Cost-Benefit Analysis of Providing Non-Emergency Medical Transportation (2005)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Chapter 6: Method for Estimating Healthcare Costs and Outcomes In this chapter, we consider the optimal method to quantify the impacts of missed medical care on healthcare costs and health outcomes in light of the constraints imposed by the data. The work builds on the literature review presented in Chapter 2, and the set of non-emergency medical services needed by the target population identified in Chapter 3 and described in Chapter 4. From both data analysis and an extensive use of the literature, a comprehensive list of conditions that relate to those with transportation difficulties has been investigated. We discuss economic evaluation in the health domain and describe the optimal methods for attacking the problem at hand. In Chapter 7, these methods are applied to the eleven medical conditions we have selected for specific analyses. To clearly illustrate the methods, we review health economic evaluation information, present a general methods discussion that argues against a global (macro) cost benefit approach, and conclude by listing the steps required for the favored approach – a series of cost-effectiveness analysis case studies. 6.1 Review of Health Costs and Outcomes Evaluation Healthcare costs and benefits (outcomes) are often difficult to distinguish from one another due to how some analysts commonly conceive of them, for example, by counting an adverse outcome as a cost or counting reduced utilization due to an intervention as a benefit. In this project we follow the conventions in the cost benefit analysis and cost-effectiveness literature that relegate expenditures to the cost side of the ledger (the numerator), and identify and value outcomes–the difference in effectiveness between an intervention and the alternatives to which it is compared–in the benefit column (the denominator) (Drummond et al., 1987; Gold et al., 1996; Jefferson et al., 1996). 6.1.1 Healthcare Costs The healthcare costs associated with missed care include both the cost of the care forgone plus the cost of any care prompted by the care that was forgone, minus any care that is no longer needed because of better primary care. Our presumption is that missed healthcare often results in subsequent, more costly care. The classic healthcare example is a costly emergency room visit (including a potential hospital stay) prompted by missed primary care that could have prevented the emergency condition. The number of hospital emergency department (ED) visits reached a record high of about 114 million in 2003, a 26% increase from a decade earlier (McCaig and Burt, 2005). The U.S. population increased about 12% over this period and the 65-and-over population increased about 10%. Using the 2003 National Hospital Ambulatory Medical Care Survey Emergency Department Summary, the National Center for Health Statistics attributed the increase in ED visits to more adult usage, including those aged 65 and older. The report notes that Medicaid beneficiaries were four times more likely to visit EDs than were those with private Final Report 41

health insurance. Similarly, there is great potential for avoiding costly hospitalizations (Kruzikas et al., 2004). Healthcare costs derive from five principal utilization categories: • Hospitalizations (inpatient stays) • Emergency room visits • Outpatient visits (including diagnostic tests and labs) • Physician and other primary care provider visits (office-based visits) • Pharmacy costs To perform a health economic analysis, the added transportation costs representing the “intervention,” must be added to the new healthcare costs to compute total costs. Cost weights are not separately computed because the cost information comes directly from the MEPS data. However, for newly engendered office visits stemming from enhanced primary care or specialist evaluation and management visits, we will use Medicare cost weights. These are found in the Medicare Physician Fee Schedule for calendar year 2005 and are shown in Table 6-6 (Federal Register, 2004). 6.1.2 Health Outcomes Quantifying the impact of missed care on healthcare costs requires detailed study, partially completed via our comprehensive literature review. We also have consulted various groups that bring together experts in the field, most notably the Disease Management Association of America, individuals at the relevant disease associations, the National Committee for Quality Assurance, the Agency for Healthcare Research and Quality, and the work by Milliman consulting. We explained that we are attempting to estimate the healthcare visit requirements for various chronic diseases (asthma, diabetes, arthritis, heart disease, etc.) by examining the disease management literature. In this way, we can match the number of trips a person with transportation barriers, suffering from a particular condition, might