National HIV Surveillance System

The Centers for Disease Control and Prevention (CDC) maintains the National HIV Surveillance System (NHSS), which provides data about the HIV/AIDS epidemic for program planning and resource allocation. Started in 1981, the surveillance system is conducted in all 50 states and the District of Columbia, as well as American Samoa, Guam, Puerto Rico, the U.S. Virgin Islands, and the Northern Mariana Islands. In addition, the three freely associated states (the Federated States of Micronesia, the Republic of the Marshall Islands, and the Republic of Palau) report HIV surveillance data to CDC. CDC funds and assists state and local health departments to collect the information, and the state and local HIV surveillance systems represent valuable additional sources of data pertinent to HIV care.² The NHSS is a population-based census of all persons diagnosed and reported with HIV infection in the United States, including both those individuals receiving HIV care and those who are not in care.³

Since April 2008, all 50 states and the District of Columbia, as well as American Samoa, Guam, the Northern Mariana Islands, Palau, Puerto Rico, and the U.S. Virgin Islands, have been using the same confidential name-based reporting standards for newly diagnosed cases of HIV. Although the NHSS only includes data from those confidential name-based reporting systems that have been collecting HIV data for at least 4 years in the national aggregate numbers it publishes (CDC, 2010),⁴ all states and areas report HIV surveillance data to CDC, and the data for each reporting area are included in the annual HIV Surveillance Report. As such, the population in the surveillance system is one of the most nationally representative and provides the largest available sample of diagnosed PLWHA in the United States.

Another advantage of the NHSS is the use of standardized definitions of variables and reporting methods. In terms of data elements of interest to the committee,⁵ the system includes date of HIV/AIDS diagnosis; information on CD4+ cell count and plasma HIV RNA (viral load) closest to diagnosis; and optional fields for HIV and substance abuse treatment referral, pregnancy status, and antiretroviral therapy (ART) status at the time of reporting. Data gathered also can be used to monitor disparities with regard to race, ethnicity, sex, gender, age, geographic area, and country of birth.

In addition, most jurisdictions report all CD4 count and viral load lab results,⁶ which permits the tracking of individuals’ health status over time. In those cases, additional information can be extrapolated or calculated from available data. For example, the time between diagnosis and initial (or second) CD4 and viral load test can serve as a surrogate for the length of time between diagnosis and entry into care, and the number of routine HIV care visits per year may be estimated from the number of HIV-related lab reports per year. One limitation of the NHSS, as noted in the NHAS (ONAP, 2010, p. 18), is the fact that although most jurisdictions report all lab results, such as CD4 and viral load results, not all do. Another limitation is the problem of incomplete or inaccurate reporting by clinicians treating PLWHA and by state and local health departments. Studies have raised questions about the accuracy and completeness of NHSS data. A study comparing self-reported dates of HIV diagnosis with those reported to the NHSS indicates that 56 percent of the date pairs agreed on the year of diagnosis, with another 17 percent differing by 1 year and 19 percent by 3 or more years (Hall et al., 2005). Thirty percent of self-reported dates were earlier than those reported in surveillance data (Hall et al., 2005). Another study comparing date of first diagnosis based on self-report, medical record, and surveillance system data showed 51 percent agreement between self-reported year of diagnosis and the surveillance system and 70 percent agreement between the years reported in medical records and in surveillance (McCoy et al., 2010). Another 21 percent of self-reported dates differed by 1 year, while 23 percent differed by 3 or more years (McCoy et al., 2010). On average the self-reported dates were earlier than those recorded in the surveillance data (McCoy et al., 2010).

APPENDIX TABLE 3-5

CD4 and Viral Load Reporting by HIV Surveillance Reporting Area (as of June 15, 2010).

According to one study of reporting completeness, 81 percent of HIV diagnoses are reported within 12 months of diagnosis (Hall et al., 2006). This figure corresponds with the observation of McCoy and colleagues that only 81 percent of the cases in their study could be matched to surveillance data (McCoy et al., 2010). Incomplete reporting may be explained by lack of timeliness in reporting, failure to comply with case reporting, or an assumption that previously diagnosed cases already had been reported (McCoy et al., 2010). Structures are in place to improve the accuracy and timeliness of reporting. For example, increased use of electronic laboratory reporting is expected to increase the completeness and timeliness of HIV surveillance reporting (Overhage et al., 2008).

Despite current gaps in accuracy and reporting, the NHSS is one of the most representative HIV data systems that exists and offers a wealth of information. In terms of the data elements required to assess the core indicators for clinical HIV care, the NHSS, as previously noted, currently captures the individual’s date of diagnosis and the dates and results of the individual’s first and most recent viral load test and the CD4 test at, or closest to, the date that the individual was determined to be HIV-infected or to have AIDS (see Appendix Table 3-2a), as well as the first CD4+ cell count <200 cells/mm³. The ongoing CD4 and viral load test dates available for most reporting areas may be used as a surrogate for dates of first and ongoing visits for routine HIV care in those jurisdictions. These data permit estimation of the indicators for linkage to and continuity of care, regular CD4 and viral load testing, and individuals in care who achieve or maintain a CD4+ cell count of greater than 350 cells/mm³ (see Appendix Table 3-3a).

A revised version of the Adult HIV Confidential Case Report was approved by the Office of Management and Budget (OMB) in June 2011. The form includes a section that asks whether the individual “has ever taken any antiretrovirals (ARVs),” ARV medications taken, and dates ARVs were taken (date begun, date of last use). This information, which is collected “if required by Health Department,” is required for state and local health department that participate in CDC’s HIV Incidence Surveillance (HIS) and Variant, Atypical and Resistant HIV Surveillance (VARHS) activities and is optional for all other surveillance areas (CDC response to IOM request for information, October 20, 2011). When available, the ARV information may permit estimation of the core indicators pertaining to ART initiation and subsequent durable virologic suppression. However, there currently is no mechanism by which the NHSS can routinely capture ARV usage longitudinally. Longitudinal individual-level ART data in conjunction with longitudinal CD4 and viral load test dates and results would more reliably permit calculation of the relevant core indicators. Enhancement of NHSS data in this way would allow its use to evaluate all of the core HIV care indicators for the majority of the population diagnosed with HIV in the United States.⁷ Information is also captured on pregnancy status at the time the form is completed, which may be used to estimate the additional clinical care indicator pertaining to proportion of pregnant women with HIV who are receiving ART.

NHSS data also permit calculation of the core HIV care indicators for subpopulations of PLWHA based on age, race and ethnicity,⁸ sex assigned at birth, current gender identity, geographic distribution, and country of birth. Although CDC does not currently capture information specifically about individuals’ sexual orientation, which relates to the NHAS target and associated indicator pertaining to the proportion of diagnosed gay or bisexual men with undetectable viral load, combining data on sex assigned at birth with data collected on sex of sexual partner(s) (sex with male) can serve as a close proxy.

As is common with disease surveillance in the United States, the HIV surveillance system also does not collect information about income level. Unlike the previous version of the Adult HIV Confidential Case Report form, which included an optional section asking about the individual’s primary source of reimbursement for medical treatment (Medicaid, private insurance or HMO, no coverage, other public funding, clinical trial or government program, unknown), the current form does not collect that information. Collection of such data, especially if the Ryan White HIV/AIDS Program were added to the list of reimbursement source checkboxes provided on the form, would permit the use of NHSS data to estimate the indicators for the subpopulations specifically identified in the NHAS and would help to facilitate the evaluation of data across data systems as discussed in Chapter 6.

Uniform reporting to CDC of ongoing CD4 and viral load test dates and results from all jurisdictions and collection of longitudinal information on ARV usage would permit the use of data from the NHSS to assess all of the core indicators for HIV care identified by the committee. Use of national surveillance system data would permit evaluation of the indicators for the vast majority of the population diagnosed with HIV in the United States, as well as for subpopulations based on race, ethnicity, sex, gender, age, and country of origin. In addition, capturing information on sexual orientation and maintaining current geographic areas of residence for HIV-infected individuals in the system would further enhance the ability of the NHSS to be used to evaluate the impact of the NHAS and health care reform on HIV care in the United States.

Medical Monitoring Project

Initiated in 2005 in response to an Institute of Medicine (IOM, 2004) report, the Medical Monitoring Project (MMP) is a CDC-sponsored population-based surveillance system designed to collect comprehensive clinical and behavioral service need, utilization, and outcomes data on a nationally representative sample of adults (.18 years of age) living with HIV/AIDS who are receiving medical care from outpatient facilities in the United States and Puerto Rico (Blair et al., 2011). MMP is the first project since the HIV Cost and Services Utilization Study (Bozzette et al., 1998) almost 15 years ago that is designed to obtain comprehensive information about HIV care from a nationally representative population of PLWHA who are receiving care. MMP employs a probability proportional to size sampling design to obtain cross-sectional probability samples of its target population. A sample of about 400 individuals from each of 26 project areas (approximately 10,400 people) was selected each year for the 2007 and 2008 data collection cycles.⁹ Data are obtained from individual patient interviews and medical record review.

MMP captures most of the data elements needed to assess all of the indicators identified by the committee (see Appendix Tables 3-2a to 3-2e and 3-3a to 3-3c), including data on supportive services, which makes it an attractive source of data. In terms of demographic data, the interview component of MMP captures self-reported data on race, ethnicity, sex at birth, gender identity (male, female, transgender), and sexual orientation (homosexual, heterosexual, bisexual). In addition to the comprehensiveness of the data currently captured, the nature of the interview component of MMP allows flexibility to modify the questionnaire to capture different data elements that are subsequently determined to be useful. Starting with the 2011 cycle, for example, MMP is capturing data on stigma and discrimination, making it the only data system to do so among those examined by the committee.

