Associated Costs Panel Report

(printer-friendly version - PDF, 295 kb)

Contents on this Page:

1. Summary of Panel Recommendations
2. Introduction
   1. Purpose of Panel
   2. Conceptual Approach
   3. Identification of Core Services
   4. Identification of Variables
   5. Completing the Templates
   6. Panel Recommendations
   7. Issues for Discussion
3. Subgroup Reports and Recommendations
   1. Geographic Variables Workgroup
   2. Clinical Variables Workgroup
   3. Demographic Variables Workgroup
4. Wage and Rent Indices for Titles I and II Grantees
5. History of the Panel

Views expressed are those of the panel participants and do not represent official positions of the Federal Government

I. Summary of Panel Reports

The Associated Cost panel recommended including three variables in the severity of need index at this point in time. The recommended variables include:

• Geographic wage index
• Geographic non-labor price index
• Substance abuse (IDU exposure category)

The geographic wage and the non-labor price indices are intended to compensate for regional variation in the cost of labor and facilities. The substance abuse risk category variable is intended to serve as a proxy for the incrementally higher costs of treating HIV infected individuals who are also substance abusers. The panel believed that the substance abuse exposure category is correlated with the prevalence of Hepatitis C among the HIV infected community and, thus, helps capture the costs of that comorbid condition as well. The panel agreed that these three variables are significant and independent determinants of per capita costs and can be appropriately measured with currently and publicly available data. (1)

Several other important variables were considered but were not recommended for inclusion in the severity of need index at this time. These variables were not recommended for inclusion on the basis that they were: (1) correlated with one the variables recommended for inclusion, (2) did not yet have a sufficient impact on per capita costs, or (3) could not yet be appropriately measured by the publicly available data. Variables considered but not recommended for inclusion at this point include:

• HIV disease stage*
• Health insurance premiums*
• Hepatitis C*
• Diabetes*
• Cardiovascular disease*
• Poverty*
• Race/ethnicity*
• Gender*
• Age*

The panel instead recommended that HRSA continue to monitor the cost associated with these variables. The panel also recommended that HRSA continue to address the limitations of available data surrounding some of the most important of these currently excluded variables (i.e., HIV disease stage) can be incorporate in the severity of need index in the future. Variables that the Associated Cost panel recommended for possible future inclusion in the severity of need index are indicated with an ‘*’ in the above list.

Of special note is the recommendation not to include HIV disease stage at this time. The panel agreed that disease progression was one of the most important determinants of per capita costs of HIV primary care. But currently available surveillance data from the CDC are not able to capture the significant improvement in CD4 count that may occur following the introduction of antiretroviral (ARV) therapies. Nor are the surveillance data able to measure the prevalence of CD4<50, which the panel members also thought was a strong predictor of costs. AIDS diagnosis and CD4 count at the time of initial diagnosis was considered by most panel members to be insufficient for measuring actual resource needs, particularly with the availability of ARVs. Furthermore, data from the HIVRN survey indicate that the incremental costs associated with disease progression (i.e., CD4<50) were attributable largely to an increase in the use of inpatient services, which not covered under the CARE Act. Most panel members also believed that ARV use, independent from CD4 count, may be a more important cost driver than actual disease progression. But data to measure state and local area variation in ARV rates do not exist. In sum, the majority of the panel members agreed that an estimate of the incidence of AIDS or CD4<50 was not a good proxy for the prevalence of these disease stage markers. It should be noted that several panel members disagreed with this decision and argued that CD4 count at the time of initial diagnosis (or AIDS diagnosis) based on CDC surveillance data is in fact an appropriate proxy for measuring HIV disease stage and should be included in the severity of need index.

^ back to top

II. Introduction

A. Purpose of Panel

The Associated Cost panel had two primary responsibilities. First, it was responsible for developing a set of geographic price indices for labor and non-labor inputs for the delivery of HIV primary care services funded under Titles I and II. Second, it was responsible for developing and assigning cost weights to a group of patient attributes considered to be important and independent determinants of the cost of care under Titles I and II.