require to be considered well- managed. We would like to know how many visits (and what type, if possible) a well-managed patient might have, per year, on a disease-specific basis (see Section 6.5). Similarly, we would like to determine the characteristics of a poorly managed patient on a disease-specific basis. In this way, we can match cost data by condition to derive the direct economic benefit of moving from poorly to well-managed care on the basis of better access to care as a result of improved transportation. Health outcomes can be divided into quantity (life expectancy or mortality) and quality components (illness or morbidity). Measuring quantity of life is unambiguous; assessing health-related quality of life is both difficult and inherently controversial. To counter these difficulties, health services researchers have adopted the quality adjusted life year (or QALY) as the primary currency in their studies. The QALY method combines duration of life and health-related quality of life into a summary measure. Researchers can then compare interventions across various diseases and affected populations. We discuss our approach to quality of life in further detail, in Appendix C. Final Report 42

6.2 Estimating Missed Trips from a Disease Perspective In this section, we discuss shortcomings in analytic approaches that rely on an estimate of the number of NEMT trips missed by the target population and explain why other approaches are superior, especially in the realm of preventive care. The resulting discussion sets the stage for use of cost-effectiveness analysis instead of a strict cost benefit analysis. 6.2.1 Identifying and Aggregating Missed Trips Estimating the number of NEMT trips missed by those persons who need non- emergency medical care and lack transportation – the target population – is a difficult task. The difficulty involves both feasibility and accuracy. Feasibility is low because the data described in Chapters 3 and 4 on the transportation-disadvantaged population in the NHIS and MEPS focus on respondents who report missing at least one medical visit over the past year, but they do not contain enough detail to estimate the number of missed trips per person or to sum across the entire population. Moreover, this survey does not include preventive care, because respondents would not perceive preventive care as missed visits. That is, by definition, one cannot self-report a missed visit that is not perceived as needed (and hence never scheduled). This point also pertains to visits that are not scheduled, because they are a component of an aggressive disease management protocol for chronic conditions that has not been instituted for a segment of the population. Based on our analysis of available data, we have concluded that there is no sound, accurate, nationally representative way to count and sum missed trips. The converse is not true; counting visits that are actually made is quite easy. On average, each American makes approximately 3.2 healthcare visits per year, excluding hospitalizations and emergency visits (Burt and Schuppert, 2004). These data can be further detailed according to factors such as age, sex, race, payment source, etc. For example, those aged 75 years and above have more than 7 visits per year. Table 4-8 shows extremely high average visits for those identified in the MEPS with the chronic conditions analyzed in Chapter 7. Even with an extensive modeling analysis that matches characteristics of individuals in our target population with data from the National Center for Health Statistics, we would still only be able to estimate the actual number of visits that were made, not those that were missed. As an alternative, we can straightforwardly compare the healthcare expenditures for individuals suffering from a disease depending on their transportation status. A separate comparison involves a synthetic approach for costing out poorly and well- managed cases by disease according to visit counts obtained from disease experts. 6.2.2 Comparison of Costs for Well and Poorly Managed Individuals The data limitations make it difficult to match transportation-disadvantaged individuals suffering from certain diseases with missed healthcare in a statistically meaningful way. To overcome this shortcoming, we have developed a method that evaluates missed trips from a disease perspective. Literature on disease management and standards of care guidelines or protocols often include data on number of visits required for a disease to be considered “well-managed.” For example, this literature Final Report 43