APPENDIX TABLE 3-2c

Data Elements for Additional Clinical HIV Care Indicators.

An additional strength of MMP is its design to generate results that are nationally representative of the population of HIV-infected adults in care in the United States, which makes it a potentially valuable tool for tracking changes in access to and quality of HIV care in the country. Although MMP only includes HIV-infected individuals who are in care, the sample is not limited to those receiving care through a specific payer (Medicaid, Medicare, Veterans Health Administration [VHA], private or HMO) as is the case with a number of other data systems.

Despite its strengths, MMP also has several limitations. One significant concern about MMP is its low participation rate. For the 2007 data collection cycle, 10,192 individuals were determined to be eligible for participation. The median participation rate was 40 percent, ranging from 3 to 76 percent depending on the project area. Interview data ultimately were reported for 3,643 of the 3,944 participants interviewed; medical record abstraction data were not reported (Blair et al., 2011). As a result of the low participation rate, the data for the 2007 collection cycle may not be nationally or locally representative of HIV-infected adults receiving care in the United States. Steps have been taken to improve participation rates beginning with the 2009 collection cycle, and CDC anticipates that future data will permit nationally representative results (Blair et al., 2011).¹⁰ It is not clear, however, that the efforts will completely resolve the issues of nonresponse bias. For example, studies have found that PLWHA who are harder to reach and/or engage for study participation are more likely to be homeless or unstably housed; to be struggling with mental health or drug use problems; to be socially isolated; and to have high rates of missed appointments. Specific efforts to engage such populations are needed to ensure their representation in the study.

A second concern about MMP is the potential for social desirability response bias in the responses to the interview questions. Since many of the the interviews are conducted “in person,” respondents may be reluctant to answer accurately if doing so means providing what they perceive to be less “socially appropriate” responses to sensitive questions (Blair et al., 2011). Providing participants with a means to enter their responses to sensitive questions directly into the computer or on the response form is one way to help counteract social desirability response bias. This approach would avoid the necessity of sharing their responses with the interviewer and could improve the accuracy of the information collected (Carr et al., 1983; Greist et al., 1973; Kobak et al., 1996; Lawrence et al., 2010; Lucas et al., 1977; Metzger et al., 2000; Petrie and Abell, 1994; Waruru et al., 2005; Willig, 2011), although a study of clients at an addiction treatment center found no significant differences in the reliability of information on drug, alcohol, or tobacco use collected through computerized interviews, face-to-face interviews, or self-report formats (Skinner and Allen, 1983).

A third concern is the potential inaccuracy of clinical data (lab values, vaccinations, ART prescription) collected through participant self-report (Blair et al., 2011). Although a problem for reports of findings based on clinical information obtained solely from interviews (e.g., Blair et al. 2011), medical record abstraction is another component of MMP (CDC, 2009, pp. 22-25), which permits comparison with and corroboration of the self-reported clinical data. It is important that data from the medical record abstraction component of the protocol be available to permit such cross-checking and confirmation of self-reported information. The 2009 MMP protocol specifies that in project areas that have the surveillance authority to abstract medical records of selected patients without their consent, medical record abstraction should be completed for all sampled patients, including those who decline to participate in the interview or who cannot be located for interview (CDC, 2009, p. 25). In project areas with a more narrow definition of surveillance, where record abstraction cannot be completed without patient consent, minimal data can be collected on all sampled patients. The minimum data set contains the same fields as the NHSS case report form, and therefore these data can be collected in all project areas under HIV/AIDS surveillance authority.

Despite its current limitations, the research infrastructure, design, and implementation efforts that are in place make MMP a promising tool for monitoring care among HIV-infected adults receiving care in the United States. The committee supports the current efforts of CDC to improve individual participation and completion rates. Other strategies to increase participation might include providing additional incentives for study participants¹¹ and participating clinics or reducing the time required to complete the full interview by selectively eliminating certain questions.¹² In addition, implementation of the “minimum data set” records abstraction could help to provide some data for individuals who decline to participate.

Beginning with the 2012 data collection cycle, medical record abstraction will focus only on the 12 months preceding the interview; earlier clinical data will no longer be captured (Personal communication, Amy Lansky, Centers for Disease Control and Prevention, October 20, 2011). Although limiting medical record abstraction to the preceding 12 months likely will expedite collection of the data, certain data elements required to estimate some HIV care indicators may no longer be captured. For example, data on hepatitis B screening, vaccination, and immunity would not be captured if the relevant testing and immunization took place more than 12 months prior. Another option for reducing the number of questions in the standard interview without undermining the breadth of information provided might be to eliminate (some of) the self-reported clinical data if the same information already is being harvested through medical record abstraction. In addition, as electronic health records (EHRs) become more prevalent, MMP may be able to increase the scope of medical abstraction while retrieving and processing the data in a timely way. Such enhancements would help to make MMP better able to fulfill its promise as an expanded surveillance system for monitoring HIV care in the United States.

The potential for MMP to provide comprehensive information for tracking improvements in access to and quality of HIV care and supportive services in the United States is great; however, the low completion rate to date and the potential for nonresponse bias raise concerns about the representativeness of the data, especially for the homeless or unstably housed population and those with mental health and/or substance use disorders. Implementation of strategies to improve participation rates, especially among hard-to-reach populations, and to expedite the processing and availability of the data obtained through medical record abstraction would significantly increase the value of the project.

Ryan White HIV/AIDS Program Data

According to the Health Resources and Services Administration (HRSA), approximately 529,000 people currently receive at least one medical, health, or related support service through the Ryan White HIV/AIDS Program each year (HHS, 2011a). The AIDS Drug Assistance Program (ADAP), under Part B of the Ryan White Program, reported 213,764 clients enrolled during FY 2009, including 33,672 new enrollees, and 190,936 clients served (NASTAD, 2011, Table 5). The Ryan White Program is the third-largest federally funded program serving PLWHA (after Medicare and Medicaid) and the largest that serves only PLWHA (KFF, 2009b). Twenty-nine percent ($5.4 billion) of federal spending for HIV care was allocated to the Ryan White Program in FY 2011 (Kates, 2011, p. 1). The majority of Ryan White Program clients are low income, with approximately 70 percent at or below the federal poverty level (FPL) (HRSA, 2010, p. 45, Table 6).

HRSA launched a new reporting scheme in 2009, replacing the Ryan White HIV/AIDS Program Annual Data Report with the Ryan White HIV/AIDS Program Service Report (RSR). The RSR captures individual client-level data annually for individuals who receive one or more Ryan White–funded services (client report), as well as grantee and service provider information (grantee report and service provider report). The RSR client report generates a unique client identifier for every Ryan White HIV/AIDS Program client based on the client’s name, birth date, and other characteristics; the identifier is then encrypted before being sent to HRSA, further protecting the client’s privacy. Use of unique client identifiers not only permits tracking of individual clients across providers, generating more accurate client counts, but also permits the capture of individual-level demographic, clinical, and service utilization data, which can be used to assess quality of care received.¹³

The client report captures many of the data elements needed to assess the core clinical HIV care indicators identified by the committee (Appendix Tables 3-2a, 3-3a). The client report captures the year of birth (but not the full date); date of death; dates of ambulatory or outpatient HIV care visits; and the dates and results of all CD4 counts and viral load tests within the reporting period. In addition, the client report captures the year, but not the full date, of HIV diagnosis and whether the individual has been prescribed ART at any time within the reporting period, but not the date of ART initiation or subsequent prescriptions. Finally, the client report records the date of the client’s first ambulatory or outpatient care visit with the provider. However, the visit need not be for HIV care, nor is it necessarily the client’s first HIV care visit following diagnosis, which is required to assess the indicator pertaining to linkage to care.

The RSR client report also captures data relevant to the indicators related to mental health, substance abuse, and supportive services (see Appendix Tables 3-2b, 3-3b, 3-3d): screening for both mental health and substance use within the reporting period; number of mental health service visits; number of substance abuse service visits (inpatient and outpatient) in each quarter of the reporting period; housing status (stable permanent, temporary, unstable); and receipt of housing, food, and (medical) transportation services in each quarter of the reporting period (HRSA, 2011). The report does not provide information on referral for mental health or substance abuse services (e.g., whether the services were received within 60 days of referral); direct (e.g., dates) assessment of housing, food, or transportation need; or the proportion of clients who are food insecure or have an unmet need for transportation services, although such information might be inferred from the number of clients who are receiving food or transportation services.

APPENDIX TABLE 3-2b

Data Elements for Core Mental Health, Substance Abuse, and Supportive Services Indicators.

APPENDIX TABLE 3-3b

Data Systems Mapped to Core Mental Health, Substance Abuse, and Supportive Services Indicators.

The client report also collects data specific to a number of the additional clinical HIV care indicators (Appendix Tables 3-2c, 3-3c). These data include whether a client has been screened for tuberculosis (TB) during the 12-month reporting period or since being diagnosed with HIV; whether a client was screened for syphilis during the reporting period (excluding those under 18 years of age who are not sexually active); whether a client was screened for hepatitis B and C during the reporting period or since diagnosis with HIV; and whether a client has completed the hepatitis B vaccination series. In addition, data are collected on the pregnancy status of HIV-infected female clients, the stage of pregnancy at which they entered prenatal care, and whether they were prescribed ART to prevent maternal-to-child transmission of HIV. Data relevant to the additional clinical HIV care indicators that are not captured by the client report include dates of chlamydia and gonorrhea screenings; dates of influenza and pneumococcal immunizations; and data pertaining to ART drug resistance testing and ART initiation in individuals with HIV nephropathy, hepatitis B or C, or TB.

