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 5: Cost of Providing Non-Emergency Medical Transportation Because this study focuses on the population of transportation-disadvantaged individuals who miss non-emergency medical care due to a lack of available transportation, we must examine the costs of providing additional transportation to serve the needs of this population to determine if the net healthcare benefits of this transportation are worthwhile. Thus, the primary purpose of this chapter is to develop cost estimates for several categories of non-emergency medical transportation (NEMT) so that these costs can be used in analyses of the cost- effectiveness of providing needed NEMT to the target population identified in Chapter 3 for treating the conditions faced by this population identified in Chapter 4. As with other transportation costs, NEMT costs vary by location (region of the country and urban or rural) and by the type of transportation service required (i.e., ambulatory, stretcher, wheelchair-equipped, etc.). Other factors can also play a role in transportation costs. For example, trip costs can depend on the technical and managerial skills of the organizations managing the transportation service. In Kentucky, for example, improved operational efficiency and control practices resulted in nearly a 20 percent drop in the cost of an average trip (O’Connell et al., 2002). To keep the current study manageable, however, we focused exclusively on cost variation by location and service type, and sought to develop nationally representative cost estimates. Specifically, we developed estimates of the cost per trip for ambulatory, stretcher, and wheelchair transportation service for both urban and rural areas, along with an estimate of per trip costs for fixed-route transportation in urban areas, though our target population is less likely to use fixed-route transit due to lack of availability, or their physical limitations. One difficulty in arriving at NEMT cost estimates for our target population stems from distinguishing the costs of non-emergency medical trips from other transit and paratransit trips serving transportation-disadvantaged persons. In many states, and for many public transportation operators, some vehicles serve a variety of trip purposes. While data from both Michigan (Wallace, 1997) and Florida (Florida Commission, 2003) indicate that medical trips dominate, they certainly are not the only trips provided. In our analysis, we minimized this potential problem, however, by focusing on medical trips in forming our cost estimates. According to federal policy, the Medicaid program will pay for transportation to non-emergency medical care for those who cannot otherwise arrange or pay for travel to this care. Medicaid, which generally serves the poor, is joined at the federal level by Medicare, which generally serves older adults. Unlike Medicaid, however, Medicare pays only for emergency transportation (i.e., by ambulance). As discussed in Chapter 2, this policy can result in non-emergency care being accessed by emergency transportation. The cost savings derived from switching service types alone can be substantial, but these are beyond the scope of the current study, which examines only missed care. For the purposes of this study, non-emergency medical transportation (NEMT) refers to trips analogous to those currently provided under the auspices of Medicaid, including a variety of transit and paratransit services provided by public agencies and private entities. One past estimate of the cost of this transportation service, deriving from a survey conducted by the Community Transportation Association of America (CTAA) Final Report 35

completed in 2000 (CTAA, 2001), showed an average cost of about $16 nationally for Medicaid trips, independent of transportation service type. In the following sections, we develop per trip costs estimates for paratransit and fixed-route transit, with the former focused on costs of Medicaid trips and divided into ambulatory, wheelchair, and stretcher services. In addition, we briefly describe how to associate transportation service types (or modes) with medical conditions. The latter is required to examine cost-effectiveness of additional transportation access for treatment of specific conditions, shown in Chapter 7. 5.1 Paratransit Cost Estimates We have determined average, per trip costs for non-emergency medical (NEM) paratransit for three service categories (ambulatory, wheelchair, and stretcher) for both urban and rural areas, resulting in six cost categories. Note that ambulatory service refers to transportation for someone who is physically capable of accessing any type of ground transportation. Wheelchair service refers to vehicles capable of safely transporting a wheelchair-bound person, such as a specially equipped van or paratransit vehicle. Stretcher service refers to NEMT for someone who cannot sit on their own; this will often mean transport by ambulance, though not necessarily one that is EMT/EMS-equipped. Thus, we retain focus exclusively on non-emergency care and trip needs; emergency trips by ambulance are not part of the current study. These costs are listed in Table 5-1 and derive from data from twenty different services in New England, the Mid-Atlantic States, the south, and the west. One Midwestern service is represented in the rural category, but the Pacific Northwest is not represented at all. Some of the services in our data are in the same states. Respecting the proprietary concerns of the organizations providing the data, we have not listed the costs per service area or the exact locales from which the data derive. As shown in the table, our one-way cost estimates are based on more than 800,000 trips (from year 2004), and they match expected trends, with higher levels of service (wheelchair and stretcher) displaying higher costs and rural areas generally displaying higher costs than urban areas. To make these estimates, we created a weighted average within each service type (or mode) and urban or rural combination (average cost per trip for the service provider based on the number of trips for that service). In fact, in the data we used, “service provider” probably is not an accurate term for most of the reported services. Rather, these data are based on service provision within a given geographic region, but several different providers were used within the region. In that respect, the data accurately reflect the diverse mix of transportation service providers drawn upon within any region, even by public transportation agencies that subcontract out various paratransit trips. As a check on the veracity and reasonableness of our estimates, we made two additional comparisons of our results to other information. First, we compared them with the costs reported in a Request for Proposals let by the Commonwealth of Pennsylvania for NEMT in Philadelphia County in early 2005. In this RFP (Commonwealth of Pennsylvania, “Request for Proposals for Philadelphia Medical Assistance Transportation Program (MATP)”), paratransit NEMT costs (undistinguished as to ambulatory or wheelchair) are reported as “almost $22 per Final Report 36