recommends that a patient suffering from mild to moderate asthma should see a primary care provider twice a year. Likewise, it recommends that patients with severe asthma see a primary care provider three times a year and a specialist once a year to ensure that their asthma is under control. Thus, for patients in the transportation-disadvantaged population (or others) who have asthma, the ideal number of visits to treat their asthma ranges from two to four per year. The number of trips missed is needed to estimate the cost of trips for the cost benefit analysis, but these trips also factor into the economic gains associated with more frequent and consistent care that prevents the contraction of a disease or the development of complications. For example, consistent access to transportation for an asthma patient over age one (2 to 4 trips) may prevent one trip to the emergency room or one hospitalization. This change in healthcare utilization provides the direct economic benefit that may offset the cost of providing the asthma care trips (and the cost of additional healthcare visits). Therefore, both the number of trips and the benefits of better health must be calculated on a disease-specific basis to provide a meaningful weighing of costs and benefits overall. We return to discussing the promise of this approach after considering issues related to a macro or global cost benefit approach. 6.2.3 Cost Benefit Analysis Issues Conducting an accurate analysis of costs and benefits by examining diseases and then aggregating trips and benefits across the transportation-disadvantaged population that miss visits, poses a number of challenges. The central problem concerns the indeterminateness and arbitrary nature of translating health benefits into monetary terms required to conduct a strict cost benefit analysis. That is, after accounting for any cost changes – increased costs from the intervention plus any added healthcare expenditures minus reduced healthcare expenditures because of better care – healthcare benefits would have to be evaluated in monetary terms to complete the cost benefit assessment. This is controversial and objectionable to most analysts as it forces one to make monetary valuations for persons suffering from various diseases. These objections have led to an almost exclusive application of on cost-effectiveness analysis in the health domain (Gold et al, 1996). Beyond the generic problems of conducting a strict cost benefit analysis, the transportation-oriented context of the current study raises additional problems. The crux of the problem with a macro-oriented analysis is that trips and visits do not relate one-to-one. Some visits will address a single, specific health concern. Others will address overlapping conditions for multiple diseases in a single visit. Therefore, estimating the number of trips per year associated with one disease can lead to inaccuracies when aggregating all of the trips and all of the disease-related benefits. In summing trips, we may overestimate the number of trips required by the transportation-disadvantaged population because one trip may provide a visit and therefore a health benefit, for a different disease that has a different benefit associated with better management. Although we can estimate the amount of trip layering in the transportation-disadvantaged population, the final results will have a wide margin of error, especially given the high prevalence of multiple, simultaneous chronic conditions found in the target population. Final Report 44

Instead of focusing on individuals with transportation barriers and aggregating their missed trips, in conjunction with changes in healthcare utilization, to get a macro- level analysis, the benefit of added transportation services is analyzed through a series of disease specific, cost-effectiveness analyses. In this way, there is no need to estimate the number of missed trips per transportation-disadvantaged individual, and the problems associated with trip- and visit-layering for persons with multiple chronic conditions are negated. These, in fact, may nevertheless inform and have implications for an aggregate cost benefit analysis. (A preliminary, macro cost benefit appraisal was produced. With plausible ranges for missed trips per disadvantaged persons, cost of added healthcare, and the added expense of transportation services, very little improvement is required in quality of health outcomes, or lowered unnecessary utilization, to derive a net positive cost benefit finding.) In short, cost-effectiveness analyses are more accurate, provide better information for policy makers, and give a better sense of program evaluation options. 