APPENDIX TABLE 3-3c

Data Systems Mapped to Additional Clinical HIV Care Indicators.

Although the data elements collected by the client report are not identical to those enumerated for the indicators identified by the committee, they provide information that may serve as a proxy for estimating many of the indicators. In addition, the RSR client report captures demographic data that specify subpopulations within Ryan White HIV/AIDS Program clients including race and ethnicity; gender (male, female, transgender, unknown; and for transgender, male-to-female, female-to-male, unknown); geographic code (first three digits of client zip code); income as a percentage of the FPL; and sources of health insurance.

ADAPs independently report data to HRSA, and ADAP reporting also is undergoing revision. The October 1, 2012 through March 31, 2013, data collection period is the first to capture individual client-level data for the ADAP Data Report (ADR), replacing the ADAP Quarterly Report. Like the RSR, the ADR will employ unique client identifiers using the same algorithm and encryption process as those used for the RSR. The encrypted client identifiers are meant to carry across the reports. In the future, although the reports will remain separate, client-level data from the RSR and the ADR will be merged into a single system, and the two reports will be linked for those clients receiving ADAP and other Ryan White–funded services (Personal communication, Faye Malitz, Health Resources and Services Administration, October 25, 2011).

Appendix Tables 3-2a through 3-2e summarize the data elements pertaining to the committee’s indicators that are captured by the ADAP Quarterly Report and the ADR, including those that are new for the ADR. Appendix Tables 3-3a through 3-3d map the committee’s indicators to the various data elements that are or will be captured by the ADAP reports.¹⁴ The ADAP reports do not supplement the data already captured in the RSR in terms of those needed to evaluate the committee’s indicators. However, for the population of ADAP clients who do not receive other Ryan White services, the ADR in particular can provide data to estimate a few of the core clinical HIV care indicators, such as the proportion of ADAP clients who have received CD4 and/or viral load testing in the past year (Appendix Table 3-3a). The ADR also may be able to provide the data to estimate the indicators pertaining to the proportion of clients with a CD4+ cell count that is less than 500 cells/mm³ who are on ART; the proportion of clients on ART for 12 or more months who have an undetectable viral load; and the proportion of female ADAP clients who are pregnant and on ART. However the data for these indicators are limited to ART drugs that are fully ADAP funded. If a client is not receiving at least one such drug, that person will not be identified as being on ART.

The ADAP reports capture no data pertaining to mental health or substance use screening or services or to the need for or use of supportive services for housing, food, and transportation. Demographic data captured in the ADAP reports are more limited than those captured by the RSR, limited to race and ethnicity, gender (as in RSR), and for the ADAP Quarterly Report, percentage of clients with an annual household income less than 200 percent of the FPL. The ADR includes year of birth and insurance status or type, as well as income as a percentage of the FPL.

As a stand-alone data system, the ADAP reports are of limited usefulness in providing the data needed to estimate the indicators identified by the committee for tracking the provision of HIV care and mental health, substance use, and supportive services in the United States. However, ADAP data may prove useful for assessing waiting time for access to ART drugs and the proportion of people who need, but do not have access to, ART. The committee supports HRSA’s intention to merge the client-level data from the RSR and the ADR into a single system.

Ryan White HIV/AIDS Program data are an important source of information for monitoring access to quality HIV care and supportive services because of the population represented and the importance of the program in providing care and services to many disadvantaged populations. By increasing health insurance options and extending Medicaid coverage to nondisabled individuals who meet the expanded income criteria, implementation of the ACA is expected to reduce the dependency of a portion of current Ryan White HIV/AIDS Program clients on the program to meet their health care service needs, although the Ryan White HIV/AIDS Program likely will continue to serve an important role in providing HIV care to individuals who remain uninsured. Reduction in the use of Ryan White funds for medical care would permit the redirection of funds to other vital Ryan White–funded services. The Ryan White HIV/AIDS Program has an established role in providing a comprehensive array of services beyond medical care, including medical case management and treatment adherence counseling, mental health and substance abuse treatment services, oral care, food assistance, medical transportation, and psychosocial support (IOM, 2011, p. 20). Increased emphasis on such services through continued funding of the Ryan White HIV/AIDS Program and decreased demand for medical care among clients would continue to advance the goals of the NHAS in important ways, for example, by supporting PLWHA “who have challenges meeting their basic needs, such as housing” (ONAP, 2010, p. 21). As one of the few data systems examined by the committee that capture data on housing, food, and medical transportation need, an increase in Ryan White funding available for such supportive services would make it an even more valuable source of data on those services.

Although the population of PLWHA receiving services through the Ryan White HIV/AIDS Program is a large and important one, it is not nationally representative of PLWHA, and use of Ryan White data to estimate the indicators will only permit tracking of the indicators for that group. Another difficulty with Ryan White HIV/AIDS Program data is that data pertaining to medical and supportive services received are reported only when the services were funded with Ryan White dollars. Such services include mental health and substance abuse treatment visits and housing, food, and transportation services. An organization might receive funding from a number of different sources, and if a client were to receive some services funded, at least in part, through the Ryan White HIV/AIDS Program and other services funded exclusively by another source, only the former would be reported to HRSA. Ryan White–funded services vary widely among and within states, depending on how state and local jurisdictions tailor services to meet the needs of local communities (Rawlings and Hopson, 2009). There are persistent dollar-per-case federal allocations to states (Martin and Keenan, 2011), which are associated with the size and scope of ADAP drug formularies (Martin and Barry, 2011). If other sources of state and local funding are used to provide these additional services, they will not appear on the client’s record. To obtain a comprehensive picture of access to needed services within the Ryan White HIV/AIDS client population, it would be helpful to have information on all pertinent services received by clients regardless of funding source, as is the case for clinical data. The clinical information reported by providers who receive Ryan White HIV/AIDS Program funding includes all of the data requested for each Ryan White HIV/AIDS Program client, regardless of how the service was paid for and who delivered it. Thus, all of a given client’s outpatient or ambulatory care visit dates, CD4 and viral load counts, and the like within the reporting period are included.

Along with MMP, the RSR is one of two data systems to provide information on the need for and utilization of supportive services for housing, food, and transportation, as well as HIV medical care and mental health and substance use services.¹⁵ The data’s usefulness is limited by the reporting only of those supportive services that are funded through the Ryan White HIV/AIDS Program. Just as all pertinent clinical data are reported by Ryan White–funded providers for clients regardless of payment source, reporting of complete data for supportive service utilization would provide more robust information for tracking the impact of the NHAS and health care reform on the provision of these services. Absent reporting of all supportive service utilization, an indication of whether clients had received any non-Ryan White-funded services would allow analyses of Ryan White HIV/AIDS Program data to be stratified accordingly.

Medicaid Statistical Information System

Medicaid is the largest safety-net health insurance program in the United States, providing health and long-term care coverage to more than 59 million low-income and disabled beneficiaries (KFF, 2011a). Although PLWHA represent less than 1 percent of the total Medicaid population, in FY 2007 Medicaid provided coverage for 47 percent of PLWHA estimated to be receiving regular medical care: 212,892 Medicaid beneficiaries were HIV infected (Kates, 2011, p. 1). Medicaid is financed jointly by the federal and state governments and represents the largest expenditure on health care coverage for PLWHA when federal and state funds are combined. Together, federal and state Medicaid expenditures totaled $9.3 billion in FY 2011, accounting for 51 percent of federal spending for HIV care (Kates, 2011, p. 1). Medicare accounts for another $5.4 billion (29 percent) of federal funding for HIV care (Kates, 2011, p. 1). Approximately 29 percent of Medicaid beneficiaries with HIV were dually eligible for Medicare in FY 2007 (Kates, 2011, p. 1).

The Medicaid Statistical Information System (MSIS) is the claims processing system for Medicaid, which captures utilization data and management information pertaining to medical care and services provided to Medicaid recipients. MSIS includes the full population of people with HIV/AIDS enrolled in Medicaid in the United States,¹⁶ and, given Medicaid’s prominent role in HIV care (covering 47 percent of PLWHA estimated to be in care), it not only captures a significant share of PLWHA but also is a critical source of care and coverage that should be assessed. Currently, to qualify for Medicaid individuals must be low income and be “categorically eligible.” Most Medicaid beneficiaries with HIV (74 percent) qualify through the disability pathway, meaning their disease is sufficiently advanced to preclude them from working (Kates, 2011, p. 4). The anticipated expansion of Medicaid under the ACA will remove the categorical eligibility requirement and extend eligibility to most people under the age of 65 who have incomes less than 133 percent of the FPL (Kates, 2011, p. 4). The resulting increase in Medicaid’s role in covering care for PLWHA makes MSIS a particularly important source of data for tracking the impact of the ACA on HIV care.

States are required to report Medicaid beneficiary and claims data quarterly to the Centers for Medicare and Medicaid Services (CMS) through the Medical Management Information System (MMIS).¹⁷ These data, which include demographic and monthly enrollment data for each person covered by Medicaid in the quarter (eligible files) and adjudicated claims data (paid claims files), are captured in MSIS. Claims files are categorized by inpatient, long-term care, prescription drug, and noninstitutional services and include data on types and dates of services, providers, costs and types of reimbursement, and epidemiological variables (CMS, 2011c).