To accomplish these goals, the panel members identified and conducted three sequential tasks. The three major tasks performed by the panel members were as follows:

• First, the panel identified a set of core services funded under Titles I and II. The purpose of this task was to identify a limited set of the most important services for which cost drivers could be determined. The criteria for selecting core services included: (1) having a demonstrated impact on care, (2) having a demonstrated impact on costs; (3) accounting for a significant share of total Title I and II funds, and (4) exhibiting geographic cost variability.
• Second, the panel identified an initial set of variables that were thought to be important determinants of per capita program costs associated with the delivery of core services under Titles I and II. The purpose of this task was to ensure that the panel considered a comprehensive set of cost drivers prior to assessing their validity, feasibility, and interdependence.
• Third, the panel completed a template for each variable, summarizing: (1) the rationale for including it in a severity of need index; (2) the sources of data for measuring the variable; (3) its level of aggregation, frequency of update, and availability for use as part of the index; and (4) its reliability, validity, and bias from measurement error. The purpose of this task was to evaluate the value of each variable and to develop a final set of recommendations for inclusion in the severity of need index.

^ back to top

B. Conceptual Approach

The panel decided early in the process that its responsibility was not to derive a standardized national per capita cost amount that could then be scaled upward or downward by a set of indices representing local cost variability. The panel agreed that any average standardized amount should ultimately be determined by the level of appropriated funds for Titles I and II per client served. The panel instead focused its attention on measuring the incremental costs associated with selected inputs (such as the price of labor and facilities) and patient characteristics (such as disease progression and comorbidities). For example, the panel agreed that grantees operating in high wage markets should receive more per capita funding than grantees in low wage markets. Similarly, the panel believed that Title I and II grantees serving a disproportionate share of patients with comorbid conditions or on antiretroviral therapies should receive more per capita funding than those serving less sick patients. The incremental costs associated with inputs would be measured and applied at the state or EMA level based on regional wage and rent data. The incremental costs of patient characteristics would be measured nationally and applied at the state or EMA level, weighted by regional prevalence rates.

The panel also believed that it should focus on “incurred” costs rather than building in policy-driven incentives for target populations, health professional shortage areas or other capacity constraints, quality of care, new treatment protocols, or optimal staffing patterns. The panel decided that rewarding grantees for quality (by allocating more funds on a per capita basis to high quality providers than to lower quality providers) or using the severity of need index to promote certain standards of care (by allocating more funds to grantees meeting recommended treatment protocols) went beyond the purpose of the severity of need index. The panel further agreed that measuring quality and incorporating it into a severity of need index would be an extremely difficult task to accomplish. However, it was commonly accepted among panel members that any needs-based funding allocation system should not penalize high quality care. For example, a grantee that succeeds in lowering the rate of AIDS progression or the incidence of comorbid conditions (and thereby avoiding the higher costs associated with AIDS or comorbid diseases) should not be penalized by receiving fewer CARE Act dollars on a per capita basis.

Finally, the panel agreed that, if the incremental costs of a given variable affected only a subset of the core services discussed above, then those incremental costs should be weighted by the share of total Title I and II funding allocated to that core service. For example, if disease progression were associated only with the introduction of ARV therapies, then the incremental cost of an AIDS diagnosis would be weighted by the share of total funding allocated to prescription medications. Weighting incremental costs by the share of funding allocated to affected services ensures that the impact of that variable on total costs is accurately captured. The panel members agreed that the Associated Cost panel was responsible for deriving both the incremental costs and the service weights. But it was assumed that the prevalence rates for the patient attributes would be provided by the Patient Characteristics panel.

^ back to top

C. Identification of Core Services

As stated earlier, the first step for the Associated Cost panel was to identify a set of core services. The panel members acknowledged that, to identify costs, first it was essential to agree on the set of services that should be included. Further, the panel members believed that understanding the services that account for the majority of Title I and II funds would be useful for identifying the kinds of factors that ought to go into a severity of need index. For example, regional variation in the prevalence of injection drug users (IDUs) is germane in part because payment for substance abuse services constitutes a relatively important proportion of Title I and II funds. Further, including substance abuse services in the core set of services highlights the importance of factoring in regional differences in the wages of substance abuse counselors.