6.3 Review of Cost-Effectiveness Analysis in Healthcare In this section we focus on the use of cost-effectiveness measures to assess the benefit side of the non-emergency medical transportation equation. A more detailed description is contained in Appendix C. To reiterate, we believe the best approach is a series of cost-effectiveness analysis studies corresponding with the most significant diseases from which the target population suffers. Cost-effectiveness analysis is a well-accepted method in health economics and health services research and is widely used to understand the value of healthcare outcomes associated with increased investment, for example, in transportation services. It serves to incorporate resource consumption into healthcare decisions, but does not directly value, in monetary terms, healthcare improvements. Rather, the cost of enhancing health is estimated and this can be compared via quality adjustments to a relative baseline. There is clear value of a condition-by-condition approach for evaluating the costs and benefits of providing non-emergency medical transportation to transportation- disadvantaged individuals. There is, however, an obvious trade-off between the number of conditions that are evaluated and the quality of these analyses. Even with relatively few conditions studied, those that contribute the most to the analysis of costs and benefits are captured. Three arguments serve to justify this statement: • An application of the Pareto rule to these conditions – a disproportionately small number of conditions account for a large proportion of costs and benefits • Recent results from the literature regarding condition costs and chronic condition overlap • New findings on the effectiveness of disease management procedures (or lack thereof). The examined diseases were drawn from the prevalence data in NHIS and MEPS, in conjunction with what we know from external evidence about disease conditions that benefit from careful monitoring and comprehensive primary care and account for high healthcare costs. In addition, the final list of conditions was reviewed and Final Report 45

approved by the panel convened for this project by the Transit Cooperative Research Program (TCRP) within TRB. Addressing the major conditions provides instructive examples and will be of sufficient value. There is limited interest in secondary conditions such as irritable bowel syndrome, especially as compared with the central diseases such as asthma, heart disease, and chronic obstructive pulmonary disease (COPD). Also, because some percentage of the supplied trips result in the treatment of two or more conditions, a correction factor can be integrated into the final analysis. While transportation cost analysis provides triangulated, reliable information, the healthcare cost-effectiveness analyses are necessarily illustrative. In addition to accounting for limitations in the available data, the approach taken is strongly supported by two recent themes in the literature. The first points to the vast potential of novel disease management strategies (U.S. Congressional Budget Office, 2004). The second aims to compute the number of persons with: (1) chronic illnesses, (2) disabilities, and (3) functional limitations; as well as the various overlaps between people with combinations of these conditions, including those with multiple chronic diseases (Anderson, 2005; Anderson and Knickman, 2001; Partnership for Solutions, 2002). More than 100 million Americans fall into one of the three groups and nearly 10 million are in all three. Chapter 4 illustrated that there is close alignment with these individuals and those that we characterize as transportation disadvantaged and missing medical care due to a lack of access to NEMT. Nonetheless, these individuals are extremely high users of healthcare despite the barriers they face getting to encounters. We have stressed that inordinately high disease prevalence, multiple simultaneous diseases, and high disease severity explain high healthcare utilization by those with transportation difficulties. Another factor is the likelihood that individuals who lack transportation, particularly those in urban settings, live in less healthy environments and therefore require more visits. Research clearly demonstrates that a significant portion of overall healthcare cost inflation derives from a small set of healthcare conditions – on the order of 30 percent of cost growth is accounted for by five conditions (Thorpe, 2004). These and related findings strongly argue that a condition-specific method, in which a selective set of conditions is intensively studied, is superior to a large set of conditions studied with insufficient detail. 6.4 Using the MEPS for Cost-Effectiveness Analysis The MEPS is used for cost-effectiveness analysis because it is the richest source of nationally representative, health utilization and expenditure data, and it also contains information suitable to estimating QALYs. These advantages are described below. 6.4.1 QALY Information in the MEPS The MEPS collects two measures of health status on all respondents, the Short-Form 12 and the EuroQol 5-D (Fleishman, 2005). These are two of the more widely used health status metrics. Each relates to QALY measurements, with the latter enabling direct QALY calculations (Gold et al., 1996). Accordingly, QALY information can be directly integrated into the individual evaluations that depend on MEPS cost and Final Report 46