MSIS data are available in two forms: MSIS files and Medicaid Analytic eXtract (MAX) files. MSIS files are organized quarterly for the federal fiscal year (October–September) and by transaction or claims adjudication date. They cover all enrollment transactions, including retroactive enrollment and corrections, as well as all interim claims records, including originals, voids, credits, debits, and the like. MAX files contain data extracted from the MSIS files and formatted to facilitate research and public policy needs. They are organized chronologically by calendar year, based on date of service, and MSIS claims records (initial, interim, voids, and adjustments) are combined or consolidated to generate final records for specific services covered by Medicaid as accurately as possible (CCW, 2011c; CMS, 2011a,d). MAX files include a person summary file, as well as inpatient hospital, long-term care, prescription drug, and other services files.¹⁸

MAX files are available to approved academic researchers and certain government agencies through the CMS Chronic Condition Data Warehouse (CCW). MAX files currently are available for 1999 through 2008, although 2008 data are not yet available for all states (ResDAC, 2011a).¹⁹ MSIS files are available through FY 2009 (48 states) and for 22 states for FY 2010 (CMS, 2011b). Although most states complete their reporting within a year, not all do, resulting in about a 2-year lag time for MSIS data files. The lag time is somewhat longer (2.5-3 years) for MAX data files because the raw MSIS data must be extracted and consolidated (CCW, 2011c). A built-in lag of at least 13 to 14 months is needed to ensure that claims for most services delivered in a given calendar year are captured, and another 9 to 10 months are needed to validate and process the data (ResDAC, 2011b).²⁰

Advantages of MSIS as a source of data for HIV care include the large number of HIV-infected individuals represented (although enrollees with HIV are identifiable only if they have a diagnosis for HIV entered in the system); strong representation of “vulnerable populations,” including racial and ethnic minorities; regular collection of data over time (currently quarterly, moving to monthly within 2 years [CMS response to IOM request for information, April 8, 2011]), linkage of data to unique personal identifiers, and an existing data processing and data retrieval structure. In addition, diagnostic and treatment information is reported by providers, which may reduce inaccuracies inherent in patient self-report, although nonclinical factors can affect provider reporting also. For example, changes in disease recognition, treatment, and prescription patterns, as well as billing or reimbursement considerations, may have an impact on provider reporting (Crystal et al., 2007). Inaccurate reporting (errors, coding variation, designation) is another nonclinical factor that can affect the accuracy of the data available from MSIS or MAX. MSIS and MAX data include eligibility and claims data and limited demographic data. As is the case with all claims databases, information is available to chart “quality of care” based solely on medical service and medication utilization. Dates of service, diagnosis and procedure codes, and provider codes are available, but core outcome measures such as CD4 and viral load test results are not. Some negative outcome indicators would be available, such as treatment for an opportunistic infection or mortality, based on date of death. Assessment for mental health treatment needs or medical comorbidities would be indicated only if a diagnosis code appears in the case file to justify treatment or medication. Appendix Tables 3-2a through 3-2e summarize which data elements of interest to the committee are captured in MSIS.

One challenge of using MSIS or MAX data is identification of the population of PLWHA who are Medicaid beneficiaries. Variations in diagnostic and other service coding may adversely affect the usefulness of any particular group of codes for accurately identifying the Medicaid population with a given condition. Therefore, use of a combination of diagnosis codes (for HIV/AIDS), common procedure codes (CD4 counts; HIV RNA tests), and prescription drug codes (ARVs) is likely the best way to identify the maximum number of PLWHA among Medicaid recipients with the greatest positive predictive value (see Crystal et al., 2007; Koroukian et al., 2003). Difficulties other than those related to coding also limit the ability to identify the full population of PLWHA within Medicaid. The fluctuating eligibility of some beneficiaries causes those individuals to move in and out of coverage during the course of a year, meaning any medical care they receive in the period during which they are not covered is not captured by Medicaid claims data. Dual eligibility with Medicare also causes claims covered by Medicare not to be captured in Medicaid data. Each of these situations makes it probable that MSIS or MAX data on encounters will not provide a complete accounting of medical services received by individuals in the group (Koroukian et al., 2003). Not only may some Medicaid recipients with HIV not be identified at all, but a number of others within these groups will have incomplete encounter data in MSIS, resulting in an underestimation of the indicators for the population of Medicaid recipients (Crystal et al., 2007).

Similar to the cases in which Ryan White HIV/AIDS Program data include only Ryan White–funded services, even if the MSIS data were complete and accurate, state variation in covered services beyond a set of “mandatory” services required to receive matching federal funds and service payment structure (fee-for-service versus prepaid plans) would mean those data still would not provide a complete accounting of service utilization by individual recipients. For example, MSIS may include data pertaining to a given type of service for some beneficiaries (those residing in a state in which the service is covered) but not capture data on the provision of the same type of service for Medicaid beneficiaries residing in a state where the service is not covered. In addition, states may place limits on the number of occurrences (prescriptions, inpatient days, provider visits) that Medicaid will cover. In both types of case, Medicaid claims data will not provide a complete picture of service utilization by individual beneficiaries.

In other cases, MSIS data may not be complete or accurate. Fee-for-service plans generate fairly complete utilization data because reimbursement depends on filing a claim for each covered service. However, in FY 2007, 71 percent of Medicaid beneficiaries with HIV received some covered services through managed care plans (Kates, 2011). Although states are required to report utilization for beneficiaries in prepaid plans (HMOs, preferred health plans), the accuracy and completeness of these data are suspect (CCW, 2011c; Crystal et al., 2007). Both situations (variations in Medicaid coverage, incomplete or inaccurate Medicaid data) may result in incomplete service utilization data being available from MSIS on specific individuals. Identification of the most common service providers for individuals with variable Medicaid eligibility or of services not covered by Medicaid (e.g., Ryan White HIV/AIDS Program) would permit use of data from these additional sources to gain a more complete measure of the full set of services received by these individuals. Likewise, combined Medicare and Medicaid data for individuals dually eligible for both programs also would provide a more complete picture of service usage. As discussed in Chapter 6, although combining data from multiple data systems to generate a more complete measure of service utilization for the purpose of estimating the recommended indicators is a theoretical ideal, doing so in practice poses numerous statistical challenges.

Demographic data of interest are limited to date of birth, date of death, gender (male, female), race and ethnicity, and zip code. MSIS also collects limited information on private payer status. Income (as a percentage of the FPL) is an optional field, although information about income level could be inferred based on eligibility criteria. Also, data are collected on whether the beneficiary received Temporary Assistance for Needy Families benefits during the month. MSIS links data to unique individual identifiers (either MSIS generated or Social Security number, depending on the state), so that information may be tracked across time for individuals, permitting evaluation of their longitudinal care experiences to the extent permitted by claims data. The demographic data collected would permit assessment of indicators for racial and ethnic subgroups of interest, as well as subgroups based on location of residence and payer status. Since data on sexual orientation are not collected, MSIS data do not permit estimation of the indicators for the NHAS-targeted subgroup of gay and bisexual men, although separate assessments could be made for men and women with data based on sex. With respect to the core indicators of HIV care, MSIS could be expected to provide the data needed to assess the indicators pertaining to continuity of care and regular CD4 and viral load testing, based on claims submitted for office visits with HIV listed as one of the diagnosis codes and claims submitted for CD4 and viral load tests, all of which capture dates of service (see Appendix Table 3-3a). However, any services received by an individual that were not reported to CMS would not be included in MSIS, resulting in gaps in the information available. MSIS captures date of death and so could provide data to calculate the mortality rate within its population of PLWHA.

MSIS captures neither the date of HIV diagnosis nor the date of first visit for HIV care; thus it cannot be used to assess the linkage-to-care indicator. Also, since MSIS does not capture clinical data, such as the results of CD4 counts and viral load tests, it cannot independently provide the information needed to assess the remaining core indicators of clinical HIV care, even though it collects data on the prescription and (re)fill dates for ART drugs, when claims are submitted.

MSIS captures data pertaining to screening and visits for mental health and substance use services covered by Medicaid, but it does not specifically capture the dates of (hence, the time between) diagnosis or referral and first visit for services. With regard to supportive services, MSIS collects data on the provision of social work or case management services, but specific information pertaining to housing, food, and transportation needs is not captured (see Appendix Tables 3-2b, 3-3b).

For the additional indicators (see Appendix Tables 3-2c, 3-3c), MSIS could provide data to assess the clinical HIV care indicators relating to TB, sexually transmitted infections (STIs), and hepatitis B and C screenings, along with influenza, pneumococcal pneumonia, and hepatitis B immunizations, although clinical information about whether the TB test results were interpreted or hepatitis B immunity was documented would not be available. Data would also be available to assess the indicators pertaining to drug resistance testing and the proportion of HIV-infected pregnant women on ART, although pregnancy status would have to be extrapolated from related diagnostic or service codes. Data to assess indicators relating to timely diagnosis of HIV infection and those involving clinical markers for ART initiation would not be readily available.

Despite the importance of data from the Medicaid population for tracking the impact of the NHAS and the ACA on HIV care, MSIS and MAX data have some limitations. As previously noted, the lag time from service utilization to reporting completion (especially for MAX data) may be problematic for time-sensitive policy evaluation. In addition, Medicaid data alone may not provide a complete accounting of service utilization by beneficiaries who receive services from multiple funding sources, and strategies must be employed to help correct for that additional encounter data. Stephen Crystal and colleagues (2007) list some “best practices” for working with Medicaid data. Development of methods for combining data from or analyzing data across additional relevant data systems (e.g., Medicare, Ryan White HIV/AIDS Program) might provide more complete information on service utilization for individuals receiving services through two or more of the programs. One such effort is the database of linked Medicaid and Medicare data developed by CMS in 2009, which contains service utilization and expenditures data for 9 million dually eligible beneficiaries (CHCS, 2010). CCW assigns a unique beneficiary identification number for the MAX and Medicare records of dually eligible beneficiaries to permit tracking and analysis of data across programs (CCW, 2011c, p. 4).