one-way trip”—a number that lies in between our urban ambulatory and wheelchair estimates. Furthermore, the American Public Transportation Association (APTA), based on the 2002 National Transit Database, indicates an average operating cost per demand-response trip (independent of trip purpose) of $18.86, and this number closely matches our estimate for urban, ambulatory trips—the category most likely to resemble NTD data. As a result, we have confidence that the estimates are representative, especially for the ambulatory and wheelchair service types. Table 5-1: Estimated One-Way Trips Costs for NEMT by Paratransit Service Category (2004) Service Type Region Number of Trips in Estimate (Sample Size) Average Cost Per One-Way Trip ($) Ambulatory Urban 420,435 19.95 Ambulatory Rural 276,705 20.95 Wheelchair Urban 111,384 28.52 Wheelchair Rural 87,121 33.02 Stretcher Urban 4,173 89.68 Stretcher Rural 7,805 86.20 5.2 Fixed-Route Cost Estimates Although much of the target population for this study is unlikely to use fixed-route transit—either because of this population’s overall poor health and high prevalence of co-morbidities or because fixed-route transit is not available—certainly ambulatory members of this population could use fixed-route transit if it were available to them, and they could afford to pay for the trip (or it were provided for free via some sort of subsidy, as is the case for Medicaid NEMT); furthermore, even those requiring wheelchair service could access fixed-route transit in some circumstances. Thus, we turned to the American Public Transportation Association’s Web site for easy access to National Transit Database (NTD) information. According to APTA (based on data from the 2002 NTD), the national average operating cost for fixed-route service (including buses, trolley buses, and rail) is $2.38 per unlinked trip (these calculations are detailed in Table 5-2). Furthermore, APTA estimates that 10-30% of trips require multiple (more than one) links. Using the middle value (20%) and assuming that exactly two links are needed for multiple link trips, we arrived at a national average cost per trip of $2.86 (1.2 times $2.38). Because the NTD best represents urban providers, we believe that limiting use of this estimate to urban fixed-route service is appropriate. Thus, in our later analyses, we will not assign any rural NEMT to fixed-route service. Final Report 37

Table 5-2: Cost Estimates for Fixed-Route Service Based on APTA Web Site (and NTD Data) Mode Operating Costs ($1000s) Number of Unlinked Trips (1000s) Cost ($) per Unlinked Trip Bus 14,065,603 5,867,945 2.40 Trolley Bus 186,714 115,968 1.61 Commuter Rail 3,003,211 414,253 7.25 Heavy Rail 4,267,460 2,687,973 1.59 Light Rail 778,274 336,531 2.31 Other Rail 187,768 26,214 7.16 Total 22,489,030 9,448,884 2.38 5.3 Linking Medical Conditions to Transportation Service Type In addition to developing cost estimates for the transportation-service types that could potentially provide unmet demand for NEMT, to complete the cost-effectiveness analyses we also need to estimate the relevant percentages of trips assigned to each service type (or mode) by location (urban or rural) and medical condition for the eleven conditions that are part of this study. To accomplish this, we made use of questions related to respondents’ functional limitations in the 2001 NHIS (2001 was selected because it is the latest dataset that includes an urban-rural designation). Based on responses to these questions for our target population (these questions deal with ability to walk a quarter mile with no difficulties, climb ten steps without need for special equipment, and the like), we estimated the likely split across transportation service type (or mode) within each condition, and then used these splits to determine an average trip cost by condition. Because these estimates are somewhat sensitive to assumptions made about the effects of these functional limitations on transportation service needs, as well as on assumptions about the percentage of trips that are provided by fixed-route public transportation, we have developed low, intermediate, and high estimates for each of the eleven conditions, as shown in Table 5-3. In the cost-effectiveness analyses presented in Chapter 7, we employ all three values to arrive at a range of possible estimates. Chapter 7 also investigates sensitivity based on compliance with prescribed care arising from healthcare visits and on the required number of NEM trips per capita. The following assumptions were made to arrive at weighted estimates based on the varying costs for paratransit and fixed-route transit by location in Table 5-3: • Low Estimate: All persons in the target population represented as having none of the functional limitations asked about are assumed to need only ambulatory services; for those in urban areas with no limitations, 100 percent are assumed to travel via fixed-route transit (the main assumption that makes this estimate “low”), while similar persons in rural areas are assumed to use ambulatory paratransit. Those represented by some functional limitations but no need for special equipment (such as a wheelchair) are also all assumed to use ambulatory paratransit. Finally, those represented as needing special equipment are assumed to make wheelchair trips via paratransit. Final Report 38

