Area Characteristics Panel Report

(printer-friendly version - PDF, 282 kb)

Contents on this Page:

1. Introduction
2. Discussion of Variables
   1. Burden of Disease
   2. Health Infrastructure
   3. Poverty and Variables from the Census
3. History of the Panel
   1. Members and Affiliations
   2. HSR/RTI Contact Information

I. Introduction

The Area Characteristics Panel was charged with identifying aggregate characteristics of the State or eligible metropolitan area (EMA) that could be predictive of variations in resource needs for Ryan White Comprehensive AIDS Resources Emergency (CARE) Act HIV/AIDS services. To accomplish this goal, the panel split into three working groups, based loosely on available data sources. The first, the Burden of Disease Group, evaluated ways to measure the number of HIV/AIDS cases in an area and their level of severity using primarily Centers for Disease Control and Prevention (CDC) surveillance data. The second, the Health Infrastructure Group, looked at ways to measure access to health care services using the Area Resource File (ARF) and Health Resources and Services Administration (HRSA) internal data. The third, the Poverty and Census Group, evaluated poverty and aggregate measures of the economic health of an area using variables drawn primarily from the U.S. Census and the Bureau of Labor Statistics.

In general, the Area Characteristics Panel recommended variables that would help enumerate the number of HIV/AIDS cases in an area and then adjust this count based on measures of access, poverty, and insurance. Variables (Table 1) were evaluated based on their importance in determining resource needs for CARE Act services and the current quality, cost, and availability of data used to measure them. Poor access, high poverty, and low rates of insurance may lead to greater need for CARE Act resources to provide services to the needy and to undercounts of HIV/AIDS cases. The Area Characteristics Panel evaluated 20 variables, of which they forwarded 5 for possible inclusion in a severity of need (SON) index.

Table 1. Variables considered for possible inclusion in an HIV/AIDS SON index, by area characteristics working group

To prioritize variables, panelists first met as a full group (13 panelists and 2 contractors) to develop a list of variables to evaluate subjectively in terms of each variable’s contribution to SON. The group eliminated 12 of 20 variables that were deemed impossible to accurately measure or not related to SON. Panelists were asked to score each remaining variable from 1 to 5, with 1 indicating a variable of the highest importance and 5 indicating a variable of the lowest importance, based on how well each variable measures the theoretical concept of SON. Panelists were also asked to consider that variables may covary or measure the same concept and were asked to prioritize similar variables as opposed to giving them all the same score. These scores were then compiled and the averages ranked (Table 2).

Following the panel’s discussions, the list of variables was discussed by a mixed panel of experts, some from this group and some from groups which had discussed other topics, at a two day meeting held in Washington, DC. The mixed group reviewed the list of variable and made recommendations to remove three variables; chlamydia prevalence, unemployment rate, and HRSA-supported clinics.

Table 2 Area characteristics variables forwarded to the full panel and panelists’ priority score

* Removed during the mixed group session.
** Removed during final meeting sessions of the panel.

Panelists in this group generally agreed on the majority of issues they faced. Panelists’ votes on these items were remarkably consistent from voter to voter. For example, all but one panelist gave HIV/AIDS disease prevalence a score of 1, all but three panelists gave the poverty rate a score of 1, and only one panelist gave STI burden a score lower than 3. Qualitatively, all panelists agreed that the poverty rate, uninsured rate, and unemployment rate measured similar concepts and should be considered together, although there were some minor differences in whether panelists thought the uninsured rate or the poverty rate was more important to consider.

The panel had some areas of disagreement, which were each resolved before forwarding recommendations to the larger group. First, in the name of parsimony, a subset of panelists believed that the group should forward the smallest possible number of variables to the larger group and suggested the group forward only HIV/AIDS disease prevalence and the poverty rate to the larger group. The larger group disagreed, and the panel’s consensus was to forward the eight variables in Table 2. Second, the group disagreed on whether to forward a variable measuring the level of personal income (described below) to the full committee. Arguments in favor of this variable suggested that it was a valuable measure of resources that could potentially be diverted to HIV care. Arguments against this variable suggested that, although the variable did accurately measure income in an area, the total wealth of an area was not descriptive of wealth that had already been or would potentially be allocated to HIV care. After discussing the issue, the consensus of the group was not to forward the variable.