Chronic Condition Data Warehouse

Although not representing as large a patient population as Medicaid, Medicare accounted for 29 percent of federal spending on HIV in FY 2011 (Kates, 2011), the largest source of federal spending on HIV care.²¹ Medicare is a federal program providing health care coverage to disabled individuals and those age 65 and older. Approximately 100,000 Medicare beneficiaries are HIV-infected, representing about 20 percent of HIV-infected individuals estimated to be receiving care in the United States (KFF, 2009a). The majority of PLWHA currently receiving Medicare qualified through the disability pathway. With the evolution of HIV into a chronic condition, many PLWHA are living longer and increasingly are expected to qualify for Medicare on the basis of age, resulting in an increase in the number Medicare beneficiaries with HIV. Given Medicare’s Part D prescription drug coverage and the increasing number of Medicare-eligible PLWHA, Medicare plays an important role in HIV care coverage.

The CCW contains fee-for-service²² claims data for 100 percent of Medicare beneficiaries from 2005 to 2009²³ and Part D drug event data from 2006 to 2009 (CCW, 2011a, About). As such, it includes fee-for-service utilization data for all PLWHA who are enrolled in the Medicare program. To expedite delivery and maximize cost efficiency, data sets are available for predetermined cohorts representing 21 chronic conditions. Although HIV/AIDS is not presently one of the predefined cohorts, it currently is under consideration for addition to the list of flagged conditions (CCW, 2011a, Chronic Conditions).

Medicare uses a unique beneficiary identification number and collects the basic demographic data of interest: date of birth, date of death, OMB-defined race and ethnicity, gender (male, female), and zip code. The Medicare Current Beneficiary Survey (MCBS) of a representative national sample of Medicare beneficiaries is used to generate two files (Access to Care; Cost and Use) each year. The Access to Care file “contains summaries of use and expenditures for the year from Medicare files along with survey data on insurance coverage, health status and functioning, access to care, information needs, satisfaction with care, and income” (CCW, 2011b, p. 10). Although MCBS data are not automatically linked to Medicare beneficiary identification numbers, information is available upon request to permit MCBS data to be merged with other CCW data at the beneficiary level. Medicare also requires assessments for beneficiaries receiving care in nursing facilities, inpatient rehabilitation facilities, and home care. These assessments provide data on certain aspects of beneficiaries’ health status, as well as other relevant information (e.g., the safety and sanitary condition of the individual’s home for those receiving home care) and likely will increase in importance as the population of PLWHA ages and individuals with HIV enter home and institutional nursing care in greater numbers.

As a source of claims data, Medicare is similar to Medicaid in terms of the data available to assess the core indicators of HIV care. It should be able to inform indicators related to continuity of care, regular CD4 and viral load testing, and mortality rate, but it does not contain the information necessary to evaluate the linkage-to-care indicator or the clinical data needed to assess the other core HIV care indicators. ART drug prescription and (re)fill data are available for Medicare Part D beneficiaries. The availability of Medicare data to assess the additional indicators for HIV care is similar to that of MSIS data as well.

Although Medicare does not have data on screening for mental health disorders or substance use, it does capture service utilization data on the diagnosis of and covered treatment for these conditions, but as with Medicaid, the data do not specifically permit calculation of the time between treatment referral and receipt of services. Medicare does not collect data on housing, food, or transportation needs, although questions pertaining to housing adequacy are included in the assessment for home health beneficiaries (OASIS).

As previously noted, efforts to link Medicare and Medicaid data for dually eligible beneficiaries will provide a more complete picture of service utilization for that group of individuals. In addition, inclusion of HIV/AIDS in the list of predetermined chronic condition cohorts for which CCW data sets are available should expedite delivery of these data for research and policy use.

North American AIDS Cohort Collaboration on Research and Design

The North American AIDS Cohort Collaboration on Research and Design (NA-ACCORD) captures data from 22 single and multisite clinical and classical epidemiologic HIV cohorts, which represent most of the HIV/AIDS cohort studies in North America, including the CFAR Network of Integrated Clinical Systems (CNICS) and the HIV Research Network (HIVRN) discussed in the following sections. Although NA-ACCORD includes data from CNICS and HIVRN, not all of the data elements captured by those systems are represented in NA-ACCORD.

NA-ACCORD collects data on more than 100,000 HIV-infected adults from more than 60 academic research and hospital- and community-based clinical sites throughout the United States (44 states and the District of Columbia) (Kitahata, 2011; NA-ACCORD, 2011; NA-ACCORD response to IOM request for information, March 30, 2011). NA-ACCORD is designed to be widely representative of HIV care in the United States,²⁴ and the population of PLWHA represented is similar to that reported by the CDC in terms of age and sex, but it includes somewhat fewer minorities²⁵ (Kitahata, 2011; NA-ACCORD response to IOM request for information, March 30, 2011).

NA-ACCORD data pertaining to the core clinical HIV care indicators include date of diagnosis (although these data are not complete); date of first visit at the clinical site and whether the individual was previously seen at another site (but not always the date of first-ever visit for HIV care); dates of routine HIV care visits; CD4 and viral load test dates and results; dates that individual ARVs were started or stopped; year of birth; and mortality information (date and cause of death) (Appendix Tables 3-2a, 3-3a). For the additional indicators of clinical HIV care, NA-ACCORD also includes dates of ART drug resistance testing; diagnoses of AIDS-defining conditions; diagnoses of and/or laboratory results relevant to renal disease (nephropathy), hepatitis C, and hepatitis B; dates of hepatitis C and hepatitis B screening; and dates of TB testing (Quantiferon-TB tests). As of 2012, NA-ACCORD collects information on pregnancy status. NA-ACCORD does not currently collect dates of screening for chlamydia, gonorrhea, and syphilis or dates of vaccination or immunization for hepatitis B, influenza, or pneumococcal pneumonia (see Appendix Tables 3-2c, 3-3c).

Information on diagnosis or referral for mental health disorders and substance abuse is included in NA-ACCORD, but not data on screening for those disorders or on first visit for mental health or substance abuse treatment, although it will capture visits for psychiatry, psychology, and counseling, beginning in 2012. NA-ACCORD does not include data pertaining to housing stability, food security, or access to transportation but could add such data to the extent they are collected by social workers in the clinical practice setting (Appendix Tables 3-2b, 3-3b).

Demographic information available in NA-ACCORD includes age, sex, race and ethnicity, and the first three digits of zip code, as well as metropolitan statistical area (MSA), and, as of 2012, country of birth. Data on sexual orientation, gender identity, income, and insurance status are not collected. (See Appendix Table 3-2d.)

APPENDIX TABLE 3-2d

Data Elements for Additional Mental Health, Substance Abuse, and Supportive Services Indicators.

Data collected by NA-ACCORD can be used to estimate all of the core indicators of clinical HIV care. The data system is less useful for estimating the additional clinical care indicators and much less so for indicators pertaining to mental health, substance abuse, and supportive services. A strength of NA-ACCORD is that new data elements can be added relatively easily if they are collected by the participating cohorts.

CFAR Network of Integrated Clinical Systems

The CFAR Network of Integrated Clinical Systems comprises a network of eight Centers for AIDS Research (CFAR) sites²⁶ that have implemented point-of-care electronic data collection systems. CNICS data are collected prospectively through these systems on PLWHA in care at the site and, therefore, characterize the rapidly changing course of HIV disease management. The CNICS cohort includes more than 23,000 individuals and represents a diverse population of patients with regard to sex, race, ethnicity, age, risk factor for HIV transmission, and geographic distribution, although the CFAR sites are located in urban areas and the data may not be representative of PLWHA in rural areas (CNICS, 2011, CNICS sites; Kitahata, 2011).

CNICS currently maintains data on 10 unique domains: (1) disease diagnoses; (2) laboratory data (viral load, CD4 count, viral hepatitis, hematologic, kidney, and chemistries or metabolic markers); (3) medication data; (4) demographics (sex, race and ethnicity, age, and risk factor for HIV transmission); (5) health care utilization (initial patient enrollment, primary care visits, and hospitalizations); (6) vital status (death date, source, and cause of death); (7) patient-reported outcomes²⁷; (8) antiretroviral drug resistance; (9) biological specimens; and (10) census block data (CNICS, 2011, Data elements).

According to the project website, an important distinction between CNICS and other cohorts is the ability to provide peer-reviewed open access to data for research from a system that prospectively collects comprehensive patient data including validated outcomes, longitudinal resistance data, and PROs (CNICS, 2011).

CNICS includes much of the data needed to calculate the core clinical HIV care indicators (Appendix Tables 3-2a, 3-3a), including dates of first and ongoing visits for HIV care, CD4 and viral load test dates and results, and dates of starting and stopping specific ARVs. Information on date of diagnosis is collected by CNICS, but these data are not available for some individuals. For the additional clinical HIV care indicators (Appendix Tables 3-2c, 3-3c), CNICS collects data on diagnosis of AIDS-defining conditions; hepatitis B and C; chlamydia, gonorrhea, and syphilis screening; ART drug resistance testing; and diagnosis of renal disease. Data also are collected on TB testing using QuantiFERON-TB tests. As of 2012, data on pregnancy status is collected, and collection of immunization information for hepatitis B, influenza, and pneumococcal pneumonia is proposed. If all these data are collected, CNICS will capture the data needed to estimate all of the core and additional indicators for clinical HIV care.

