninsured Private 1.583 REFERENCE 1.161 2.159 4.0633 0.0438 Income Categories <$10,000 1.107 0.818 1.499 2.5465 0.1105 $10,001 $19,999 1.091 0.815 1.461 2. 3511 0. 1 315 $20,000 $29,999 1.093 0.811 1.474 3.0224 0. 1 821 $30,000 $59,999 >$60,000 0.978 REFERENCE 0.792 1.207 1 .6559 0.418 Receiving Service C onnected Disability P ayments ( No ) 1.223 0.974 1.536 1.0004 0.146 Married ( No ) Employed ( No ) 0.876 1.047 0.712 0.862 1.015 1.274 0.1397 1.6132 0.1669 0.6419
178 Table 6 25. Needs Based Odds of ER Use for Research Question 2A Effect Effect 95% Confidence Limits Chi Square p value PERCEIVED HEALTH Excellent REFERENCE Very Good 1.3 15 0.834 2.073 0.1357 .2157 Good 1.766 1.143 2.728 2.3156 .00 3 Fair 2.898 1.895 4.431 6 0536 <.0001 Poor 6.235 3.935 9.878 13.0061 <.0001 EVALUATED HEALTH Presence of ADL ( No ) 1.67 1.225 2.276 0.0004 Presence of IADL ( RE No ) 0.921 0.705 1.203 0.3199
179 Table 6 26 Predisposing Odds of Outpatient Use for Research Question 2A Age Categories O ver 75 years of age 65 74 years of age 1.189 REFER ENCE 0.972 1.456 0.9858 .0976 55 64 years of age Y ounger than 55 years of age Education Categories No College Degree College Degree and Beyond 0.771 0.488 REFERENCE 1.6 67 0.651 0.404 1. 23 4 0 914 0.59 1.88 5 0.2855 2.679 1.325 .035 <.0001 <.0001 Race Categories White ( No ) 0.929 0.56 1.543 .0257 0.777 Black ( No ) 0.764 0.413 1.41 .3590 0.389 Served in Combat or War Zone ( No ) 0.876 0.716 1.073 2901 0.2014 Exposed to Dead/Dying/Wounded ( No ) Exposed to Hazardous Chemicals ( No ) 1.236 1.259 1.02 1.046 1.498 1.515 .5478 1.290 0.0309 0.0148 Effect Estimate 95% Confidence Limits Chi Square p value
180 Table 6 27 Enabling Odds of Outpatient Use fo r Research Question 2A Effect Estimate 95% Confidence Limits Chi Square p value Insurance Categories VA Only 1.025 0.43 2.446 0.0405 .21 01 Multiple 1.812 1.472 2.231 3 .4224 <.0001 Uninsured Private 0.778 REFERENCE 0.572 1.058 0.2276 .11 01 Income Categories <$10,000 0.522 0.412 0.662 29.8986 <.0001 $10,001 $19,999 0.63 0.459 0.864 11.3795 0.0007 $20,000 $29,999 0.637 0.507 0.799 15.0914 0.0001 $30,000 $59,999 >$60,000 0.645 REFERENCE 0.533 0.78 21.4436 <.0001 Receiv ing Service Connected Disability Payments ( No ) 1.845 1.419 2.398 22.638 <.0001
181 Table 6 28 Needs based Odds of Outpatient Use for Research Question 2A Effect Estimate 95% Confidence Limits Chi Square p value PERCEIV ED HEALTH Excellent REFERENCE Very Good 1.44 1.124 1.845 8.0594 0.0045 Good 1.728 1.317 2.267 14.8464 0.0001 Fair Poor 3.891 9.118 2.866 5.863 5.282 14.181 66.8338 83.3393 <.0001 <.0001 EVALUATED HEALTH Presence of ADL ( REFEREN No ) 1.06 0.754 1.49 0.0565 0.375 Presence of IADL ( No ) 0.524 0.371 0.74 11.5843 <.0001
182 Table 6 29. Predisposing Odds of Inpatient Use for Research Question 2A Effect Estimate 95% Confidence Limits Chi Square p value Age Categories O ver 75 years of age 65 74 years of age 1.323 REFERENCE 1.099 1.593 8.7661 0.0031 55 64 years of age Y ounger than 55 years of age 0.842 0. 643 0.614 0.455 1.154 0.908 1.1423 6.2936 0.2852 0.0121 Education Categories No College Degree College Degree and Beyond Race Categories White ( No ) Black ( No ) Served in Combat or War Zone ( No ) Exposed to Dead/Dying/Wounded ( No ) REFERENCE 1.009 1.279 1.214 0.918 1.154 0.852 0.735 0.668 0.753 0.941 1.196 2.227 2.205 1.118 1.414 0. 2868 0.5702 0.5 324 0. 1546 0. 0981 0 238 0.397 0.417 0.237 0.132 Exposed to Hazardous Chemicals ( No ) 1.071 0.846 1.357 0. 1385 0.346
183 Table 6 30 Enabling Odds of Inpatient Use for Research Question 2A Effect Estimate 95% Confidence Limits Chi Square p value Insurance Categories VA Only 0.829 0.397 1.731 0.1345 0.7138 Mult iple 0.921 0.761 1.113 0.3343 0.5631 Uninsured Private 0.982 REFERENCE 0.693 1.39 0.0033 0.6541 Income Categories <$10,000 0.893 0.649 1.228 0.3732 0.5412 $10,001 $19,999 0.961 0.714 1.293 0.0292 0.8644 $20,000 $29,999 1.095 0.82 1.462 0. 4509 0.5019 $30,000 $59,999 >$60,000 0.89 REFERENCE 0.728 1.088 1.3872 0.2389 Receiving Service Connected Disability Payments ( No ) 1.177 0.926 1.496 1.7267 0.173 Married ( No ) 0.943 0.765 1.163 0.3168 0.233
184 Table 6 31 Odds of Inpatient Use for Research Question 2A Effect Estimate 95% Confidence Limits Chi Square p value PERCEIVED HEALTH Excellent REFERENCE Very Good 1.748 1.079 2.834 4.6405 0.0312 Good 3.51 2.172 5.672 24.9771 <.0001 Fair 5.479 3.412 8.798 47.0818 <.0001 Poor 12.382 7.29 21.03 1 81.5083 <.0001 EVALUATUED HEALTH Presence of any ADL ( No ) Presence of any IADL ( No ) 1.598 0.775 1.15 0.546 2.221 1.099 7.345 2.1317 0 0002 0.316