Finally, one panelist was concerned that the group was paying insufficient attention to measuring aggregate need for substance abuse and mental health services and the burden of sexually transmitted diseases (STDs) and other comorbidities. This panelist also believed that aggregate measures of substance abuse, mental health service use among all people in an area, and rates of STDs in an area were relevant to the need for HIV/AIDS resources. The other members of the panel agreed these issues were important but that (1) individual characteristics, such as STDs and substance abuse, would be covered by another panel and (2) many of these factors were “colinear” with other measures of indigence and lack of care services (e.g., ADAP adequacy, 100 percent poverty level).

Panelists were in strong agreement that the need for substance abuse and mental health services among CARE Act clients indeed would lead to increased resource needs. However, panelists believed the need for substance abuse services was in part measured by the intravenous drug use exposure category forwarded by another panel. Other reliable sources of information to measure these needs among HIV-infected patients were not identified. High rates of STD rates may be predictive of future high rates of HIV and AIDS, and for this reason, these rates may measure the need for future services. However, the relationship between STD infections and new cases of HIV is largely unquantified and likely differs regionally. Furthermore, the primary responsibility of the CARE Act is to provide medical services for those currently diagnosed with HIV and AIDS. Therefore, the panel ultimately thought that aggregate measures of need for services among the entire population (including those without HIV) would not help understand the need for these services among HIV-infected patients.

This report outlines the rationale for recommending each of the variables above and describes variables that were not forwarded for consideration and the reasons these variables were excluded. The format of the report reflects the work of three workgroups:

• Burden of Disease Workgroup, which considered ways to measure HIV and AIDS cases at the community level
• Health Infrastructure Workgroup, which considered measures of an area’s capacity to offer access to care
• Poverty and Census Workgroup, which considered measures from the census to measure the underlying poverty in an area.

Each section is divided into two subsections, the first discussing variables that were accepted by the entire group as potential elements for an SON index and the second discussing variables that were not. Each section briefly describes each variable considered and then presents a completed template that guided the evaluation of all variables.

^ back to top

II. Discussion of Variables

^ back to top

A. Burden of Disease

Variables considered: prevalence of HIV/AIDS disease, prevalence of sexually transmitted infections, AIDS-specific mortality, and relative survival.

The Burden of Disease Group considered variables that would measure the degree of HIV and AIDS in an area. Current CARE Act allocation algorithms use the 10-year weighted AIDS case count to define the level of disease burden in an area. This group decided to use the cumulative count of living HIV and AIDS cases to measure burden. The workgroup also suggested using reported rates of chlamydia as a possible adjustment to HIV disease rates because higher rates of chlamydia infections may be indicative of a higher degree of incident and potentially unreported HIV cases (Pinkerton et al., 2003).

Variables related to AIDS mortality were not forwarded primarily based on the inadequacy of the data used to measure it. First, deaths are not reported to the CDC with consistent timeliness from all jurisdictions, and second, deaths among patients with HIV/AIDS reported to the CDC may reflect death from any cause. The panel felt that, without adjusting for reporting delays in death rates and the relative survival patterns across jurisdictions, death data would be meaningless at best and potentially misleading at worst.

1. Variables forwarded for consideration

Prevalence of HIV/AIDS disease: Panelists recommend using the enumerated number of living HIV and AIDS cases per jurisdiction, as reported to CDC surveillance, to measure the SON for CARE Act services. Specifically, the panel recommends using the number of documented living HIV and AIDS cases reported by States using name-based reporting systems in the most recent calendar year.

Current CARE Act allocations are based on the number of estimated living AIDS cases in an area over the past 10 years. The current system excludes HIV cases altogether. In addition, the formula for estimating living AIDS cases includes counts of people who are deceased when a jurisdiction’s actual death rate is higher than the national average and excludes individuals who are still alive when a jurisdiction’s actual death rate is lower than the national average. The majority of panelists felt that moving from the current system to a system that allocates funds based on living reported HIV and AIDS cases would represent a vast improvement from the status quo.