The four criteria used for selecting the core services are listed below. Services that satisfied several (though not necessarily all) of these criteria were considered for inclusion in the core set.

• Share of Title I and II Funding. Core services had to be covered under Titles I or II. Largely non-covered services (e.g., inpatient care) were not included.
• Impact on Costs. Core services had to have a significant impact on total costs, as measured by their share of total allocations. Services that account for a very small share of total CARE Act spending (e.g., outpatient rehabilitation services) were not included.
• Impact on Care. Core services had to have a significant impact on the quality of care and health outcomes for the average client.
• Variability in Costs. Core services had to exhibit significant variation in unit costs across grantees. A service that may not have a significant impact on costs or quality may have average costs that vary tremendously between regions. Panelists therefore decided that it should be included, even though it is not one of the major drivers of costs.

The recommended list of core services, along with their proportions of FY2004 Title I funding and their ranking by criteria, is presented in Table 1. The list includes six medical services (ambulatory/outpatient medical care, specialty care, pharmaceuticals, substance abuse services, mental health services, and oral health care).(2) In addition, the list includes three social support services (housing assistance and services, transportation services, and food assistance). Finally, core services include case management services, which can be either medical or psychosocial case management. The medical services combined account for nearly half (48%) of total Title I and II funding. The social support services account for an additional 13% of total funding, and case management represents slightly over 10% of total funding.

Table 1: List of core services included in assessment of costs

NOTES:
* Title I allocations based on FY2004 expenditures as reported by HAB.
KEY: 1=high importance, 2=moderate importance, 3=low importance

Not surprisingly, ambulatory/outpatient primary medical care constitutes nearly one-quarter (24%) of total Title I and II funding. The panelists agreed that primary medical care has an important impact on both cost and quality and exhibits significant variation in per capita spending across grantees. Drug assistance (over and above ADAP expenditures) also accounts for a significant share of total Title I and II allocations (10%) and has an important impact on cost and quality. But panel members agreed that, because grantees face a national drug market and have access to the national 340b drug pricing program, unit drug costs should not vary substantially by state or EMA. In slight contrast, oral health accounts for only 3% of total spending and has a moderate impact on cost and quality. The unit cost of oral health was also assumed by the panel members to exhibit relatively less variation than medical care. Nonetheless, the panel members concurred that these 10 core services represent the majority of Title I and II spending, are critical components of the HIV primary care delivery system, and exhibit sufficient variation in unit costs to warrant inclusion in the severity of need index.

The panel considered several other services important to the HIV primary care delivery system but decided that these services either did not have a major impact on costs or did not demonstrate sufficient variation between grantees to warrant inclusion. Two important examples include outreach, retention, and adherence programs, and testing and counseling programs. Both of these services are critical components of the primary health care system for people living with HIV and AIDS. But first, at this point they constitute a relatively small proportion of total Title I and II funding, and second, the cost of these services does not vary significantly across regions. Variations in labor costs for these services are likely to be correlated with the variation in wages for other, more fundamental labor categories such as physicians and nurses. However, the panel members agreed that, because the core set of services may change over time, HRSA should periodically re-evaluate all services covered under the CARE Act using these criteria. The panel also agreed to exclude the cost of administrative activities and those associated with data collection and reporting requirements from HRSA.