185 Table 6 32 Predisposing Odds of ER Use for Research Question 2B Age Categories Over 75 years of age 1.371 1.121 1.676 7.5875 0.0059 65 74 REFERENCE 55 64 years of age 1.005 0.738 1.368 0.0123 0.9119 Younger than 55 years of a ge Race Categories 1.622 1.134 2.319 6.882 0 .0087 White ( No ) 0.895 0.565 1.418 2.6188 0.7355 Black ( No ) Education Categories 1.156 0.709 1.885 2.3227 0.6885 No College Degree REFERENCE College Degree and Beyond 1.058 0.762 1.469 1 9156 0.492 Served in C ombat or War Z one ( REFERENCE No ) 1.013 0.83 1.235 1 0023 0.6741 Exposed to Dead/Dying/ Wounded ( No ) 1.16 0.962 1.398 1. 5645 0.0991 Exposed to Hazardous C hemicals ( No ) 1.006 0.791 1.28 1. 0023 0.9559 Effect Estimate 95% Confidence Limits Chi Square p value
186 Table 6 33. Enabling Odds of ER Use for Research Question 2B Income Categories <$10,000 1.091 0.787 1.511 $10,001 $19,999 1.154 0.835 1.594 $20,000 $29,999 1. 105 0.799 1.528 $30,000 $59,999 0.987 0.786 1.238 >$60,000 REFERENCE Receiving Service Connected Disability Payments ( No ) Married ( No ) Employed ( No ) 1.211 0.874 1.037 0.957 0.723 0.855 1.532 1.055 1.257 0.146 0.1669 0.6419 Insurance Categories VA only 1.08 0.586 1.992 0.2483 0.6803 VA + Medicare + Medicaid 1.631 0.734 3.624 0.0013 0.2262 Uninsured 1.637 1.114 2.406 10.8388 0.001 Private ins urance REFERENCE VA + Medicare 1.517 0.984 2.34 1.4644 0.0634 VA + Medicaid 0.952 0.382 2.371 0.8067 0.2 067 Medicare + Medicaid 1.08 0.456 2.556 0.1698 0.6183 Medicare VA + Private 1.381 1.692 0.983 1.221 1.94 2.343 7.3467 3.4472 4.2 254 0.4923 0.5431 2.9697 2.6701 2.5333 0.9714 0.0067 0.1 067 0.4829 0.4611 Effect Estimate 95% Confidence Limits Chi Square p value
187 Table 6 34 Needs based Odds of ER Use for Research Question 2B 95% Confidence Chi Effect Estimate Limits Square p value PERCEIVED HEALTH STATUS Excellent REFERENCE Very good 1.247 0.781 1.991 0.9973 0.318 Good 1.63 7 1.056 2.537 6.0706 0.0137 Fair 2.735 1.783 4.196 22.3815 <.0001 Poor EVALUATED HEALTH STATUS 5.631 3.526 8.99 56.9334 <.0001 Presence of any ADL ( No ) 1.807 1.317 2.481 0.0014 0.0004 Presence of any IADL ( No ) 0.856 0.651 1.125 0.4601 0.3199
188 Table 6 35 Predisposing Odds of Outpatient for Research Question 2B Effect Estimate 95% Confidence Limits p value Age Categories 0.0067 Over 75 years of age 1.252 1.002 1.563 65 74 years of age REFERENCE 55 64 years of age 1.144 0.861 1.519 Younger than 55 years of age Race Categories 0.832 0.612 1.131 White ( No ) 0.929 0.56 1.543 0.777 Black ( No ) Education Categories No College Degree College Degree and Beyond Served in C ombat or War Z one ( No ) Exposed to Dead/D ying / Wounded ( No ) Exposed to Hazardous C hemicals ( No ) 0.764 REFERENCE 1.567 0.876 1.236 1.259 0.413 1.324 0.716 1.02 1.046 1.41 1.855 1.073 1.498 1.515 0.389 <.0001 0.2014 0.0309 0.0148
189 Table 6 36 Enabling Odds of Outpatient Use for Research Question 2B 0.7921 0.0049 0.155 <.0001 0.2743 0.0685 0.0001 0.0004 0.0006 0.0015 0.0411 0.0317 0.0001 0.1335 0.1682 Effect Estimate 95% Confidence Limits Chi Square p value Insurance Categor ies VA only VA + Medicare + Medicaid Uninsured Private insurance VA + Medicare VA + Medicaid Medicare + Medicaid Medicare VA + Private Income Categories <$10,000 $10,001 $19,999 1.085 7.887 0.768 REFERENCE 3.289 1.547 1.629 1.652 1.639 0.557 0.6 18 0.426 1.601 0.555 2.143 0.584 0.714 1.174 1.241 0.428 0.438 2.759 28.243 1.061 4.806 4.095 3.715 2.051 2.192 0.725 0.874 0.0695 37.0636 12.4739 15.0167 7.9019 2.0223 1.1951 3.3194 2 8 1254 1 0 2456 $20,000 $29,999 0.724 0.568 0.924 1 3 0468 $30,000 $59,999 0.711 0.58 0.872 2 0 7898 >$60,000 REFERENCE Receiving Service Connected Disability Payments ( No ) 1.727 1.305 2.285 22.638 Married ( No ) Employed ( No ) 1.133 0.879 0.963 0.732 1.333 1.056 2.5671 0.0 864
190 Table 6 37 Needs based Odds of Outpatient Use for Research Question 2B PERCEIVED HEALTH S TATUS Excellent REFERENCE Very good 1.462 1.131 1.889 7 .0 526 0.004 3 Good 1.811 1.367 2.4 1 2 4568 0.0001 Fair 4.129 2.943 5.792 6 3 8756 <.0001 Poor EVALUATED HEALTH STATUS 8.88 5.545 14.222 8 2 4568 <.0001 Presence of any ADL ( REFERE No ) 1.06 0.754 1.49 0.7388 0. 8354 Presence of any IADL ( No ) 0.524 0.371 0.74 0.0002 0.000 5 Effect Estimate 95% Confidence Limits Chi Square p value
191 Table 6 38 Predisposing Odds of Inpatient for Resea rch Question 2B Effect Estimate 95% Confidence Limits Chi Square p value Age Categories Over 75 years of age 1.35 1.107 1.647 9.0639 0.0026 65 74 years of age REFERENCE 55 64 years of age 0.845 0.612 1.166 1.3577 0.2439 Younger than 55 year s of age Race Categories 0.688 0.48 0.987 6.6982 0.0097 White (Yes, No) 1.279 0.735 2.227 0.3078 0.3009 Black (Yes, No) Education Categories No College Degree College Degree and Beyond 1.214 REFERENCE 1.009 0.668 0.852 2.205 1.196 0.429 0.7156 0.5323 0.2568 Served in C ombat or War Z one ( REFERENCE = No ) 0.918 0.753 1.118 0.8964 0.1521 Exposed to Dead/Dying/ Wounded ( REFERENCE = No ) 1.154 0.941 1.414 0.9801 0.1526 Exposed to Hazardous chemicals ( REFERENC E = No ) 1.071 0.846 1.357 0.8655 0.5832