Further, no previous CARE Act allocations have incorporated CDC HIV information. Currently, HIV disease data are available for only 38 areas and 13 additional areas have a non-name-based reporting system (from which the CDC does not accept data). In addition, variation in completeness of HIV reporting exists across jurisdictions based on the maturity of their name-based surveillance system. Basing allocations on all HIV disease data (inclusive of all HIV and AIDS cases) would reward jurisdictions with the most mature name-based surveillance systems but would not address real differences in underlying need for CARE Act services. However, excluding HIV cases would ignore a substantial element of variation in need between areas altogether.
Unlike other variables considered by the panel, reported HIV was the only variable which certain jurisdictions lacked data by choice. Many States chose to begin reporting HIV data by name following a directive to do so in the previous round of CARE Act legislation. A few areas chose not to report this data. Panelists did not feel that areas that reported HIV cases could be fairly denied funding for those cases simply because other jurisdictions had chosen to not implement similar systems.

The majority of panelists do not recommend adjusting reported AIDS cases to account for additional currently undocumented or unreported cases of HIV in States without mature name-based systems. However, in the event such an adjustment becomes a political necessity, the panel would strongly advise policy makers to convene a scientific panel to investigate the most fair means to make such adjustments.

2. Variables not forwarded for consideration

Prevalence of sexually transmitted infections: The prevalence of STD infections has been requested from grantees by the CARE Act in the past to assess an area’s relative SON. A high level of STDs may indicate a high degree of sexual risk activity that would be predictive of incident HIV infections, although the precise quantitative link between STDs and HIV is not known.
According to the CDC,

“Individuals who are infected with STDs are at least two to five times more likely than uninfected individuals to acquire HIV if they are exposed to the virus through sexual contact. In addition, if an HIV-infected individual is also infected with another STD, that person is more likely to transmit HIV through sexual contact than other HIV-infected persons. There is substantial biological evidence demonstrating that the presence of other STDs increases the likelihood of both transmitting and acquiring HIV. STDs probably increase susceptibility to HIV infection through two mechanisms: genital ulcers (e.g., syphilis, herpes, or chancroid); and non-ulcerative STDs, such as chlamydia, gonorrhea, and trichomoniasis, which increase the concentration of cells in genital secretions that can serve as targets for HIV. In addition, studies have shown that when HIV-infected individuals are also infected with other STDs, their infectiousness is increased. For example, men with both gonorrhea and HIV are more than twice as likely to shed HIV in their genital secretions than those who are infected only with HIV” (http://www.cdc.gov/std/hiv/STDFact-STD&HIV.htm).

Still, the panel questioned whether data on prevalent STD infections were valuable as an indicator of HIV disease that would require CARE Act assistance in light of the fact that the CDC provides direct estimates of the number of prevalent HIV and AIDS cases. Prevention of incident infections was thought to be an important issue to address but one that was ultimately not the central mission of the CARE Act. Of all prevalent STDs, the prevalence of chlamydia was thought by the panel to be the most highly related to HIV disease. This is supported by some evidence (Pinkerton et al., 2003). State-level estimates of chlamydia prevalence are available freely from the CDC, whereas county-specific estimates require a special request from the CDC. The panel suggested using chlamydia rates at the State level as an additional, potentially useful indicator of undiagnosed HIV disease. The panel voted to forward this variable for consideration for use in an SON index but suggested that its weight or value in such an index should be low, if in fact it was included at all.

At the final meeting in Washington, DC, the mixed-group panel questioned the purpose of chlamydia prevalence, and argued in favor of its removal.

AIDS-specific mortality: Mortality resulting from AIDS was considered by the panel as a possible indicator of poor quality of medical care. The panel was concerned that no such estimate of deaths specifically caused by HIV/AIDS exists, only estimates of total deaths from all causes among patients with HIV and AIDS. Aggregate mortality data are fairly good, but patients with HIV disease are at an elevated risk of death from a number of causes, including substance abuse, violence, and accidents. Recently, renal failure and hepatic diseases have become major causes of death among patients with HIV disease. Cause of death information listed on patient death certificates is also not useful because AIDS often may not be listed as a cause of death because of the stigma that is associated with the behaviors that cause AIDS. The degree to which this occurs likely varies across jurisdictions. Without adjusting for these sources of error, the panel felt that the aggregate number of deaths among patients with HIV disease would not be a valid indicator of deaths resulting from HIV or AIDS. The panel also thought that for the purposes of an SON adjustment, AIDS-specific mortality described a variable outside of the scope of the Area Characteristics Panel. The panel forwarded both this variable and a possible adjustment to it (relative survival) to the Patient Coverage Panel for consideration.