• Intermediate Estimate: Again, all persons in the target population represented as having none of the functional limitations asked about are assumed to need only ambulatory services; half of those in urban areas are assumed to use fixed-route transit, and all in rural areas are assumed to use ambulatory paratransit. As with the low estimate, those represented by some functional limitations but no need for special equipment (such as a wheelchair) are also all assumed to use ambulatory paratransit, and those represented as needing special equipment are assumed to make wheelchair trips via paratransit. • High Estimate: For both urban and rural members of the target population, all those represented as having no functional limitations and all those represented as having one or more limitations but no need for special equipment are assumed to travel via ambulatory paratransit service. Those needing special equipment are all assumed to require stretcher transportation service, thereby producing a high estimate that is not likely to be exceeded for any region or locale. Thus, it is a true high estimate. Table 5-3: Average per NEMT One-Way Trip Cost by Condition Medical Condition Low Estimate ($/one- way trip) Intermediate Estimate ($/one-way trip) High Estimate ($/one-way trip) Depression or Other Mental Health Problem 19.58 22.22 41.24 Hypertension 21.40 22.33 43.46 Heart Disease 21.94 22.83 46.05 Asthma 21.07 21.85 37.93 Chronic Obstructive Pulmonary Disease 21.18 21.67 33.92 Diabetes 19.58 22.22 41.24 End-stage Renal Disease 22.39 22.87 42.20 Dental Problems 13.08 16.63 20.18 Cancer 21.93 22.22 37.43 Premature Births 13.08 16.63 20.18 Vaccinations 21.93 22.22 37.43 For three of our conditions (vaccinations, dental problems, and depression), MEPS lacked data on functional limitations. For these, we assigned the costs from those for known conditions using a one-to-one mapping of unknown costs to known costs. Specifically, we assigned cancer transportation costs to vaccination transportation costs (both preventive care), prenatal care transportation costs to dental-problem transportation costs (again, both preventive), and diabetes transportation costs to depression (both chronic conditions). Because the costs do not vary considerably by condition, we believe that this compromise is minor and necessary to allow further examination of these three conditions. Final Report 39

5.4 Discussion of NEMT Cost Estimates While we have used available data as judiciously as possible to develop our cost estimates, numerous uncertainties still exist in combining these costs with healthcare costs and benefits, as is done in Chapter 7. This section comments on some of these uncertainties. By using the spreadsheet tool that we have developed and that we describe in Chapter 8, state, regional, and local agencies, as well as other interested entities, will be able to tailor analyses specific to their own environment. To the extent that these uncertainties are less pronounced for well-defined transportation service areas, users of the spreadsheet tool will be able to specify more precise transportation cost estimates for their geographic region and thereby minimize the effects of these uncertainties. One limitation in our transportation cost estimates is that currently unmet trips, by definition, may differ from those that have been taken and are represented in existing datasets. For example, trips not taken may be longer in time or length or occur at peak travel times, thereby increasing their cost relative to the average. Thus, though we focused on Medicaid trips and believe that these are a good analog for our targeted missed trips, the latter may well be more or less expensive to provide in any specific locale. Thus, we have also developed a spreadsheet tool to accompany this report that will allow for local or regional tailoring of these cost estimates to accurately reflect local conditions. A second limitation relates to the difference between average and marginal costs. Because we use average costs for our healthcare analyses, we chose to use average transportation costs to facilitate like comparisons. In the transportation industry, however, marginal costs can differ significantly depending on the percentage of available capacity that is being used. For example, the incremental cost of providing an additional trip for an under-utilized service will be very low (vehicles, vehicle operators, etc., are already paid for, and labor is often the most expensive component in paratransit, thereby increasing trip cost as trip distance increases). As another example, the marginal cost of providing one more ambulatory or wheelchair trip on fixed-route transit is effectively zero. The incremental cost of an additional trip for a service already at capacity, on the other hand, or one for which trip distances are large (such as in rural areas), may be extremely high. Further complicating matters, and leading to a third limitation in our estimates, both average and marginal costs can drop through the application of more efficient scheduling routines and procedures, and for many transportation providers it may be possible to provide additional trips through improved service coordination without the addition of any more resources, as described in TCRP Report 91 (Burkhardt et al., 2003). Indeed, improved coordination is a critical component of increasing the supply of available paratransit trips and the topic of much ongoing research. Finally, we also did not link our target population to specific transit-mode criteria arising from the Americans with Disabilities Act (ADA). This was because the available MEPS data were not sufficient to allow this linkage, especially given that the MEPS provides no geographically specific data. Final Report 40