192 Table 6 39 Enabling Odds of Inpatient Use for Research Question 2B Effect Estimate 95% Confidence Limits Chi Square p value Insurance Categories VA only VA + Medicare + Medicaid Uninsured Private insurance VA + Medicare VA + Medicaid Medicare + Medicaid Medicare VA + Private Income Categories <$10,000 $10,001 $19,999 0.784 1.302 1.028 REFERENCE 1.011 0.335 1.961 1.159 1.028 0.881 1.005 0.351 0.518 0.715 0.677 0.096 0.909 0.82 7 0.715 0.613 0.724 1.747 3.274 1.478 1.508 1.164 4.228 1.624 1.478 1.265 1.396 0.1648 0.2668 0.0019 0.0084 0.0182 2.2996 2.1913 0.7284 0.3615 0.0173 0.6848 0.6055 0.965 0.9268 0.8928 0.1294 0.1388 0.3934 0.5477 0.8955 $20,000 $29,999 1.122 0. 816 1.543 0.5226 0.4697 $30,000 $59,999 0.896 0.727 1.105 1.3115 0.2521 >$60,000 REFERENCE Receiving Service Connected Disability Payments ( REFERENCE = No ) 1.165 0.91 1.491 1.5599 0.2268 Married ( REFERENCE = No ) Employe d ( No ) 0.947 0.951 0.757 0.771 1.184 1.174 0.2313 0.2 012 0.5901 0.6325
193 Table 6 40. Needs based Odds of Inpatient Use for Research Question 2B Effect Estimate 95% Confidence Limits Chi Square p value PERCEIVED HE ALTH STATUS Excellent REFERENCE Very good 1.7 1.042 2.775 4.5703 0.0325 Good 3.3 2.017 5.399 24.6008 <.0001 Fair 5.246 3.219 8.55 45.8053 <.0001 Poor EVALUATED HEALTH STATUS 11. 416 6.566 19.848 80.0557 <.0001 Presence of any ADL ( No ) 1.645 1.162 2.33 7.4309 0.0049 Presence of any IADL ( No ) 0.754 0.524 1.086 2.3198 0.1418
194 CHAPTER 7 CONCLUSIONS Discussion These findings support previo us research findings which make reference to distinct variables purported to drive healthcare utilization. Through the use of a modified ABM referencing enrollment into the VHA, this research was able to uncover s everal important distinctions. Succinctly p ut, predictor variable of interest, proved to be one of the most influential driving variables of both utilization and enrollment. As a result, this research helps bridge t he gap in literature involving uptake rat es widely researched in the civilian literature but, only recently explored within the Veteran population. Research Question 2A used the m odified ABM to explore associations between utilization and insurance across three broad categories of coverage; the uninsured, Veterans who use the VHA IDS only, Veterans with multiple coverage, and the comparison category, private insurance The results only marginally confirmed previous There are several plausible reasons why these results are less revealing. On one hand, Veterans may be self selecting into the VHA IDS based on other predi sposing and needs based factors with VHA IDS coverage being less important in this decision. On the other h and, there remains uncertain ty into whether other types of insurance statuses are associated with different patterns of utilization This may be minimizing the association within this broader category of Veterans who may be using the VHA IDS and other alte rnative coverage as a means of gaining ATC. This is a cause of concern for the VHA IDS. Its primary care model structure relies on extensive coordination of care to provide the best
195 possible quality of care available to the Veterans it serves. As more Vete rans gain alternative ATC with other institutions, there is greater reliance placed on out of network coordination to deliver uncompromising care. Research Question 2B was used to uncover possible associations between utilization and insurance across seve ral mutually exclusive categories of coverage. This analysis uncovered several important associations. Firstly, that it c onfirms previous findings that insurance status is associated with varying odds of use of the ER and primary care. These results provid e some evidence that Veterans are likely self selecting into the VHA IDS and other insurance plans based on predisposing, enabling, and needs based factors. This also e xpands the knowledge base on associations with Veterans ha ving multiple types of coverag e, particularly among Veterans with more than two types of coverage. As a result, it r aises further inquiry into whether having multiple simultaneous insurance coverage are having a negative effect on the qua lity of care and access to care. Specifically, a mong Veterans enrolled in the VHA IDS and have Medicare and Medicaid, the odds of outpatient use is more than double that of enrolled Veterans with Medicare, the next highest odds ratio for outpatient care. Additionally, dually covered Veterans enrolled in the VHA IDS with either Medicare or private coverage were more likely to use outpatient care and emergency care. Within these same cohorts, the odds of ER use are similar to the uninsured population. It is possible that this category of Veterans may be us ing more primary care and are having adverse health effects associated with greater ER use or have self selected into these insurance options due to a higher risk of sickness or disease.