utilization data. Table 6-1 gives the means of the EuroQol 5-D broken down by whether the individual falls into the transportation-disadvantaged target population, and according to insurance status. Table 6-1: EuroQol 5 D Results from the MEPS Transportation UNINSURED EuroQoL 5-D Status ALL OF 2001 N Population Mean Non-Target Population Yes 3,045 22,456,521 0.8548 Non-Target Population No 17,087 164,035,341 0.8195 Target Population Yes 76 526,597 0.7421 Target Population No 210 1,769,281 0.5601 N/A N/A 11,704 95,459,587 N/A Total 32,122 284,247,327 6.4.2 Using the Richness of the MEPS for Cost and Benefit Analysis The initial analytical plan involved linking the NHIS data to the MEPS so that the NHIS’s detailed condition information would be supplemented with the MEPS’s rich expenditure data. Special linkage disks were obtained via a data user’s agreement with the U.S. Agency for Healthcare Research and Quality. Unfortunately, after reviewing the preliminary linked data, it was discovered that the final sample of MEPS respondents would be effectively cut in half (from about 32,000 to 15,000), because only an NHIS sub-sample contains the crucial transportation question that carries through to the linked MEPS data. Given the other virtues of the MEPS data, including its own transportation- disadvantaged designation and the close agreement between this measure and the one in NHIS, we determined that it could stand on its own for the subsequent cost- effectiveness analyses. The MEPS is the preeminent, nationally representative healthcare cost and utilization dataset in the United States, and it includes extensive encounter and cost data broken down into five categories: 1. Inpatient stays 2. Outpatient visits 3. ER visits 4. Office-based visits 5. Pharmacy costs. The MEPS data provide significant information on transportation-disadvantaged persons and their use of health services. The study population – those who miss healthcare visits due to a transportation barrier – is described in detail, and contrasted with the rest of the U.S. population, in Table 6-2. The weighted frequencies project the survey sampling onto the entire U.S. population using sophisticated statistical procedures. Final Report 47

6.4.2.1 Demographic Information Comparing those who we believe to have missed healthcare due to transportation factors, with all others in the survey, Table 6-2 shows that the former group has more older adults, includes more females and minorities, and its members are more likely to have come from households with yearly income under $20,000 (this figure is low due to the focus on individuals, hence children, in the MEPS v. families or households). Table 6-2: Demographic Review of the Target Population and Rest of the U.S. Weighted Frequencies Weighted Percentages Age Rest of U.S. Population Target Population Rest of U.S. Population Target Population 0-15 63,940,806 848,675 22.8% 24.6% 16-24 34,529,988 419,293 12.3% 12.2% 25-39 59,061,624 683,714 21.0% 19.8% 40-64 86,959,976 985,529 31.0% 28.6% 65+ 36,308,457 509,266 12.9% 14.8% Totals 280,800,851 3,446,477 100.0% 100.0% Sex MALE 137,147,041 1,483,896 48.8% 43.1% FEMALE 143,653,809 1,962,580 51.2% 56.9% Totals 280,800,850 3,446,476 100.0% 100.0% Race AMERICAN INDIAN 2,558,716 51,171 0.9% 1.5% ALEUT, ESKIMO 99,946 ASIAN OR PACIFIC 11,513,628 142,118 4.1% 4.1% BLACK 35,483,711 464,591 12.6% 13.5% WHITE 231,144,849 2,788,595 82.3% 80.9% Totals 280,800,850 3,446,475 100.0% 100.0% Personal Income $20,000 or more 110,680,755 489,435 39.4% 14.2% Less than $20,000 170,120,095 2,957,041 60.6% 85.8% Totals 280,800,850 3,446,476 100.0% 100.0% Source: 2001 MEPS Data 6.4.2.2 Insurance Status of the Target Population Transportation-disadvantaged persons who miss healthcare due to a lack of access to NEMT are more likely to be uninsured than those who do not miss healthcare for transportation-related reasons. Of the target population, 22 percent were uninsured for all of 2001, while only 12 percent of the others were uninsured for the entire year. Table 6-3 shows the proportion of people in each of the insured categories. Final Report 48

Table 6-3: Insurance Status of the Target Population Rest of U.S. Population Target Population Uninsured 32,357,569 772,098 Weighted Frequency Insured 248,443,282 2,674,378 Uninsured 12% 22% Weighted Percentage Insured 88% 78% 6.4.2.3 Utilization of Healthcare Services The target population is much more likely to have an inpatient stay and emergency room visit as well as have more prescriptions written for them. Table 6-4 shows that for each of these indicators of utilization, the target population was about twice as likely to use these services. This finding confirms earlier analysis that indicates the target population suffers from diseases at a higher rate and also experiences multiple, chronic conditions. Table 6-4: Utilization of Services Inpatient Stays (per 1,000 pop) Outpatient Visits (per 1,000 pop) ER Visits (per 1,000 pop) Office Based Visits Rx Scripts Target Population 212 652 464 7.1 17.0 Rest of U.S. 105 524 190 4.7 8. 