^ back to top

D. Identification of Variables

The second step was to identify the important determinants of cost variation for these core services. The panel identified three types of variables for consideration. The three broad categories identified and considered by the panel, and the individual variables within each category, were:

• Variables based on geographic cost differences
  • Labor
  • Non-labor inputs such as rent and facilities
  • Health insurance premium
• Variables based on patient clinical characteristics
  • HIV disease stage
  • Hepatitis C
  • Substance abuse (IDU exposure category)
  • Diabetes
  • Cardiovascular disease
• Variables based on patient sociodemographic characteristics
  • Race/ethnicity
  • Gender
  • Age
  • Poverty

The first set of variables captures the impact of geographic variation in wage rates, non-labor input prices, and health insurance premiums on pre capita grantee costs. Labor accounts for the major share of total grantee expenditures and represents an important external source of variation in per capita spending that, the panel agreed, should be included in the severity of need index. The panel also agreed that building and facility costs vary by region and, thus, should also be considered in the severity of need index. Similarly, several states use a proportion of their ADAP funds to cover the cost of health insurance with prescription medication coverage. Since the cost of health insurance also varies by region and state based on non-actuarial factors (which would be captured under the patient case mix variables), the panel agreed that health insurance premiums represent another important external source of per capita spending that should be considered for inclusion in the severity of need index. The ADAP Health Insurance Purchasing and Maintenance Program (HIP) is also a cost-saving initiative. Per capita ADAP expenditures in states with a HIP program are likely to be lower than in states without a HIP program, a cost differential that panelists believed should be considered in the severity of need index.

The second and third sets of variables measure the impact of patient characteristics on per capita expenditures among Title I and II grantees. The panel agreed that HIV disease progression and Hepatitis C are highly correlated with per capita expenditures: clients with low CD4 counts (generally less than 50) or a diagnosis of Hepatitis C have higher costs of care than those without such conditions. However, since CD4 counts can change over time after initial testing in ways that are not captured in the current data, the panel concurred that the prevalence of clients on antiretroviral (ARV) therapies was a more accurate indicator of costs. The IDU exposure category is an indirect measure of substance abuse, which was also considered an important and positive correlate with per capita spending. The panel members agreed that diabetes and cardiovascular disease were also important correlates with HIV infection, both as sequelae of HIV infection as well as age-related comorbidities among an aging HIV population. Finally, the panel agreed it was important to consider such sociodemographic characteristics as age, gender, race/ethnicity, and poverty. Certain age, sex, and race/ethnicity subgroups may have higher or lower resource needs than others, based in part on their access to health care services and their underlying health care needs. Poverty was considered an indicator of both resource needs and access to care. The panelists agreed that low-income people are less likely to be insured and, thus, less likely to seek and remain in ongoing care. While the short-term costs of low-income and disadvantaged populations may be lower, delays in seeking care initially and lack of continuity and adherence once in care lead to higher per capita expenditures over time. The inclusion of the poverty variable is based on the need of the poor who have HIV for CARE Act support once they seek care.

^ back to top

E. Completing the Templates

After identifying the variables to be considered, the Associated Cost panel then divided itself into three workgroups to discuss and evaluate the variables in each of the three categories in greater depth. Each workgroup was responsible for completing a template for each variable in its category. The purpose of the template was to define the variable; identify the rationale for its inclusion; identify the potential sources of data for measuring the variable; assess the validity, reliability and potential bias of each variable; and suggest ways to address any underlying bias. Based on the results of the small group discussions, each workgroup then forwarded to the larger panel a list of recommended variables to be included in the severity of need index. The full members of the Associated Cost panel then discussed the small group recommendations and identified the variables to be included in the panel’s final recommendation to the larger severity of need expert panel.

A description of the variables, the potential sources of data, and the final recommendation from the Associated Cost panel are summarized in Table 2. Most variables require data on both the incremental costs and the prevalence of the variable. Incremental costs are usually measured at the national level and prevalence rates are measured at the EMA or state level. (A more complete discussion of the variables, sources of data, and panel recommendations are presented in Section II.)

Table 2: Description of variables, sources of data, and panel recommendations

The main sources of data for the cost variables are the HIV Research Network and Medicaid claims data. The HIV Research Network (HIVRN) survey collects annual clinical and health resource utilization data for about 15,000 patients a year across 17 non-representative sites across the country; data have been collected for some 38,000 unique patients over a 5-year period. Data elements are individual-level and include inpatient and outpatient utilization, prescribed medications, some substance abuse and mental health visits, and other information. The Network also conducted a client interview (sample size = 951) to collect information on additional services and services received at other sites. Costs were assigned to the reported utilization rates for each service category based on HCSUS estimates trended forward. The survey was conducted with a stratified random sample of those in care at 14 institutions receiving IRB approval. The sample included those with public, private, and no insurance.