196 Implications From an academic perspective, this research e xpands t he traditional use of the ABM and its focus on utilization to include enrollment within the full value chain of realized healthcare delivery. It also e xpounds upon associations between enrollment and insurance status on utilization patterns As hypothesize awareness of VHA IDS benefits has a positive association with enrollment. This is important for the VA as it pursues a clearer understanding on how knowledge of the VA is disseminated among Veterans. The use of the modified ABM dra ws important inferences into the VA policy problem of expanded eligibility and its consequential effects. In particular, the accessibility, availability, accommodation, and acceptability of services among a more diverse population of e ligible Veterans are important considerations within an institution focused on primary care and constrained by a capitated set budget. By addressing these concerns, the VA may be able to focus its resources on targeted enrollment and utilization interventions designed to addre ss need and predisposition over enabling factors. Additionally, f uture health policy interventions may then be designed around altering utilization patterns in a proactive manner and whenever necessary. Furthermore, t he results of this research support on going VA research on the need for health information exchange initiatives between VA and non VA providers The VA and the DoD have recently begun testing several pilot exchanges nationwide in the hopes that they will soon provide a standardized platform fo r the flow of secure patient information within a network of institutions (Weiner & Haggstrom, 2013) The demand for these exchanges stems from the mounting concern on how to coordinate care for
197 larger populations of patients with co morbidities and ATC wi thin a rural setting. Mueller and Lampman ( 2011 ) described how the VHA IDS is already looking into using presently established state health information exchange networks in combination with the National Health Information Network as a model. While this res earch is unable to draw causal reasons into why certain groups of Veterans are using more care, it is possible to infer from the previous literature that Veterans with more insurance are in a higher risk category. Specifically, these Veterans might be pron e to more adverse effects as a result of higher utilization patterns These behavioral patterns may be to better health, or the perception that he or she is sicklier and anticipate s offset ting futur e costs of care with greater coverage. Additionally, multiple insurance may also permit high risk Veteran s who are unable to gain ATC for primary care to gain access through the ER when medically unnecessary Although this practice might be discouraged amo ng non Veterans through the use of high deductibles and co payments, Veterans with a purple heart, or have a high Priority Group ratings do not have OOP payments for care received at the ER (VA Priority Groups, 2013). At this time, this study could not con trol for geographic or market variations and thus should be studied in greater depth in subsequent research to determine what may be causing more utilization At the heart of the PPACA is the expansion of healthcare insurance across the nation If great er insurance coverage among Veterans is truly found to be a contributor of healthcare utilization policymakers should be cautious about how expansion policies are implemented within this group. The literature points to how o veruse and underuse
198 may be lead ing to higher total costs of care, a red uction in quality of care as a consequence to a deficiency in the coordination of care and undesirable patient outcomes. While this research is unable to determine how overuse or underuse is classified, i t may be po ssible that further expansion of healthcare insurance covera ge may have complex implications among Veterans with multiple coverage As such, the VHA IDS primary care model should be consider ing the association between Veterans with multiple insurance cover age and their greater odds of use as well as uninsured Veterans and their reduced odds of use. Additionally, the V A should consider other predisposing, enabling, and need factors across different groups of Veterans when conducting outreach to a growing div ersity of new Veterans. Future Research As the PPACA is fully implemented over the next few years and Medicaid expands health insurance market exchanges are fully operational and individual tax penalties are fully enforced different groups of Veterans may likely find greater value in enrolling and using the VHA IDS. These changes consider key structural transformations within the insurance industry which consequentially affects enrollment and utilization. In effect, certain groups of Veterans may discov er new incentives or disincentives to enroll in the VHA IDS as well as other available insurance options. Research may using social network analysis and other means of exploring cultur al variations and possible causal relationships As the VA continues to develop its healthcare resources, it will need to understand how knowledge is effectively disseminated among a diverse population of Veterans
199 Tax penalties and tax credits apply to Am ericans who choose to have insurance or not have insurance in 2014. According to CMS market exchange guidelines (2013), Veterans do not receive a tax credit in the new insurance market exchanges if they are enrolled in the VHA IDS. As a result, certain Vet erans may be more likely to remain enrolled and us e care at the VHA IDS and less likely to pursue multiple insurance coverage Conversely, there may also be other Veterans who choose to dis enroll from the VHA IDS if they have a greater financial or health care advantage to buying care on the market exchange. This is doubtful, however, given that the VHA IDS does not require an insurance premium to remain enrolled and individual insurance is projected to increase in cost in the next few years. It is possible that as more pressure is placed on decreased payments to FFS, ACOs and vertically integrated may become more attractive to providers as value based purchasing becomes more prevalent causing different Veterans to s elf select into other insurance optio ns over time. A s Medicaid expands under the ACA and redefines the physician payment and delivery system research should evaluate the effect s of having multiple insurance coverage on utilization and patient outcomes in a longitudinal study. Moreover ther e is a certain need to further ex plore possible reasons why overuse of the ER and overuse/underuse of primary care may be occurring. A robust attempt should be made to l ink enrollment and multiple coverage data with evidence of state an d other applicable risk adjusters to identify possible causal links between self selection and utilization