7 % Difference 103% 24% 144% 49% 96% 6.4.2.4 Per Capita Expenditures by Category While the median per capita costs of healthcare for the target population is significantly higher than the cost for the rest of the U.S. population, the cost categories that appear to drive the total per capita cost are home health and prescription costs. Table 6-5 shows the weighted per capita cost for each of the cost categories included in the MEPS database. It is not surprising that the per capita costs for outpatient care are less for the target population. This further demonstrates how difficult it is for transportation- disadvantaged persons who miss medical care due to a lack of access to transportation to obtain care. Table 6-5: Weighted Median Per Capita Healthcare Costs by Category Unweighted Sample Total Healthcare Expenses Inpatient Out patient ER Rx Office-Based Medical Provider Dental Home Health Other Target Population 454 $1,874 $4,862 $310 $336 $644 $446 $184 $2,156 $141 Rest of U.S. 31,668 $1,095 $5,281 $547 $357 $312 $307 $178 $928 $157 % Difference 71% -8% -43% -6% 107% 45% 3% 132% -10% Final Report 49

6.5 Establishing the Benefits of Well-Managed Care There are established standards of care intended to prevent complications for chronic diseases widely prevalent in the United States. When a patient receives well- managed care, his or her disease is under control, complications are minimized, costly care is avoided, and quality of life is enhanced. A disease that is uncontrolled may be a product of patient non-adherence (or noncompliance) with prescribed care or clinical mismanagement or both. Transportation barriers can be attributed both to external barriers for clinical management and to issues of patient access and adherence to treatment (Javors et al., 2003). Because transportation barriers are a factor in poor disease management, it is reasonable to assume that data on poorly managed patients will include transportation-disadvantaged patients. Data from MEPS confirm that transportation-disadvantaged patients experience higher rates of disease complications than the general public. Thus, for the purposes of this economic evaluation, transportation-disadvantaged individuals are considered part of a poorly managed patient population, and the utilization rates and costs will be derived from this perspective. To efficiently analyze the diseases on our list, we focused our effort on diseases and conditions that have proven financial benefits through better-managed care. Conditions such as diabetes, cardiovascular disease, and asthma, have been widely discussed in the disease management literature and are considered to be the best targets for preventing complications and costly healthcare services using early intervention (Ofman et al., 2004). Other diseases, when the literature on the number of visits and the economic benefits of disease management are relatively unknown, will not be considered. In Chapter 7, we will analyze the data on well-managed patients compared with poorly managed patients suffering a particular disease and estimate the economic benefits of moving a patient into a well-managed state. This gain, reduced to account for patient and provider compliance, will represent the benefit of providing non- emergency medical transportation for the specific disease being analyzed. The data on disease management programs and effects provides information on how poorly managed patients can become well managed through more frequent preventive care visits and monitoring (nurse phone calls to home or employing sophisticated remote monitoring equipment), home healthcare, prescription drug adherence, etc. It is critical that the number and type of visit required to achieve proper management is understood. To obtain insights into this issue, we contacted the research director of the Disease Management Association of America. Through a series of consultations with experts in disease management research, we received suggestions on how to determine the appropriate number of visits per year for each disease, e.g., Karen Fitzner, research director of the Disease Management Association of America, www.dmaa.org, contacted via email on October 15, 2004. From our discussions, we have concluded that the Milliman Care Guidelines also provide useful analyses of disease-specific healthcare utilization. The guidelines include utilization data for ambulatory, inpatient, surgical, and home care health services to provide “best practice” information. They are “drawn from analysis of thousands of abstracts, articles, databases, textbooks, nationally-recognized guidelines and practice Final Report 50