The Medicaid claims are obtained from the Medicaid Analytic eXtract (MAX) files. The MAX files are a standardized Medicaid claims database containing fee-for-service claims from all 50 states and the District of Columbia and are available from the Centers for Medicare & Medicaid Services (CMS). The MAX database includes an enrollment file with demographic characteristics and dates of enrollment and a set of claims for inpatient, outpatient, prescription drug, and long-term care services. The MAX suffers from several weaknesses. First, there is a serious lag between the date of service and the availability of the MAX database. The most currently available MAX file is for services rendered in CY2001. Second, the MAX database contains only claims paid under Medicaid fee-for-service; encounter data for enrollees covered under Medicaid managed care plans are not included. Third, the Medicaid population may not be representative of the CARE Act population. Fourth, the MAX database contains only claims for Medicaid-covered services; the costs of services not covered under Medicaid are not included. Other sources of data include the Bureau of Labor Statistics’ Occupational Employment Statistical Survey (OESS), the Medicare Geographic Practice Cost Index (GPCI), and the Department of Housing and Urban Development’s Fair Market Rent (FMR).

^ back to top

F. Panel Recommendations

After completion of the variable templates and a thorough discussion by the full panel members, the Associated Cost panel agreed to recommend only three variables for inclusion in the severity of need index at this point. The variables recommended for inclusion are:

• Geographic wage index
• Geographic non-labor price index
• Substance abuse (IDU risk factor)

These three variables were considered to be major and independent determinants of per capita costs. The panel also agreed that the existing data for measuring these variables also supported their inclusion. Several variables, such as disease progression and health insurance premiums, were also considered very important and independent determinants of severity of need, but the majority of the panel members argued that the data necessary for capturing their true impact on costs do not yet exist. (See Section II for a more complete discussion of the rationale for not including selected variables in the initial severity of need index.) Hepatitis C was considered important, but the panel members felt it was closely correlated with variables already included in the model (especially the substance abuse exposure category) and, thus, should not be included as a separate indicator. In contrast, age-related comorbidities were considered important and independent but do not yet have a sufficient impact on costs (and also lack appropriate data to measure) to warrant inclusion. Thus, they should be considered in the future as the infected population continues to age and the prevalence of these diseases increases, and better data for measuring the impact of age and age-related comorbidities become available. The panel also agreed that the inability to differentiate the impact of age-related comorbidities on HIV-related service use further complicates the issue. The panelists agreed that CARE Act funds should only be used to cover the cost of comorbidities as they affect HIV-related care. Panel members also agreed that impact of gender on the cost of HIV care is difficult to ascertain with currently available data. While women have higher costs of medical care than men in general (because of ob/gyn-related care and a greater propensity to seek care), studies have not yet been able to demonstrate that they have higher HIV-related costs. Finally, the panelists agreed that poverty was a better measure of access and service use than race/ethnicity, but patient-level data on income do not yet exist.(3)

The panel recommended that HAB continue to monitor the impact of HIV disease progression, age and age-related comorbidities, poverty and gender on the resource needs and continue to address the data limitations that hinder their inclusion in the severity of need index at this time.