200 APPENDIX A ITEM MEASURES Variables to be Measured Description of Variable Items Used to Measure Variable Predisposing Factors: The socio cultu ral characteristics of individuals that exist prior to their illness Demographic: Age and Gender O1. What is your gender? Dichotomous variable: 1 =Male or 2=Female O2. What is your year of birth? (Age) Continuous variable recoded into four categories: Over 75 years of age 65 74 years of age 55 64 yea rs of age Younger than 55 years of age Social Structure: Education, occupation, ethnicity, social networks, social interactions, and culture O4. What is the highest degree or level of school you have completed? Categorical variable referenced b y numeric code from 1 9: 1= Less than high school 2= High school diploma / GED 3= Some college credit, but less than 1 year of college credit 4= 1 or more years of college credit, no degree 5= example, AA, AS ) 6= example, BA, BS) 7= example, MA, MS, MEng, MEd, MSW, MBA) 8= Professional degree beyond example, MD, DDS, DVM, LLB, JD) 9=Doctorate de gree (for example, PhD, EdD)
201 Predisposing Factors Continued: Social Structure Continued: Items Used to Measure Variable Continued: O6. What is your race? Categorical variable referenced by numeric codes from 1 11: 1=White 2=Black or African American 3=American Indian or Alaska Native 4=Asian Indian 5=Chinese 6=Filipino 7=Other Asian (for example, Hmong, Laotian, Thai, Pakistani, Cambodian, and so on) 8=Native Hawaiian 9=Guamanian or Chamorro 10=Samoan 11=Other Pacific Islander (for example, Fijian, Tongan, and so on) O5: Are you of Hispanic, Latino, or Spanish origin? Categorical variable referenced by numeric codes from 2 5 re operationalized to dichotomous: 1=codes 2 5 (Yes) 2=No A3. When did you serve on active duty in the U.S. Armed Forces? Categorical variable referenced by numeric code from 1 9: 1=September 2001 or later
202 Predisposing Factors Continued: Social Structure Continued: Items Used to Measure Variable Continued: 2=August 1990 to August 2001 (includes Persian Gulf War) 3=May 1975 t o July 1990 4=Vietnam era (August 1964 to April 1975) 5=February 1955 to July 1964 6=Korean War (July 1950 to January 1955) 7=January 1947 to June 1950 8=World War II (December 1941 to December 1946) 9=November 1941 or earlier A4. Did you deploy in support of Operation Enduring Freedom (OEF) or Operation Iraqi Freedom (OIF)? Dichotomous variable: 1=Yes or 2= No A7. Did you ever serve in a combat or war zone? Dichotomous variable: 1=Yes or 2= No A8. During yo ur military service, were you ever exposed to dead, dying, or wounded people? Dichotomous variable: 1=Yes or 2= No A9. Were you ever a prisoner of war? Dichotomous variable : 1=Yes or 2= No
203 E nabli ng Factors: Personal/Family: T r equired to access health services, income, health insurance, a regular source of care, travel, extent and quality of social relationships Justification for use as enabling variables: T he understanding of VA health benefits would play a significant role in enrollment into the VHA IDS. Predictor variable Knowledge of VA healthcare benefits/coverage. This variable is used for two reasons; 1. To follow suit with the Congressional mandate requiring the VA to collect understand ing of their health benefits and the implied association with utilization of VA resources a nd, 2. To test the presumption that knowledge of available benefits has a significant positive affect on reducing access to care barriers controlling for other predisposing, enabling, and need factors. Items Used to Measur e Variable Continued: A10. During your military service, were you ever exposed to environmental hazards such as Agent Orange, chemical warfare agents, ionizing radiation, or other potentially toxic substances? Categorical variables coded 1 4 and 8: 1=DE FINITELY YES 2=PROBABLY YES 3=PROBABLY NO 4=DEFINITELY NO 8=DON'T KNOW Predictor Variable for Enrollment (Hypothesis 1): B1. Please indicate how much you understand about the following statements regarding the Veterans benefits provided by the Departmen t of Veterans Affairs (VA). b. The Veterans health care Categorical variables coded 1 4: 1=A lot, 2=Some, 3=A little, 4=Not at all
204 Enabli ng Factors Continued: Personal/Family Continued : Income is used for two reasons; a. to establish the means by which a Veteran may be able to enroll and access care and b. to determine some level of self selection into differen t types of plans (VHA IDS, other available plans, or un insurance status) Items Used to Measure Variable Continued: N2. Which income range cate gory represents the total combined income of all members of this family during the past 12 months? Categorical variables coded 1 16: 1= LESS THAN $5,000 2= $5,000 TO $7,499 3= $7,500 TO $9,999 4= $10,000 TO $12,499 5= $12,500 TO $14,999 6= $15,000 TO $19,999 7= $20,000 TO $24,999 8= $25,000 TO $29,999 9= $30,000 TO $34,999 10= $35,000 TO $39,999 11= $40,000 TO $49,999 12= $50,000 TO $59,999 13= $60,000 TO $74,999 14= $75,000 TO $99,999 15= $100,000 TO $149,999 16= $150,000 OR MORE
205 Enabling Factors Continued: Personal/Family Continued : Serv ice connected disability Rating (SCDR): Prior to June 2009, SCDR was used as a p rimary qualifier of eligibilit y for enrollment into the VHA IDS. Following June 2009, the VA began permitting all Honorably discharged Veterans with an opportunity to enroll. This variable remains an enabling factor because SCDRs represent means of determining cost sharing between the VA and the Veteran. Marital status has been used extensively throughout the ABM literature. It has become generally accepted that greater family support indicates greater social structure and hence, better access to healthcare. Items Used to Measure Variable Continued: N1: Please indicate whether your family received income (past 12 months) in any of the categories listed below. connected disability compensation payments. Dichoto mous variable: 1=Yes or 2= No O 7. What is your current marital status? Recoded to dichotomous yes or no ('yes' indicates married or in civil union). C ategorical variable coded 1 6: recoded into Married or in Civil Commitment Union or No.
206 Enabling Factors Continued: Personal/Family Continued : Employment Status Items Used to Measure Variable Continued: H vacation or sick leave from work; Not working, but looking for work; Not working and not looking for work. Recoded to dichotomous 'working 'or 'not working'. H ealth insurance is an important enabler to accessing healthcare. By assessing VHA IDS independently and in combination with other health insurance, we may be able to predict the utilization of ER, inpatient, and outpatient care with greater accuracy J ustification: This is a s elf sele ction control variable for Aim 2 Predictor Variable f or Utilization (Specific Aims 2A and 2B ): covered by any of the following types of health insurance or Categorical variable coded 1 10 Code 6 is cross referenced with 'enrolled in VA' then reduces any member with any other type of insurance and recoded as Only' C ode 6 is cross referenced with 'enrolled in VA' and matched to any me mber with Medicare or Medicaid insurance and recoded as 'multiple enr olled' C odes 2, 3 and 7 are recoded cross referencing with any member w ith any other type of insurance. Code 1 is considered ninsured
207 Enabling Factors Continued: Personal/Family Continued : Items Used to Measure Variable Continued: 1= NONE 2= THROUGH CURRENT/FORMER EMPLOYER/UNION 3= PURCHASED DIRECTLY FROM INSURANCE CO 4= MEDICARE 5= MEDICAID OR OTHER GOV ASSISTANCE PLAN 6= VA 7= TRICARE/MILITARY HEALTH CARE 8= INDIAN HEALTH SERVICE 9= OTHER TYPE OF HEALTH INSURANCE 10= OTHER SPECIFY