observations, they synthesize the latest medical knowledge and best practices across the United States” (Milliman Consulting, 2004). To prevent overestimation of the benefits of providing non-emergency medical transportation to a poorly managed, transportation-disadvantaged population, we included a noncompliance factor in the analysis. This factor will account for the providers who do not adhere to standards of well-managed care, patients in the disadvantaged population who do not adhere to treatment, and those patients whose disease is considered uncontrollable, despite the best efforts of the provider and patient. In a study of compliance with clinical management guidelines for cardiovascular disease (CVD), diabetes, and asthma, 90 percent of patients were compliant with their care. Healthcare providers varied in their adherence to national standards from 71.4 percent compliance with CVD to 42.9 percent compliance with diabetes guidelines (Javors et al., 2003). The compliance factor is equal to the percentage of patients who had improvements in their health as a result of better disease management. The noncompliance factor, therefore, is the percentage of patients enrolled in a disease management program whose healthcare did not change. The hypothesis is that patients with transportation barriers are likely to receive poorly managed care, thus using a compliance factor based on disease management program effects is consistent. This factor is used to reduce the net benefit for the disadvantaged patients who may receive well-managed care through non-emergency medical trips. The equation is: [Compliance factor * (poorly managed cost – well managed cost)] – [# of visits * (cost of transportation + cost of medical visit)] = Net Costs 6.6 Benefits and Costs of Providing Transportation for Chronic Medical Conditions: Analytical Steps The steps to calculate benefits and costs of providing transportation for the chronic medical conditions that are analyzed in Chapter 7 are described below. 1. Select all survey respondents in MEPS who did not miss a trip to the doctor due to transportation problems and who had health insurance during the entire year (2001). 2. Review the characteristics of well- and poorly managed care, for each condition, through the literature. Well-managed patients use appropriate drugs, have their care monitored via a sufficient number of healthcare provider visits, and take other steps to stay healthy and avoid unnecessary emergency room visits. Poorly managed patients fail to adhere to drug therapy, do not have sufficient provider visits, and otherwise do not take steps to remain as healthy as possible. 3. Isolate the well-managed patients and review their per capita cost of care. These patients represent the expected costs for well-managed patients with transportation deficiencies addressed. Similarly, identify poorly managed patients. These patients represent expected costs for patients who are not Final Report 51

well managed for many reasons, including transportation deficiencies. Likewise, review their per capita cost of care. 4. Determine the compliance factor(s) for the disease in question from the literature. This will typically involve a large range so sensitivity must be addressed. 5. Determine from the literature review the number and type of visits (if available) required to manage a patient with the given disease adequately. For example, seeing a primary care provider alone is often insufficient to properly treat a chronic condition. 6. Use the Medicare Fee Schedule cost weights to determine the average medical cost of the required visits. The fee schedule contains five levels of evaluation and management physician visits for established patients, ranging from basic to extensive. (We use the costs for established patient visits instead of new patients because of the underlying assumption that these patients are being well managed and thus have a usual physician provider.) The middle level visit cost is typically applied. The codes and payment amounts are shown in Table 6-6. Table 6-6: Medicare Fee Schedule Evaluation & Management Visits for Established Patients Fee Codes Description & Typical Visit Length Payment ($) 99211 Minimal presenting problem(s); may not require physician; 5 minutes 21.60 99212 Self limited or minor presenting problem(s); 10 minutes 38.66 99213 Low to moderate severity presenting problem(s); 15 minutes 52.68 99214 Moderate to high severity presenting problem(s); 25 minutes 82.62 99215 Moderate to high severity presenting problem(s); 40 minutes 120.14 7. Determine the cost of required paratransit trips. This cost depends on rural or urban location and whether the patient is mobile or requires a modified vehicle for travel. 8. Incorporate health-related quality of life adjustments so that the analysis will correspond to the QALY methodology. For each respondent, MEPS collects EuroQol data that comprises an accepted quality of life measure. This provides a preference-based index that ranges on a scale from 0 (“worst possible health”) to 100 (“best possible health”). Because we can compute an average score across any subset of the population, we are able to obtain quality of life measures for the poorly- and well-managed subsets. 9. Summarize the results in a table that illustrates ranges for compliance and other factors. Note that while cost savings from healthcare expenditure reductions are anticipated, they are not required for a cost-effective outcome due to the expectation of quality of life improvement. Final Report 52