A summary of the significance of each variable for capturing the variation in local resource needs, as well as the availability of appropriate data that can be used to measure each indicator at this time are presented in Table 3. The availability of data includes both cost estimates and prevalence estimates. The panel agreed that all three geographic variables had a moderately significant impact on per capita cost variation among Title I and II providers, but only labor and non-labor input prices had data that supported their inclusion at this time. The non-age-related clinical variables were considered to have a high significance on per capita cost variation, but the panel believed that only the substance abuse variable could be measured in an appropriate manner to justify its inclusion at this time. HIV disease stage was considered extremely important to severity of need, but neither the prevalence of CD4<50 or of ARV use could yet be measured in a way to warrant inclusion. The age-related comorbid conditions were considered to have a low impact on cost variation at this point in time, but could be measured with currently available data with some degree of accuracy with Medicaid or Medicare claims once the prevalence of these diseases increases among the infected population. The panel agreed that age and gender could not be measured with current data. While age is likely to have a significant impact on HIV costs in the future, the impact of gender on HIV-related costs is difficult to ascertain with currently available data. Neither age nor gender should be included until better data become available and a demonstrated and generalizable impact on HIV-related costs of care can be measured. Race/ethnicity had a potentially more significant impact on costs, but the reasons for including the variable in a severity of need index (i.e., inferior access to care and the additional costs associated with recruitment and retention of hard-to-reach populations) were more accurately captured through the poverty variable. Unfortunately, state and local prevalence estimates of poverty among those with HIV infection are not available.

Table 3: Significance and Availability of Data for Measuring Variables

Table 4 presents the panel’s primary reasons for recommending the inclusion or exclusion of each variable at this point in time. Those variables characterized by a good rationale and supported by adequate data have been recommended for inclusion. Those variables characterized by a good rationale but not supported by adequate data have been recommended for exclusion at this point in time only, but are accompanied by a recommendation that HAB continue to monitor their impact on costs and to address the data limitations that preclude their current inclusion. Those variables that are characterized by a weak rationale are recommended for exclusion from the severity of need index more generally. As the table makes clear, there are many variables that the Associated Cost panel viewed as important determinants of per capita costs, but could not recommend for inclusion at this point in time solely because of the inadequacy of the available data.

Table 4: Variables considered and forwarded for possible inclusion in the SON index by the Associated Costs Panel

^ back to top

G. Issues for Discussion

As stated above, the panel recommended inclusion of only three cost-related variables at this time: two for capturing geographic differences in input prices and one for the incremental costs associated with the substance abusing population. The panel agreed that several other variables were important determinants of per capita costs but decided that additional information was required before they could be measured in an accurate and meaningful way for the severity of need index. Some of the limitations of the variables not included in the panel’s recommendations, as well as several remaining issues that need to be considered in the development of a severity of need index, are discussed below.

HIV disease progression. The panel agreed that HIV disease progression is one of the most important determinants of per capita cost variation. However, the panel acknowledged that both AIDS diagnosis and CD4 counts, with treatment, could improve over time such that even AIDS patients could have a CD4 count above 200. In fact, the state and local prevalence estimates of a CD4 count of less than 50 (found to be a highly significant predictor of per capita costs in the HIVRN survey) are currently not available. The panel also noted that the HIVRN survey data show that most of the incremental costs associated with a low CD4 count are incurred in the inpatient setting and the CARE Act does not cover the cost of inpatient care. The panel concurred that the true cost-driver associated with disease progression was not AIDS diagnosis or CD4 count, but rather whether an individual was receiving ARV therapies. Clients on ARV drugs, independent of their diagnosis or CD4 counts, are more expensive to care for than those not on ARV medications. Further, the incremental costs associated with ARV use are likely be incurred on an outpatient basis and thus eligible for CARE Act funding.

The panel then determined that there is no consistent and current information on ARV prevalence at the state or local level. The rate of ARV use from the annual CARE Act Data Reports (CADRs) is duplicated and inconsistently reported among CARE Act providers. One panel member pointed out that ADAP reports submitted annually by each Title II grantee may be useful for estimating ARV rates at the state level. Others argued that CDC surveillance data on AIDS as a percent of new HIV cases may be a sufficient indicator of the relative burden of people who are in need of ARVs, especially given current standards of when to start prescription drug therapy. However, after much discussion, the panel decided that indicators of HIV disease progression should not be included in the severity of need index until consistent and unduplicated state-level prevalence estimates of either CD4<50 or ARV use are available.