CNICS also collects certain data on mental health and substance abuse disorders, including dates of screening and diagnosis or referral. Although CNICS does not specifically capture the date of first visits for mental health treatment services, it does capture visits for psychiatry, psychology, and counseling. Since it also captures date of screening, a proximate visit for services would suggest date of first visit, but it would not be flagged as such. CNICS captures whether an individual received substance abuse treatment in the past year, but not specific dates of service. CNICS does not currently include data on housing, food, and transportation needs assessment or status, but data on housing stability (stable or permanent, temporary, unstable, and homelessness) are collected in the clinical practice setting and could be added to CNICS (CNICS response to IOM request for information, April 11, 2011).

Demographic data captured in CNICS include age, sex, race, and ethnicity. Data on sexual orientation and gender identity are not currently collected. CNICS collects MSA of residence and is adding the three initial numbers of individuals’ zip codes, which provide state and at least county of residence, as permitted by the Health Insurance Portability and Accountability Act. Country of birth is collected beginning in 2012. Insurance status is collected, and income data are collected and could be added to CNICS. PROs provide an opportunity to collect qualitative data on satisfaction with provider care and on stigma or discrimination, as well as other information of interest, including food security and transportation needs.

HIV Research Network

The HIV Research Network compiles data electronically from health records and through manual medical record review in order to obtain, analyze, and disseminate current information on the delivery of services to PLWHA. HIVRN captures these data longitudinally to assess trends in areas such as accessibility, quality, utilization, safety, and costs of HIV-related health care services. HIVRN is primarily supported by the Agency for Healthcare Research and Quality with additional support from other agencies of the U.S. Department of Health and Human Services (HHS).

HIVRN represents a consortium of 16 academic and community-based sites that provide primary and subspecialty HIV care in 13 cities throughout the United States, with 8 in the eastern United States, 1 in the Midwest, 3 in the South, and 4 in the West (HIVRN, 2011). Data are collected annually on the clinical and demographic characteristics of approximately 21,000 adults, adolescents, and children receiving HIV care at the participating sites (HIVRN response to IOM request for information, March 30, 2011). (Five of the sixteen sites are devoted to pediatric care.) The data from each site are sent to the data coordination center at the Johns Hopkins School of Medicine, where they are consolidated into a single uniform database.

HIVRN data can be used to estimate all of the core clinical HIV care indicators identified by the committee. In addition, data for estimating the core indicators for mental health and substance abuse are available for a subset of the participating sites. HIVRN does not collect data on the dates of screening for mental health or substance abuse disorders, nor does it collect any data pertinent to the committee’s core or additional indicators for housing, food security, or unmet need for transportation. As with NA-ACCORD and CNICS, HIVRN data are collected from urban areas and may not be representative of PLWHA in rural areas.

Clinical Case Registry: HIV

The Veterans Health Administration within the Department of Veterans Affairs is the largest provider of HIV care in the United States, serving more than 24,000 veterans with HIV in 2010 (VA, 2011). The Clinical Case Registry (CCR): HIV is an administrative and clinical database containing population-based data on HIV-infected individuals who receive care through the VHA.²⁸ Local reporting allows clinicians with access to the database to monitor clinical outcomes and resource utilization. The national database permits quality of care, as well as outcomes and utilization, monitoring. Data on all veterans with a confirmed diagnosis of HIV/AIDS in VHA care during the calendar year are included in the database. The sample is limited to those veterans who receive care within the VHA system and is older and predominantly male, compared with the overall population of PLWHA in the United States (VHA response to IOM request for information, April 11, 2011).

The VHA has a sophisticated EHR system that captures utilization and outcomes data. Data needed to calculate all of the core clinical HIV care indicators and most of the additional clinical care indicators are available from the EHR if the services are performed within the VHA system (Appendix Tables 3-2a, 3-2c, 3-3a, 3-3c). Date of diagnosis, CD4 count at diagnosis, and date of first visit for HIV care are all available for individuals diagnosed and treated within the VHA, but for those who transfer into the system following diagnosis, data on linkage to care and stage of disease at diagnosis are not available. Information on prescriptions and refills written by VHA providers is available as are dispensing data for prescriptions (re)filled through the VHA pharmacy system.

Although prenatal care is covered by the VHA, prenatal care services are provided outside the system by community providers. Although the VHA EHR does not capture data from external providers, the information pertaining to ART prescription for pregnant women would be available for prescriptions filled through the VHA pharmacy system.

Data pertaining to screening for mental health disorders and substance use are not captured in the VHA data system (VHA response to IOM request for information, April 11, 2011), but it does include data on diagnosis of or referral for mental health and substance abuse disorders, as well as date of first visit for treatment services if they occur in the VHA. Data pertinent to the supportive services indicators are not collected, although some data pertaining to social work or case management are captured. (Appendix Tables 3-2b, 3-3b). Demographic data collected include age, sex, race, ethnicity, and address. Data on gender identity, sexual orientation, income, insurance status, and country of birth are not collected. (See Appendix Tables 3-2d, 3-2e.)

As an EHR, the VHA data system contains comprehensive clinical data on test and treatment services provided within the system, including prescription and pharmacy dispensing data, although information on services provided outside the system is not reliably captured.²⁹ As an integrated health care system, the VHA is well poised to respond to challenges raised by the NHAS, as demonstrated by its recent efforts to implement routine HIV testing. Although the population of PLWHA served by the VHA is disproportionately male and older compared to the national population of PLWHA, recent collaborative efforts between the VHA and Kaiser Permanente (KP) (discussed in the following section) may be the first step in addressing concerns about representativeness.

Kaiser Permanente

Kaiser Permanente is one of the largest not-for-profit health plans in the United States, providing coverage to more than 8.7 million members in nine states and the District of Columbia (KP, 2011). Health outcome and utilization data are collected on all members through EHRs and databases. The second-largest private provider of HIV care in the United States in 2006, with more than 16,000 HIV-infected individuals in care, KP data represent a diverse population of individuals with private insurance in California, Hawaii, and selected metropolitan areas, including Baltimore, Maryland; Washington, DC; and Atlanta, Georgia. Analyses have shown that KP is representative of the HIV-infected population in California (KP response to IOM request for information, March 30, 2011), a state with more than 6.5 million KP members (KP, 2011), and as the largest provider of HIV care in Hawaii, KP is representative of the population there as well (KP response to IOM request for information, March 30, 2011).

KP has a sophisticated EHR system that facilitates the capture and retrieval of detailed clinical data. A major benefit to a robust EHR system is the availability of data on both service utilization and clinical outcomes. KP captures all of the data elements necessary to assess the core indicators of clinical HIV care, although the date of HIV diagnosis is only captured for individuals diagnosed within the KP system (Appendix Tables 3-2a, 3-3a). Thus, the linkage-to-care indicator can be calculated reliably only for those in the system at the time of diagnosis. KP also captures the data needed to assess the additional indicators for clinical HIV care (Appendix Tables 3-2c, 3-3c). Data on the prescription of ART drugs are available, as are most pharmacy (re)fill data. No data are available on prescriptions (re) filled at pharmacies outside of the KP system.

KP also records data on screening for mental health disorders and substance abuse, although the screenings are performed as indicated and not according to a predetermined schedule. Data on referral for services for mental health disorders and substance abuse are included in individuals’ EHRs, as are data on receipt of treatment services that are provided within the KP system. Data relevant to supportive services indicators are not routinely collected within the KP system.

As an integrated health care system with a comprehensive EHR system, KP, like the VHA, captures comprehensive clinical test and treatment data for services provided within the system. Although the population of PLWHA served by KP is not nationally representative, it is representative of the privately insured HIV-infected population in areas of the United States with access to Kaiser (Hawaii; California; Oregon; Mid-Atlantic region, including the District of Columbia; Atlanta, Georgia).

In December 2009, KP and the VHA launched a pilot project in San Diego, California, for electronically sharing EHR files of individuals who receive care from both systems with patient permission (KP, 2009). Now part of the Nationwide Health Information Network Exchange project (NwHIN Exchange, 2011), this type of data sharing not only should improve patient care but also could permit the capture of similar types of data from a larger and more diverse population than those represented by the individual participating systems.

National Vital Statistics System

Most of the data systems reviewed by the committee collect date of death. In particular, the NHSS would serve as the most nationally representative source of data for estimating the committee’s recommended mortality indicator: all-cause mortality rate among PLWHA. As indicated in Chapter 2, the committee selected all-cause mortality for the indicator because of the inherent difficulties in determining and recording in every instance whether deaths among PLWHA were related to the disease or another cause. Mortality rate due to HIV nevertheless may be a useful measure for some purposes. In such cases, the National Vital Statistics System (NVSS) is the best source of data for estimating mortality related to HIV infection. Although some of the data systems examined by the committee, such as NA-ACCORD, record information on cause of death, the NVSS regularly calculates HIV mortality. NVSS operates under the auspices of CDC’s National Center for Health Statistics, which collects vital statistics data through contracts with the registration systems in jurisdictions that are legally responsible for recording vital events, such as births and deaths (NVSS, 2011a). Mortality data from the NVSS provide uniform, nationwide demographic, geographic, and cause-of-death information for individuals who die in the United States (NVSS, 2011b). Standard forms (e.g., death certificate) and model procedures are developed and recommended for nationwide use to promote the collection of uniform national data. The death certificate requires a single immediate (final) cause of death and allows for as many as three underlying causes of death to be listed sequentially (CDC, 2011c). Although reporting errors of various types may occur for cause of death, CDC provides extensive information on writing cause-of-death statements for death certificates (NVSS, 2011c). Preliminary HIV mortality data currently are available for 2009 (Kochanek et al., 2011).