During the mixed-group panel meeting in Washington, DC, there was some discussion that including a death rate measure may create disincentives for offering quality care. However, virtually all panelists, including the panelist who raised that point, agreed that a high death rate was at least as indicative of a disenfranchised population that failed to utilize services, a population with a greater number of patients with advanced disease, as it was of a population that lacked access to medical services. The panel noted that even in areas with highly generous Medicaid programs, many disenfranchised patients simply fail to enroll in State programs and therefore lack access to services. The panelists agreed that a measure of deaths among only those with AIDS could be a useful indicator of lack of access and severe case mix and supported the patient coverage group’s suggestion to include this variable in the index.

In extended conversation, the patient coverage panel developed a measure of the death rate from HIV and AIDS that could be used as a proxy for either severe case mix, or the failure of patients to receive adequate primary care. That discussion is reflected in that report.

Relative survival: Relative survival describes a methodology to adjust raw mortality rates from a given disease, in this case HIV/AIDS, by the mortality characteristics of the areas in which the deceased individuals resided (McDavid et al., 2003). By definition, this variable would always be inferior to an ideally collected measure of mortality caused by HIV/AIDS. The advantage of relative survival is that it can allow HIV/AIDS-specific mortality to be estimated given imperfect collection of the causes of patient death. The panel discussed using this variable to adjust reported mortality of AIDS cases in a given area. This is important because the CDC collects information only on the fact of death and not its cause for patients in the HIV/AIDS surveillance system. In other words, deaths among patients with HIV/AIDS reported by the CDC are from all causes. This issue is not trivial because many patients with HIV disease live high-risk lives and are much more likely to die from such causes as overdoses, homicide, suicide, and acute injuries than the general population, so attributing the raw death rate among them solely to complications of HIV disease could be highly misleading.

The panel was concerned that CDC surveillance data provided an inadequate amount of information from which to apply this adjustment. The panel decided that AIDS-related mortality in general, and relative survival as an adjustment to that rate, were intended to measure the concept of poor quality of health care and therefore were not issues for the Area Characteristics Panel to consider. They forwarded the issue and their research to the Patient Coverage Panel for review. However, the panelists who knew CDC surveillance data the best were highly skeptical that mortality data could be used to indicate deaths caused by AIDS.

^ back to top

B. Health Infrastructure

Variables considered: access to primary care providers, HRSA-supported clinics, hospital location, homeless assistance providers, and number of people without conventional housing.

The Health Infrastructure Workgroup evaluated the structural capacity of an area to care for patients with HIV and AIDS. They considered variables that evaluated the presence of medical facilities and services to house and assist the indigent, such as housing programs for the homeless. The workgroup thought areas that lacked services would require additional assistance from the CARE Act to serve the patients that lived there.

The workgroup believed that access to primary care providers was the best source to measure lack of health care access, because specialty care measures such as access to infectious disease physicians were of poorer quality and hospital location and the number of hospitals primarily measured access to inpatient services, which are not paid for by the CARE Act. The workgroup believed that measures of housing and homelessness were extremely important in measuring patients with the greatest need for CARE Act services but unfortunately could not identify data sources with adequate measures to include in the index.

1. Variables forwarded for consideration

Access to primary care providers: Access to primary care providers measures the ratio of primary care physicians to the general population. Panelists thought it was an important indicator of need because patients may have difficulty obtaining needed outpatient care in areas with provider shortages. Data on the number of primary care physicians are available from HRSA Bureau of Health Professionals, Health Professional Shortage Area (HPSA), and Primary Care Shortage Area (PCSA) databases.

2. Variables not forwarded for consideration

HRSA-supported Clinics: The variable, HRSA-supported clinics, measures the availability of HRSA-supported service centers that provide HIV care, often financed through the CARE Act. Measuring their availability may be helpful for an SON index, because it would indicate areas with few services relative to need. Data from this source also can be used to calculate the number of HRSA-supported providers in an area, and this value can be represented as a ratio compared to CDC-reported cases. The mixed group panel recommended the removal of this variable based on an unclear rationale for its inclusion.

Availability of health care services: The location of hospitals was at first thought to be a potentially useful indicator of health care access. However, given that the CARE Act does not reimburse inpatient services and that measures of primary care providers and CARE Act supported clinics are available, the additional value of hospital location as an indicator of access is low.