208 Need based Factors : The most immediate cause of health service use, from functional and health problems that generate the need for hea lth care services. "Perceived need will better help to understand care seeking and adherence to a medical regimen, while evaluated need will be more closely related to the kind and amount of treatment that will be provided after a patient has presented to a medical care provider." (Andersen, 1995) E valuated/Perceived Need: Represents professional judgment about people's health status and their need for medical care. (Andersen, 1995) How people view their own general health and functional state, as well as how they experience symptoms of illness pain, and worries about their health and whether or not they judge their problems to be of sufficient importance and magnitude to seek professional help." (Andersen, 1995) This category is broken down into the following subcategories: 1. Perceived gen eral health status 2. Evaluated need determined by perceived levels of assistance further defined as; a. Activities of daily aaaa living (ADLs) b. Instrumental aaaa activities of daily aaaa living (IADLs) Items Used to Measure Variable Continued: Perceived N eed: E1. In general, would you Categorical variable coded 1 5: 1=Excellent 2=Very good 3=Good 4=Fair 5=Poor Evaluated N eed: D4. Are you currently in need of the aid and att endance of another person? Dichotomous variable: coded 1=Yes or 2=No D3. In the past week, how much assistance did you require in the following activities due to a you require in the following activities due to a health condition NOTE: There are six basic ADLs: eating, bathing, dressing, toileting, ambulation (i.e. walking), and continence. There are 7 used in this item of which the following categorical variables were re coded from a g to d ichotomous existence of any ADL
209 Need based Factors Continued : Evaluated/Perceived Need Continued : Items Used to Measure Var iable Continued: Categorical variable coded from: E valuated N eed Continued : 1= Bathing 2= Eating 3= Transferring from bed or a chair 4= Using the toilet 5= Walking around your home 6= Dressing NOTE: The following self reported answers for D3 are Activities of Daily Living" (IADLs) which are not necessary used to define fundamental functioning, but permit the individual to live independently in a community. D3 continued for IADL. Categorical variable coded from: 7=g. Preparing meals 8=h. Managing your money 9=i. Doing household chores 10=j. Using the telephone 11=k. Taking medications properly
210 Outcome Variables: Enrollment in the VHA IDS Health Services Utilization Consisting of: ER utilization wit hin the past six months (self reported) Outpatient utilization within the past six months (self reported) Inpatient utilization within the past six months (self reported) Aim 1 and Aim 2 Items Used to Measure Variable Continued: Outcome Variable for Aim 1: Justificati on: Enrollment and utilization in the VHA IDS occur simultaneously. The use of the ABM is an ideal match for predicting enrollment variations. Enrollment: E1. Have you ever been enrolled in VA health care? Dichotomous outcome variable: 'Yes' or 'No' Outcome Variables for Aim 2 : Justification for use: Tests the VA claim that there is an institutional focus on a primary care delivery model which reduces unnecessary, emergency and inpatient care. It also tests this claim in conjunction with a f ragmented system of care outside the VHA IDS. E4. In the last six months, have you stayed in a hospital for medical or surgical care? Dichotomous outcome variable: 'Yes' or 'No' E5. In the last six months, have you had outpatient care for doctor visits, urgent care, routine exams, medical tests, or shots? Dichotomous outcome variable: 'Yes' or 'No' E15. In the last six months, have you visited or had care in an emergency room? Dichotomous outcome variable: 'Yes' or 'No'
211 AP PENDIX B DIFFERENCES IN MISSING VARIABLE DISTRIBUTIONS Missing Enrollment Values b y Missing Covariates Variable Missing Frequency Chi Square p value White Black Served in Combat or War Zone Exposed to Dead/Dying/Wounded Deployed to OEF /OIF Perce ived Health 258 258 167 152 373 117 2 .9839 2 .2009 0.015 8 0.0515 0.3213 3.5 008 0.988 8 0.9 387 0. 8905 0.7652 0.9 526 0. 0913 39 missing enrollment response values Missing ER Use Values b y Missing Covariates ** Variable Missing Fr equency Chi Square p value White Black Served in Combat or War Zone Exposed to Dead/Dying/Wounded Deployed to OEF /OIF Perceived Health 258 258 167 152 373 117 6.6755 2.6648 4.1424 0.3502 5.5107 7.1604 0.0598 0.1026 0.0618 0 .554 0.0589 0.1277 ** 160 missing ER response values
21 2 Missing Outpatient Use Values b y Missing Covariates *** Variable Missing Frequency Chi Square p value White Black Served in Combat or War Zone Exposed to Dead/Dying/Wounded Deployed to OEF /OIF Perceived Health 258 258 167 152 373 117 3 .3 775 3 .3053 0.012 0.0008 2.9775 5 .3053 0.0632 0 .0 611 0.9768 0.9126 0.0844 0.0708 *** 393 missing enrollment response values Missing Inpatient Use Values b y Missing Covariates **** Variable Missing Frequency Chi Square p value White Black Served in Combat or War Zone Exposed to Dead/Dying/Wounded Deployed to OEF /OIF Perceived Health 258 258 167 152 373 117 0.9839 0.2009 0.3339 0.4151 2.5958 7.4242 0.3212 0.654 0.5634 0.5194 0.1071 0.1151 **** 488 missing enrollment response values
213 LIST OF REFERENCES 1. L.A. Aday, R. Andersen., "A Framework for the Study of Access to Medical C are," Health Services Research 9, no. 3 (1974): 208. 2. P.D. Allison, "Missing data: Quantitative Applications in the Social S ciences. British Journal of Mathematical and Statistical Psychology 55 no. 1 (200 2): 193 196. 3. R.M. Andersen, "Revisiting the Behavioral Model and Access to Medical Care: Does it Mat ter?" Journal of Health and Social Behavior 36, no. 1 (1995): 1 10. 4. S.M. Asch, E.M. Sloss, C. Hogan, R.H. Brook, et al., "Measuring Underuse of Necessary Care Among Elderly Medicare Beneficiaries Using Inpatient and Outpatient Claims" Journal of American Medical Association 284, no. 18 (2000): 2325 2333. 5. C.M. Ashton, "Geographic Variations in Utilization Rates in Veterans Affairs Hospitals and Clinics," The New England Journal of Medicine 341, no. 1(1999): 32. 6. D.A. Barr, "Strategic Management or Env ironmental Change Which Determines Success in Health Care Organizations?" Administration & Society 30, no. 4 (1998), 374 392. 7. J.R. Baumgardner, "The Interaction between Forms of Insurance Contract and Types of Technical Change in Medical Care," The RAND Journal of Economics Vol. 22, No. 1 (Spring, 1991), pp. 36 53. 8. J. Blustein, E. C Hoy "Who is enrolled in for profit vs. nonprofit Medicare HMOs?" Health Affairs (Millwood) 19, (2000): 210 220. 9. S.J. Borowsky, D.C. Cowper, "Dual Use of VA and Non VA Primary Care," Journal of General Internal Medicine 14, no. 5 (1999): 274 280. 10. A. Bookman Family Caregiver Handbook Cambridge, MA: Massachusetts Institute of Technology, 2007 11. R.S. Brown, K. Langwell, "Enrollment patterns in Medicare HMOs: implications for access to care," Advances in Health Economics and Health Services Research 9, no. 1 (1988): 69. 12. J.F. B urgess, D.A DeFiore "The effect of distance to V A facilities on the choice and level of utilization of VA outpatient services," Social Science & Medicine 39, no. 1 (1994):95 104. 13. A.M. Burnam, L. Meredith T. Meredith, L Jaycox, "Mental Health Care For Iraq And Afghanistan War Veterans" Health Affairs 28, no. 3 (May/June 2009): 771 782.