A minority of panelists disagreed with this decision and advocated using AIDS incidence as reported by the CDC as a proxy for HIV disease progression. However, one concern with this recommendation is that the CDC surveillance data are not yet mature enough across all states to identify unduplicated cases of newly diagnosed AIDS. Using surveillance data to measure AIDS incidence is likely to result in an undercounting of true cases in states with an immature data reporting system and, thus, an under funding of those Title I and II grantees.

Age-related comorbidities. The panel agreed that age-related comorbidities of HIV infection will become an increasingly important component of the total cost of care as the infected population continues to age. Particularly important are cardiovascular disease (CVD) and diabetes, both of which can become complicated to treat because of the HIV virus. Nonetheless, the panel decided that the incremental costs of age-related comorbidities were not yet sufficiently large to warrant their inclusion in the severity of need index. The panel recommended that HRSA continue to monitor the cost of these comorbidities and consider including them in the severity of need index in the future.

Demographic characteristics. The panel also decided that while health care costs are likely to be positively correlated with age and being female, the incremental costs of HIV care associated with these demographic subgroups may not be sufficiently large to warrant inclusion in the severity of need index. This conclusion was supported by the results of the HIVRN survey, which showed no statistically significant association between costs and age or gender. Again, the panel agreed that HRSA should continue monitoring the impact of age and gender on HIV care costs and consider including them in the future if the evidence suggests they are important and independent determinants. The panel expressed less consensus on race/ethnicity. Several members argued that racial and ethnic minorities have higher long-term costs of care because they are more likely to enter care later in their disease progression, are less likely to remain in care on a consistent basis once they initially seek care, and are less likely to remain adherent to treatment protocols. As a result of inadequacy of care, they have poorer health status and greater health care needs in the long run. Providers also incur additional costs identifying HIV infected people and conducting outreach, retention and adherence programs among communities of color. However, the panel agreed that the underlying rationale for including race/ethnicity in the severity of need index is better captured by poverty.

Socioeconomc characteristics. Again, the panel identified two reasons for considering poverty. The first reason had to do with measuring the burden on care by low income subpopulations eligible for CARE Act services. Some regions serve a disproportionately lower income subpopulation than others and should be appropriately compensated for this additional burden. However, the group agreed that this reason is unrelated to costs per case and, thus, not appropriate for the Associated Cost Panel. Rather, the panelists agreed that issues related to burden associated with poverty are better handled by the Area Characteristics Panel. The second reason for including poverty is the extent to which low income people are inherently more costly to serve than those in higher income groups. Costs may be higher for low income populations for several reasons, including the costs of program activities related to identification, outreach and retention in care, as well as the higher treatment costs associated with delays in seeking care and lack of continuity in care or adherence to treatment. The panel agreed that these incremental costs are probably not sufficient (relative to overall treatment costs) to warrant inclusion in the SON index at this time. The panel also agreed that it would be difficult to obtain data on the costs and incidence related to timing of initial treatment and lack of continuity and adherence. Nor are there good estimates of the incremental costs of HIV care for people with incomes below the federal poverty level. Thus, the Associated Cost panel decided to defer a decision on poverty until better data are available.

Drug prices. In addition to labor and non-labor inputs, the panel also considered geographic variation in drug prices. The panel acknowledged that prescription medications account for a major and growing share of total CARE Act spending, and thus, even minor variations in drug prices could have an important impact on total spending. The panel members also reported that CARE Act grantees differ in their ability to obtain negotiated discounts from drug companies. However, after further consideration, the panel agreed that regardless of their ability to negotiate drug price discounts, all states have access to the same level of discounts under Section 340B of the Public Health Service Act of 1992. The 340B drug pricing discount program requires pharmaceutical manufacturers participating in the Medicaid program to provide front-end discounts on covered outpatient drugs purchased by ‘covered entities.’ State-operated AIDS Drug Assistance Programs and the Ryan White CARE Act Title I, Title II, and Title III programs qualified as covered entities. Because of states’ eligibility to participate in the 340B drug pricing program, the panel decided that grantees should not be compensated for any observed differences in drug prices.