Resource and Patient Management System

Nationally, AI/ANs represent less than 1 percent of PLWHA (between 3,039 and 3,083 individuals in 2009) (CDC, 2012, Commentary, Table 15a). Yet, AI/ANs are disproportionately burdened by the epidemic in several ways. The rate of HIV diagnoses among AI/ANs was 9.7 (per 100,000) in 2010, compared with 6.5 for Asians and 7.3 for whites (CDC, 2012, Table 1a).³⁰ Compared with other racial and ethnic groups, AI/ANs also have one of the shortest timelines from AIDS diagnosis to death (CDC, 2012, Commentary, Table 14a; Hall et al., 2005). Impoverishment and conditions such as alcoholism and diabetes that occur at higher rates among AI/ANs (Chartier and Caetano, 2010; IHS, 2008) may complicate care and adherence to treatment.³¹ The IHS is the federal agency responsible for providing comprehensive health care services to approximately 2.0 million AI/ANs representing 566 federally recognized tribes (IHS, 2012). Most IHS facilities are primary care clinics. Two IHS-funded hospitals together treat the majority of HIV patients in the lower 48 states (Personal communication, Lisa Neel, Indian Health Service, October 13, 2011). Combined with an additional two sites, these facilities account for 61 percent of the IHS HIV/AIDS case load (Personal communication, Lisa Neel, Indian Health Service, October 13, 2011). Although some IHS clinics provide limited HIV care services, most refer their HIV clients to outside providers for HIV care (GAO, 2007; IHS response to IOM request for information, March 28, 2011). As of June 2011, there were 289 HIV/AIDS patients on record at federal IHS health care service sites, of which 224 were “active,” having received CD4 and viral load testing within the preceding 12 months. Patients receiving care in tribally operated and urban Indian health care programs are not included in these numbers due to administrative constraints (Personal communication, Lisa Neel, Indian Health Service, February 29, 2012).

The IHS uses an electronic record system called the Resource and Patient Management System (RPMS) to manage clinical, administrative, and financial information on patients and resources and improve the quality of care provided at federal, tribal, and urban IHS facilities throughout the United States (Cullen, 2006). Data are entered into RPMS by providers during patient visits.³² An optional automated module within RPMS called the HIV Management System (HMS) may be used by HIV care providers and case managers in the IHS system to capture data related to HIV/AIDS and to assist nonspecialist providers with decision making through the use of clinical reminders, provider guidelines, and quality-of-care audit reports. HMS captures HIV-specific information such as date of HIV diagnosis, CDC classifications, and ART status. Lab, radiology, and pharmacy data are available through linkage with RPMS. HMS also may be used to report HIV/AIDS cases to public health authorities through a state surveillance form and report (Cullen, 2006). HMS was first implemented in 2006, and personnel at 12 IHS facilities had been trained in how to use the system by October 2007 (GAO, 2007). In 2009, it was integrated into RPMS (IHS, 2011a, Home, Tech Support), but HMS usage is not mandatory (IHS, 2011d). Although 283 IHS facilities have downloaded HMS as part of the RPMS update, only the two large hospitals that treat the majority of HIV-infected patients are known to use the system and contribute to its ongoing development (Personal communication, Lisa Neel, Indian Health Service, October 13, 2011).

Another component of the RPMS, called the Clinical Reporting System (CRS), is used for national, local, and area monitoring of clinical performance measures. The CRS draws from local RPMS databases to create printed or electronic reports of clinical performance measures, including HIV screening and HIV quality of care, as well as a number of other conditions (e.g., STI, depression, and alcohol screening) that may be relevant for monitoring HIV care (see IHS, 2011c). According to 2011 guidance on the CRS, HIV screening information is reported nationally (IHS, 2011b,c). Reported information includes data on HIV screening among pregnant women and among patients age 13 to 64 with no recorded HIV diagnosis prior to the report period, broken down by gender and age groups, as well as the percentage of patients with documented HIV screening refusals (IHS, 2011b,c). In addition, information is reported on the percentages of patients with positive, negative, or indeterminate test results and on the number of HIV tests given to patients during the report period where the patient was not diagnosed with HIV any time prior to the screening (IHS, 2011b). HIV screening is also incorporated into a syphilis, gonorrhea, chlamydia, and HIV screening measure where diagnosis of one of these STIs prompts screening for the other three (IHS, 2011b).

CRS also captures HIV quality-of-care data for the user population of patients age 13 and older with at least two direct care visits (i.e., visits within the IHS system) during the report period with HIV diagnosis and one HIV visit in the last 6 months (IHS, 2011c). These measures are not reported nationally, however. The quality-of-care measures assessed are (1) the percentage of patients who received the CD4 test only (without HIV viral load) during the report period; (2) the percentage of patients who received HIV viral load only (without CD4) during the report period; (3) the percentage of patients who received both CD4 and HIV viral load tests during the report period; and (4) total numerators 1, 2, and 3 (IHS, 2011c). The first collection period for these variables was July 2010–June 2011 (Personal communication, Lisa Neel, Indian Health Service, October 13, 2011). IHS recently added “newly diagnosed HIV” to CRS, but the measure has not yet been validated (Personal communication, Lisa Neel, Indian Health Service, October 13, 2011).

As with other clinically based EHR systems (e.g., KP, VHA), the IHS collects all of the data needed to calculate the core clinical HIV care indicators for services provided within the IHS (Appendix Tables 3-2a, 3-3a), and the HMS attempts to include historical data about tests and services provided outside of IHS facilities. Similarly, the IHS captures the data pertinent to the additional clinical HIV care indicators (Appendix Tables 3-2c, 3-3c). Even if certain data (e.g., date of influenza vaccination) are not currently captured in the HMS, they may apply to quality measures for other subpopulations (e.g., individuals ages 50 and older, individuals with diabetes) (IHS, 2011c). In addition, data would be captured at the patient level and could be applied to HIV-specific indicators in the future.

Data for calculating the mental health and substance abuse indicators also are captured by the IHS (Appendix Tables 3-2b, 3-3b, 3-3d). Information pertaining to the need for and provision of supportive services may be recorded in the provider narrative section of the EHR. Demographic data collected include age, sex, race, ethnicity, and locality of residence. Data on gender identity and sexual orientation are not routinely collected but might be recorded in the provider narrative section of the EHR.

IHS captures the data necessary for estimating most of the indicators identified by the committee, which could be used to track improvements in HIV care and access within the population of HIV-infected individuals receiving care in IHS facilities. The limited number of IHS facilities providing comprehensive HIV care can affect the size of the HIV-infected population represented in the data system. Facilities that transition from federal to tribal management no longer automatically report data to IHS. In addition, some tribal facilities have moved from IHS to private-vendor EHRs, making data sets incompatible.

Bureau of Prisons Electronic Medical Record

When both state and federal prisons are considered, 21,987 inmates (1.5 percent of total inmates) were HIV infected or had confirmed AIDS as of the end of 2008. Of those inmates, 1,538 were federal prisoners (Maruschak, 2010).

The BOP, which is responsible for ensuring access to health care services for the almost 217,000 individuals incarcerated in federal correctional institutions throughout the United States (BOP, 2011), uses the Bureau of Prisons Electronic Medical Record (BEMR), a point-of-care direct entry web-based system record, to collect health information on inmates housed at 116 federal correctional institutions (BOP response to IOM request for information, April 14, 2011). The BEMR includes a fully integrated pharmacy capability (computerized order entry through prescription administration records, BEMRx) as well as a dental module (DOJ, 2011; Price, 2011). The BEMR tracks CD4 count and viral load for prisoners with HIV/AIDS. Other information contained in the BEMR that may be useful for tracking HIV/AIDS care received by prisoners includes demographic, prescription drug, substance use, and mental health data. The BOP is in the process of enhancing the BEMR by programming key HIV data elements for the extraction and analysis of HIV data that currently are available only in individuals’ records (BOP response to IOM request for information, April 14, 2011).

Due to the much larger number of HIV-infected inmates in state prisons, it would be necessary to track HIV care data from the state inmate population as well in order to gain a more complete picture of HIV care within the U.S. corrections system. Gathering and integrating data from the individual state systems poses a significant challenge, however. A 2007 survey of state electronic health initiatives found that although 22 of 42 states responding had implemented some sort of health information technology use in their state prison systems, only 3 states reported the use of EHRs and/or electronic medical records (Smith et al., 2008). Kentucky had implemented an EHR system across all of its state-operated correctional facilities. Virginia had planned implementation of an EHR system for its correctional facilities, and Washington State was exploring the feasibility of a single, integrated EHR for all of its correctional institutions, including state prisons, city and county jails, and juvenile corrections facilities (Smith et al., 2008).

Integrating EHRs across all types of correctional institutions would provide a rich source of data for tracking the provision of HIV care in the incarcerated population. For individuals already diagnosed with HIV, linkage to and maintaining continuity of care and treatment adherence upon release is a significant challenge. Development of methods for capturing data on the provision of transitional services and associated outcomes for HIV-infected prisoners upon release would be important in this regard (Rich et al., 2011).