214 14. D.F. Burrelli, Military Health Care and the Issue of 'Promised' Benefits (Washington, Congressional Research Service, Library of Congress, 2001). 15. K. Carey, M.E Montez Rath A.K Rosen C.L Christiansen et al., Health Care a nd Outcomes Report "Use of VA and Medicare Services by Dually Eligible Veterans with Psychiatric Problems," Health Services Research 43, no. 4 (2008): 1120 1164. 16. M.R. Chassin, R.W. Galvin, the National Roundtable on Health Care Quality, "The Urgent Need to Improve Health Care Quality," Journal of the American Medical Association 280, no. 11 (1998): 1000 1005. 17. Children's Health Insurance Program (CHIP), CHIP Program Information, http://www.medicaid.gov/Medicaid CHIP Program Information/By Topics/Childrens Health Insurance Program CHIP/Childrens Health Insurance Program CHIP.html (accessed 11Jan 2013). 18. D.G. Clement, S. Retchin, R. Brown, M. Stegall, "Access an d Outcomes of Elderly Patients Enrolled in Managed Care," Journal of the American Medical Association 271, no. 19, (1994): 1487 1492. 19. Congressional Budget Office (2007), The Health Care System for Veterans: An Interim Report, http://www.cbo.gov/sites/default/files/cbofiles/ftpdocs/88xx/doc8892/12 21 va_healthcare.pdf (accessed 1 December 2012). 20. United States http://etd.fcla.edu/UF/UFE0008224/cowper_d.pdf (accessed 4 June 2013). 21. Dartmouth Atlas of Health Care S tate of New York, http://www.dartmouthatlas.org/ (accessed 3 March 2013). 22. Department of Defense, TRICARE Benefits 2013, http://tricare.mil/ (accessed 11 Dec 2012). 23. Department of Defense Guidan ce, WarriorCare Fact Sheet, 2013, http://warriorcare.dodlive.mil/files/2011/10/IDES_Trifold_Brochure.pdf (accessed 1 May 2012). 24. Department of Veterans Affairs, Nationa l Center for Veterans Analysis and Statistics, "Veteran Population Projections Model," 2007, http://www.va.gov/vetdata/veteran_population.asp (accessed 11 January 2013).
215 25. Department of Vetera ns Affairs, National Center for Veterans Analysis and Statistics, "Department of the VA, FY 07 Information Pamphlet," 2007, http://va.gov/vetdATA/ (accessed 2 April 2012). 26. Depart ment of Veterans Affairs, Priority Group Ratings (2012) http://www.va.gov/healthbenefits/resources/priority_groups.asp (accessed 2 June 2013). 27. J.E. DeVoe, A. Baez H. Angier L. Krois et al. "Insura nce + Acc Typology of Barriers to Health Care Access for Low Income Families," Annals of Family Medicine 5, no. 6 (2007): 511 518. 28. DoD Instruction 6000.11, Patient Movement May 4, 2012 http://www.dtic.mil/whs/directives/corres/pdf/600011p.pdf (accessed 1 June 2013). 29. J.M. Eisenberg, E. J. Power, "Transforming Insurance Coverage into Quality Health Care," The Journal of the American Medical Association 284, no. 16 (2000): 2100 2107 30. A. Fontana, R. Rosenheck, "Treatment of Female Veterans with Posttraumatic Stress Disorder: The Role of Comfort in a Predominantly Male Environment," Psychiatric Quarterly 77, no. 1 (2006): 55 67. 31. J. Fortney, K Rost J. Warren Comparing Alternative Methods of Measuring Geographic Access to Health Services. Health Services and Outcomes Research Meth odology 1, no. 2 (2000): 173 184. 32. J.C. Fortney, B.M. Booth, F.C. Blow, J.Y. Bunn, "The Effects of Travel Barriers and Age on the Utilization of Alcoholism Treatment Aftercare, American Journal of Drug and Alcohol Abuse 21 (1995): 391 406. 33. P. Girard, "Military and VA Telemedicine Systems for Patients with Traumatic Brain Injury," Journal of Rehabilitation Research and Development 44, no. 7 (2007): 1017. 34. R. Grol, Improving t he Quality of Medical Care. Journal of the American Medical Association 286, no. 20 (2001): 2578 2585. 35. A.L. Grubaugh, K.M. Magruder H.M. Zi nzow B.C. Frueh, et al., "Use of Dual Health Care Services Among Veterans Seen in Veterans Affairs Primary Care Clinics," General Hospital Psychiatry 31, no. 6, (2009): 589 591. 36. J.H. Hibbard, C.R. Pope, "Age Differences in the Use of Medical Care in an HMO: An Application of the Behavioral Model," Medical Care 24, no. 1 (1986): 52 66.
216 37. D.U. Himmelstein, K. Lasser D. McCormick , D. Bor MD, et al., "Lack of Health Coverage Among US Veterans From 1987 to 2004," American Journal of Public Health 97, no. 12 (2007): 2199 2203. 38. J.J. Hisnanick., "The Relationship Between Veterans Socio Economic Factors, State Level Health Care Reform Legislation and the Deman d for VA Health Care," Policy Studies Review 17, no. 1, (2000): 32 52. 39. J.J. Hisnanick, J. John, G.S. Surinder, "Veterans' Health Insurance Status and Their Use of VA Medical Facilities: A Joint Choice Analysis," Social Science Quarterly (University of Te xas Press), 77, no. 2 (1996): 393 406. 40. C.W. Hoge, "Combat Duty in Iraq and Afghanistan, Mental Health Problems, and Barriers to Care," The New England Journal of Medicine 351, no. 1 (2004): 13 22. 41. J.S Hughes, "Can the Veterans Affairs Health Care System Continue to Care for the Poor and Vulnerable?" Journal of Ambulatory Care Management The Crisis Confronting Medicaid 26(4), (2003): 344 348. 42. D.M. Hynes, K. Koelling, K Stroupe, N. Arnold, et al., "Veterans' Access to and Use of Medicare and Veterans Aff airs Health Care," Medical Care 45, no. 3 (2007): 214 223. 43. New England Journal of Medicine 348, no. 22 (2003): 22 18 2227. 44. L.E. Kazis, D.R. Miller J. Clark K. Skinner et al., "Health related Quality of Life in Patients Served by the Department of Veterans Affairs," Archives of Inte rnal Medicine 158, (1998): 626 632. 45. L.E. Kazis, D.R. Miller K.M. Skinner A. Lee, et al., "Patient reported measures of health: the Veterans Health Study," The Journal of Ambulatory Care Management 27(1), (2004): 70 83. 46. K.W. Kizer, R.A Dudley, "Extreme Makeover: Transformation of the Veterans Health Care System." Annual Review of Public Health 30, no. 1 (2009): 313 339. 47. C.F. Liu, M.L. Maciejewski A.E. Sales, "Chang es in Characteristics of Veterans Using the VHA Health Care System between 1996 and 1999," Health Research Policy and Systems 3, no.1, (2005): 5. 48. K.N. Lohr, R. H Brook, C. Kamberg, G.A Goldberg, et al., "Use of Medical Care in the RAND Health Insurance E xperiment: Diagnosis and Service specific Analyses in a Randomized Controlled Trial. Medical Care 24, no. 9 (1986): S1 S87.