ADAP health insurance purchasing and maintenance program. One panel member noted that under the CARE Act Health Insurance Purchasing and Maintenance Program (HIP), 26 states use a portion of their ADAP funds to underwrite the cost of health insurance for eligible enrollees that includes prescription medications. The panel believed that the per-enrollee cost of the HIP program could vary for reasons other than client case mix, such as health insurance regulations, market competition, and coverage policies. For this reason, the panelists agreed that the geographic variation in the per-enrollee costs of the program (or the ‘premium’) is a legitimate factor to consider in the severity of need index. The panel also agreed that, because the HIP program is intended to be cost-saving, per enrollee costs under ADAP are likely to be lower in states with a health insurance purchasing/maintenance initiative than in states that rely solely on the direct purchase of drugs. However, several problems with this variable were noted. First, some programs rely on state-only funds to supplement the expenditures, leading to inconsistencies in data reporting. Second, enrollment and expenditure reports exhibited a large and unexplained variation in per capita HIP costs. Third, and most importantly, panelists were concerned that states operating HIP programs not be penalized by receiving fewer Title I and II funds. As a result, the panel agreed that adjustments for per-enrollee HIP expenditures should be deferred until more states implement the program and more consistent data are available.

Medicaid generosity. The panel spent some time discussing Medicaid generosity and ways to ensure that states with more generous Medicaid benefit and coverage policies would not be penalized by receiving fewer CARE Act dollars. The panelists were in agreement that the severity of need index should not create a disincentive to expand Medicaid eligibility and enhance covered services. It was also recognized that the way in which Medicaid coverage was incorporated into the model would affect how certain cost-related variables, such as poverty, were measured. The panel agreed to defer this issue until the recommendations of the Patient Coverage panel were available.

CARE Act Data Reports (CADR) data. The panel recognized that many of the variables considered for inclusion in the severity of need index were reported on the CADRs. All CARE Act grant recipients are required to submit a CADR annually to HRSA. The CADR contains provider-level summary data on the number of unduplicated clients served by demographic and clinical characteristics, and the number of services used by type of service. For instance, the CADR contains counts of unduplicated clients by age categories, gender, race/ethnicity, income, insurance status, living arrangements, and risk factor. The CADR also includes counts of unduplicated HIV-infected clients newly in care, the number with an AIDS diagnosis, the number on ARVs, and the number receiving certain procedures and screenings. While the CADR provides a potentially important source of data at the local level that could be used to measure severity of needs, inconsistencies in reporting practices among grantees make this database unusable in its current form. In addition, the panelists further pointed out that the CADRs often exclude the cost of contracted services or services not paid for by CARE Act dollars and thus in general underreport the actual costs of care. Finally, individuals receiving care at multiple CARE Act providers are counted separately in each CADR.

Budget neutrality. As stated earlier, the panel agreed that, rather than deriving a national annualized cost for HIV care, the base rate should be determined by the appropriated funding amount divided by the number of qualifying individuals who need care. Adjustments would then be made to this per capita allocated funding amount for such factors as input prices and client casemix. For example, grantees treating a disproportionate share of clients with substance abuse problems would receive more on average than those serving patient populations with fewer comorbidities and/or greater adherence with both medical appointments and drug regimens. Similarly, grantees in areas with more people below the federal poverty level would receive more per capita funding than those in areas where more people have insurance and resources. However, the per capita allocation adjustments would have to be budget neutral (i.e., total spending under the program could not exceed the total Title I and II budget allocations). Therefore, the panelists noted that per capita allocations to grantees with a severity of need index above the national average would be higher than under the previous case-based system, while those with a severity of need index below the national average would receive less funding than before. Similarly, providers in high wage markets would presumably see their per capita funding increase, while those in low wage markets would experience a decline in per capita funding compared to what they received under the case-based system.

^ back to top

III. Subgroup Reports

A. Geographic Variables

1. Overview of Key Issues

The first set of variables was intended