6.7 Benefits and Costs of Providing Transportation for Preventive Health The expenditure data contained in the MEPS, while tremendously useful for the chronic conditions, cannot be straightforwardly applied to the analysis of preventive cases. Accordingly, we apply a literature-based approach to these. 6.8 Summary and Discussion of Healthcare Cost Methods Our preferred method to address the central concern of this study relies on established and peer-reviewed criteria for judging well- and poorly managed care that are applied to separate the population into two groups and to obtain the per capita cost results for the important conditions under review. Only the population that is insured and not transportation-disadvantaged is used for this data analysis. Additional parameters are required to complete the cost-effectiveness studies, especially the integration of QALY information, and variations to key cost factors must be considered. Alternately, cost differentials could be calculated by directly comparing transportation-disadvantaged individuals to advantaged ones (with or without insurance). These variations are pursued in the next chapter and comprise an essential role in the consideration of sensitivity analysis, that is, a comprehensive comparison of results considering appropriate ranges for key variables. Numerous studies document the benefit of early interventions and frequent physician contact for chronic conditions like asthma. However, these benefits, or rather, the avoided costs of complications, do not manifest in the patient immediately. In the case of diabetes, hypertension, congestive heart failure, cancer and dental problems, the benefits of prevention are delayed – in fact some of the complications associated with poor management or prevention of these conditions do not develop for decades. Complications, such as heart failure, stroke, disability or early death, are quite costly, and cost-effectiveness literature supports disease management for each of these conditions, as discussed further in Chapter 7. The cost-effectiveness model developed for the chronic conditions presented in this report relies on the comparison of healthcare costs for well and poorly managed patients at a disease-specific level within MEPS. The current analysis compares the costs for patients within a one-year time frame using MEPS data. This is a suitable structure for asthma, depression, ESRD, and COPD, all of which present an immediate benefit to the well-managed patient, and an immediate cost to the poorly managed patient. MEPS data clearly show higher emergency service or hospital utilization due to complications in the poorly managed populations with conditions like asthma, depression, ESRD, and COPD. A significant limitation of this method is the lack of longitudinal data required to truly characterize the complications avoided by disease management for diabetes, hypertension and congestive heart failure. It is possible patient with diabetes can be poorly managed yet not exhibit complications (and higher healthcare utilization) until many years have passed. Since the MEPS data only shows the healthcare costs for one year, patients who are poorly managed may have much lower utilization because Final Report 53

they do not regularly see a physician, despite the likelihood of much higher future healthcare costs. While MEPS provides significant detail on healthcare costs and disease burden, there is insufficient data to determine, within diseases, the number of years the patient has had the condition, or the severity of the disease. Therefore, comparing well and poorly managed patients with diabetes, hypertension, or congestive heart failure is likely to discount the true future value of patients in the well-managed group, and will largely underestimate the future costs of those in the poorly-managed population. Despite these limitations, we are confident this methodology will be useful to decision makers who are considering NEMT based on immediate costs and benefits. For chronic conditions like diabetes, hypertension, and congestive heart failure, the benefits of providing NEMT will not become cost saving for some time. Instead, the one-year snapshot of the benefits of NEMT at the condition level presents the real costs of ensuring patients receive preventive or timely care over a one-year timeframe. The literature on the longitudinal benefits of disease management for diabetes, hypertension, and congestive heart failure are presented in the cost effective analyses in Chapter 7 and indicate the long-term benefits of early intervention for these conditions. Final Report 54