217 49. J.A. Long, D. Polsky, J.P. Metlay "Changes in Veterans' Use of Outpatient Care American Journal of Public Health 95, no. 12 (2005): 2246 2251. 50. P. Longman, Best Care Anywhere: Why VA Health Care is Better Than Yours 3 rd ed, San Francisco, Berrett Ko ehler Publishers, 2012. 51. J.F. McCarthy, F. C Blow M.Valenstein E. P Fischer et al., "Veterans Affairs Health System and Mental Health Treatment Retention Among Patients with Serious Mental Illness: Evaluating Accessibility and Availability Barriers," Health Services Res earch 42, no. 3 (2007): 1019 1042. 52. I. Moscovice, M. Casey, S. Krein, "Expanding Rural Managed Care: Enrollment Patterns and Prospects," Health Affairs 17, no. 1 (1998)172 179. 53. Central Reg ion, Issue Brief #7 , http://www.ruralhealth.va.gov/docs/issue briefs/eHealth Information Exchange 040311.pdf (accessed 21 Ju ly 2013 ). 54. K.M. Nelson, G.A. Starkeb aum G.E. Reiber "Veterans Using and Uninsured Veterans Not Using Veterans Affairs (VA) Health Care," Public Health Reports (1974 ) 122, no. 1 (2007): 93 100. 55. A. Oliver, "Th e Veterans Health Administration: An American Success Story?" Milbank Quarterly 85, no. 1, (2007): 5 35. 56. W.F. Page, "Why Veterans Choose Veterans Administration Hospitalization: A Multivariate Model," Medical Care 20, no. 3, (1982): 308 320. 57. R. Penchan sky, JW Thomas "The Concept of Access: Definition and Relationship to Consumer Satisfaction," Medical Care 19, no. 2, (1981): 127 140. 58. J.D. Penrod, T.D. McBride, K.J. Mueller, "Geographic Variation in Determinants of Medicare Managed Care Enrollment," Health Services Research 36, n o. 4 (2001): 733. 59. A. Percy, "A CBO Paper, Quality Initiatives Undertaken by the Veterans Health Administration," August 2009 http://www.cbo.gov/sites/defau lt/files/cbofiles/ftpdocs/104xx/doc10453/08 13 vha.pdf (accessed 3 June 2012). 60. M.C. Probus (Major), "Analyzing the Requirement for a Deployed Warrior Medical Management Center Table of Distribution and Allowances at Landstuhl Regional Medical Center" US Army Baylor University, Unpublished Thesis, February 2004, http://www.dtic.mil/cgi bin/GetTRDoc?AD=ADA432724 (accessed 4 June 2013).
218 61. Y. Shen, A. Hendricks F. Wang J. Gardner et al., "The Impact of Private I nsurance Coverage on Veterans' Use of VA Care: Insurance and Selection Effects," Health Services Research 43, no. 1, part 1, (2008): 267 286. 62. Y. Shen, A. Hendricks, S. Zhang, L.E. Kazis, "VHA Enrollees' Health Care Coverage and Use of Care," Medical Care Research and Review 60. no. 2, (2003): 253 267. 63. E.F. Taylor, J.T. Kullgren, C.G. McLaughlin, "Who Enrolls In A Program For Parents Of Publicly Insured Children?" Health Affairs, (Sep. 2003 ): 460 471. 64. U.S. Department of Health & Human Services Healthcare Fact Sheets, December 2012, http://www.healthcare.gov/news/factsheets/2010/07/veterans affordable care act.html (acce ssed 1 Dec. 2012) 65. U.S. Government Accountability Office, Department of Veterans Affairs: Key Management Challenges in Health and Disability Programs (Washington: GAO 03 756T, 2003a). 66. U.S. Government Accountability Office, Department of Veterans Affairs: Military Retiree Health Benefits: Enrollment Low in Federal Employee Health Plans under DOD Demonstration (Washington: GAO 03 547, 2003b). 67. U.S. Government Accountability Office, Department of Veterans Affairs: Limited Support for Reported Health Care Mana gement Efficiency Savings (Washington: GAO 06 359R, 2006). 68. U.S. Government Accountability Office, Department of Veterans Affairs: VA Uses a Projection Model to Develop Most of Its Health Care Budget Estimate to Inform the President's Budget Request (Washi ngton: GAO 03 547, 2011b). 69. U.S. Government Accountability Office, S.P. Backhus, VA Health Care: Progress and Challenges in Providing Care to Veterans (Washington: T HEHS 99 158,1999). 70. search http://www.hsrd.research.va.gov/publications/forum/apr13/apr13 4.cfm (Accessed 21 July 2013) 71. Westat, National Survey of Veterans, Active Duty Service Members, Demobilized National Guard and Reserve Members, Family Members, and Surviving Spouses: Department of Veterans Affairs Final Report (2010), http://www.va.gov/SURVIVORS/docs/NVSSurveyFinalWeightedReport.pdf (accessed 4 June 2013). 72. A.F. Wooten, Access to Mental Health Services at Veteran Affairs Community Based Outpatient Clinics. Military Medicine 167, no. 5 (2002) 424 426.
219 73. E.M. Yano, B. Simon , A. Lanto , and L. Rubenstein "The Evolution of Changes in Primary Care Delivery Underlying the Veterans Health Administration's Quality Transformation," American Journal of Public Health 97, no. 12 (2007): 2151 2159.
220 BIOGRA PHICAL SKETCH Robert Flemming received his bachelor's degree in International Business from the State University of New York in 1998, his master's degree in Healthcare Administration and master's degree in Business Administration degrees from the State Uni versity of New York in 2005. He received his Ph.D. in Health Services Research from the University of Florida in the Summer of 2013. Mr. Flemming began his career as a health services research assistant upon his acceptance into the Ph.D. program in Health Services Research, Management, and Policy at the University of Florida in August 2009. In addition to his research interests in organizational theory and health behaviors, Robert has lectured in class and developed r graduate students in the Management of Healthcare Organizations. Complementing these proficiencies is his continuing work as a US Air Force Medical Service Corps Officer, serving four years on Active Duty status as a healthcare administrator at Sheppard Air Base (AB), Texas; Yokota AB, Tokyo, Japan; and Keesler AB, MS. Captain Flemming was discharged in 2009 from Active Duty status and retained his commission as a Medical Service Corps Officer in the Reserves serving at Patrick AB, FL since this time. Mr. Flemming presently works for the Center for Medicare and Medicaid Innovation (CMMI) as an operations, research, and policy liaison for ongoing Federal and State run primary care medical home transformation demonstrations.