Housing Opportunities for Persons with AIDS

The HOPWA program, managed by the U.S. Department of Housing and Urban Development’s (HUD’s) Office of HIV/AIDS Housing, provides funds for housing assistance and other supportive services. The additional supportive services most relevant to the indicators identified by the committee include meals and nutritional services, transportation services, mental health services, and alcohol and drug abuse services, as well as approved health, medical, and intensive care services (HOPWA, 2011a,b). HOPWA programs provide assistance to low-income households with one or more PLWHA along with other members of the household. By the end of FY 2010, HOPWA had provided resources for housing assistance to 60,669 unduplicated households (HOPWA response to IOM request for information, April 4, 2011). The population served by HOPWA is generally representative of low-income PLWHA (HOPWA response to IOM request for information, April 4, 2011).

HOPWA grantees report aggregated data on program performance outcome measures related to maintenance of housing stability, improved access to care and support, and reduced risk of homelessness for low-income persons and their families living with HIV/AIDS. HOPWA Competitive Program grantees submit an Annual Progress Report (APR) and Formula Program grantees submit a Consolidated Annual Performance Evaluation Report (CAPER) measuring performance outcomes.

APR and CAPER (HOPWA, 2011a,b) report information on the number of households with an unmet need for housing assistance,³³as well as the type of subsidy assistance needed. They also report on the number of households served by HOPWA and other funding sources that provide housing assistance and support to PLWHA and their families and the number of households that received other supportive services through HOPWA funds (e.g., meals or nutritional services, transportation, mental health services, alcohol and drug abuse services). Aggregate data are reported on the number of PLWHA who qualified their household to receive HOPWA housing assistance, their prior living situation, the number of other PLWHA who reside with the HOPWA-eligible individuals, and the number of persons not diagnosed with HIV who reside with the eligible individuals. The reports also record the number of HOPWA-eligible individuals and other beneficiaries by race and ethnicity, as well as by sex (male, female) and gender (transgender male to female, female to male) within given age ranges. Additional information reported includes number of households that demonstrated “a housing plan for maintaining or establishing stable on-going [sic] housing”; contact with a case manager or benefits counselor as specified in the client’s individual service plan; contact with a primary health care provider as specified in the client’s individual service plan; access to and maintenance of medical insurance or assistance; and sources of income. The reports also include the number of households receiving assistance by percentage of area median income.

In addition to the aggregate data reported by HOPWA grantees, HUD developed the Homeless Management Information System (HMIS) to store longitudinal standardized individual-level data on persons receiving housing assistance and homeless prevention services through Continuum of Care programs. Program-level data on homeless service usage is reported as well. Aggregate HMIS data provides information about the size, characteristics, and needs of the homeless population at the local, state, and national levels. Although HMIS is not an HIV-specific data system, one of the client-level universal data elements it captures is “disabling condition,” which includes AIDS and AIDS-related conditions.

HUD requires the collection in HMIS of a minimum set of data elements from all individuals receiving homeless assistance and prevention services. These data are required to generate unduplicated estimates of the number and basic demographic characteristics of individuals accessing services and patterns of service use. These “universal data elements” include, among others, name, date of birth, race, ethnicity, gender (male, female; transgender male to female, female to male), presence of a disabling condition, residence prior to program entry, zip code of last permanent address, and housing status (HUD, 2010, pp. 40-63). Additional “program-specific” data elements are variously required from specified homeless assistance programs, including those funded through HOPWA. The program-specific data elements include the amounts and sources of income, if any, in the preceding 30 days; receipt of noncash benefits (e.g., Medicare, Medicaid, Supplemental Nutrition Assistance Program); and information on physical and developmental disability, chronic health conditions, HIV/AIDS, mental health, substance abuse, domestic violence, and destination upon program exit (HUD, 2010, pp. 64ff.). Optional program-specific data elements, not required for APR reporting, include employment status, education, general health status, and pregnancy status (HUD, 2010, pp. 93ff.).

The data from HOPWA’s APR and CAPER and those captured in HMIS provide important information about access to housing and other supportive services for PLWHA, including an assessment of unmet housing needs for HOPWA-eligible households (i.e., those with income below 80 percent of the area median income and documented HIV/AIDS status) (HUD, 2011a). The unmet needs assessment is limited, however, to those individuals identified as being HOPWA-eligible and may not represent the full scope of need for housing assistance among PLWHA. In terms of access to other supportive services (e.g., nutrition or food, transportation), the data are limited to households receiving HOPWA-funded services and do not reflect need or access to services among other PLWHA.

The collection of longitudinal individual-level data in HMIS permits assessment of patterns of housing service usage and outcomes over time. In addition, a number of the client-level data elements captured may allow for linkage or cross-matching to additional information in other pertinent data systems (e.g., MSIS, CCW, Ryan White HIV/AIDS Program).

Enhanced Comprehensive HIV Prevention Planning Project

Launched in September 2010 in response to the NHAS, the ECHPP Project is a 3-year demonstration project funded by the CDC’s Division of HIV/AIDS Prevention. The program targets the 12 MSAs that have the highest AIDS prevalence, cumulatively accounting for 44 percent of cases in the United States (DHAP, 2011).³⁴ Following the NHAS, the overarching goals of the project are to maximize the impact of HIV prevention strategies in these geographic areas, reduce the incidence of HIV infections, improve the quality of HIV care, and reduce HIV health disparities.

The 12 ECHPP grantees are evenly divided between state or territorial health departments and directly funded local health departments. During the first year of the project, each grantee was required to conduct a local situational analysis, taking account of available resources, epidemiologic profiles, priority areas, and cost and cost-effectiveness data for specific interventions and strategies. Based on these analyses, the grantees created a set of goals and strategies that would best aid in the accomplishment of NHAS goals. These enhanced prevention plans, which have been approved by CDC, include interventions and public health strategies designed to prevent new HIV infections and to promote HIV care and treatment. By the end of 2011, the jurisdictions had begun implementing their plans and had submitted funding applications for the second and third years of the project.

Although prevention of new infections is the primary emphasis of ECHPP, seven of its nine required prevention strategies for PLWHA address treatment concerns such as linkage to care, retention or reengagement in care, provision of ART consistent with current guidelines, adherence to antiretroviral medications, STI screening, prevention of perinatal transmission, and linkage to other medical and social services (DHAP, 2011). ECHPP has a comprehensive evaluation plan that incorporates process, outcome, and impact indicators to assess progress in these prevention and treatment areas that will be collected in the 12 MSAs, as well as supplemental data from a subset of the cities (Fisher and Hoyte, 2011).

12 Cities Project

Created by HHS to work in conjunction with CDC’s ECHPP initiative, the 12 Cities Project is a demonstration project designed to promote prevention and treatment of HIV in the 12 cities (MSAs) disproportionately affected by the epidemic through cross-agency collaboration and coordination with state and local health departments and other organizations (HHS, 2011b). Ultimately the lessons learned through the 12 Cities Project will help to improve HIV care in other jurisdictions. The project expands upon the foundation laid by ECHPP, engaging additional federal partners and increasing focus on HIV care and treatment. Whereas ECHPP’s emphasis is on local plans to improve prevention and care in the 12 jurisdictions, the 12 Cities Project emphasizes better coordination of services and funding of federal efforts to improve HIV prevention and care within the jurisdictions and the development of a common set of measures (indicators), in conjunction with streamlining reporting requirements, to evaluate the efforts with respect to the goals of the NHAS.

Motivated by the need to develop common metrics for tracking program outcomes for the 12 Cities Project, HHS undertook a broader effort to develop a streamlined set of cross-agency, core indicators that can be used to monitor the prevention, treatment, and care services of all federally funded programs providing HIV/AIDS services. HHS identified the need for indicators in seven domains: HIV+ diagnosis, early HIV diagnosis, initial linkage to care, sustained engagement in care, initiation of ART, viral load suppression, and housing (Valdiserri and Forsyth, 2011; Personal communication, Andrew Forsyth, Department of Health and Human Services, January 24, 2012). The committee has recommended indicators in each of these areas. Although the jurisdictions included in the 12 Cities Project represent a large percentage of the U.S. population of PLWHA, use of a common set of core indicators across all federally funded HIV/AIDS programs nationwide will generate a more complete picture of HIV care in the United States.

Nationwide Health Information Network Exchange

Developed under the auspices of the Office of the National Coordinator for Health Information Technology, the Nationwide Health Information Network (NwHIN) Exchange is a public-private partnership designed to promote the exchange of health information from patient health records (ONC, 2011). Federal agencies participating in NwHIN Exchange include CDC, Department of Veterans Affairs, and Department of Defense. Nonfederal entities include KP, various hospitals, health information organizations, and state health information exchanges.

Health Care Cost Institute

Another type of data sharing partnership is the Health Care Cost Institute (HCCI), launched in September 2011. The HCCI is an independent, nonprofit entity whose goal is to create a comprehensive database of health care cost and service utilization data to promote and support research on the drivers of escalating health care costs and utilization. HCCI will make available de-identified claims records from four of the largest private health insurers in the United States (Aetna, Humana, Kaiser Permanente, United-Healthcare), as well as Medicare Advantage data from each of those plans (HCCI, 2011). Currently the HCCI database contains more than 5 billion medical claim records from over 5,000 hospitals and 1 million service providers from 2000 through the present (Merrill, 2011). Eventually HHCI plans to add data from additional private insurers, as well as public payers such as Medicaid (Merrill, 2011). This type of cooperative arrangement among private insurers and between the private insurance industry and the public serves as another example of the type of data sharing enterprise that would help to expand the pool of data available to estimate the indicators beyond those available from any individual data system.