Principles of Unlinked Surveys

Methods. Unlinked surveys are conducted by using residual specimens collected from eligible survey participants for routine clinical purposes. There is no contact between persons in the surveys and the investigators conducting the surveys. Before HIV testing, demographic and risk information abstracted from routine medical records and intake forms are linked to the residual specimens through a unique study number. The specimens, which otherwise would be discarded, are anonymously tested for antibodies to HIV after the permanent removal of all personal identifiers. Neither the HIV test results nor the information obtained from medical records and intake forms can be linked to individuals. The results of HIV testing provide prevalence data that are unbiased by test-seeking behavior. This information is used to track HIV infection in the population but gives no information on individuals. The surveys are conducted in settings that either provide or offer referral for voluntary counseling and testing so that each person included in the survey has the opportunity to learn his or her HIV status and receive appropriate services for care and counseling for behavior changes.

Ethical considerations. The Office for Protection from Research Risks (OPRR), to which the Secretary of Health and Human Services has delegated the protection of the rights and welfare of human research subjects, has determined that anonymous unlinked surveys are ethical if (1) no interaction takes place with the survey participant solely for the purpose of the surveys and (2) information that may inadvertently identify a person is not retained. In June 1998, the Office of the Director of CDC and the Division of HIV/ AIDS Prevention–Surveillance and Epidemiology, National Center for HIV, STD, and TB Prevention, convened an external panel to conduct an ethical review of anonymous unlinked surveys for HIV in order to address questions that have arisen in recent years concerning the ethics of these surveys. The members of the panel felt that the unlinked surveys are ethical and are necessary to provide unbiased, accurate data on the current status and direction of the HIV epidemic in specific populations and also to provide reliable information for prevention planning and the allocation of resources.

CDC protocols for unlinked surveys are submitted to the CDC Institutional Review Board (IRB) for review and approval. State and local health departments consult with, and submit their protocols for approval to, their local IRBs before conducting the surveys.

Collection and Interpretation of Data

Standardized protocols and laboratory procedures are used for each of the surveys. Data are summarized by geographic area, sex, age, race/ethnicity, and, when possible, by major behavioral risk groups. The participating states and the District of Columbia are divided into four geographic regions as defined by the United States Bureau of the Census: Northeast, Midwest, South, and West. Data for Puerto Rico are included in the state and territory analyses, but not in the regional analyses. All rates in this report are reported as percentages.

Because participating clinics represent a convenience sample rather than a probability sample, clinic data are presented as medians. Ranges of rates are given for each clinic type. Prevalence data from a single site should be interpreted with caution because persons who attended participating clinics and hospitals may not be representative of all persons attending clinics and hospitals in the area. However, the collective data from these surveys suggest geographic patterns of HIV infection.

Laboratory methods. In the unlinked surveys, residual specimens collected for routine medical purposes are tested for HIV type 1 (HIV-1) antibodies by using an HIV-1 or HIV-1/HIV-2 enzyme immunoassay (EIA) screening kit licensed by the Food and Drug Administration (FDA) after all personal identifiers have been removed. Sera that are reactive according to the manufacturer’s instructions are retested in duplicate by using fresh samples from the original unlinked specimen. Repeatedly reactive sera are tested with an FDA-licensed Western blot assay. Software provided to the sites by CDC automatically generates Western blot interpretations from recorded band patterns according to the recommendations of the Association of Public Health Laboratories and CDC.

Behavioral risk categories. Limited information on demographic characteristics and major behavioral risk categories is abstracted from client medical records and intake forms only. For the purpose of this report, behavioral risk groups for HIV infection are classified into three mutually exclusive categories. Men whose medical records indicate that they have had homosexual or bisexual contact are categorized as MSM. Clients who are not MSM and whose medical records indicate that they have ever injected drugs are categorized as IDUs. Persons for whom neither of the previous risk categories is recorded are classified as "others."

STD Clinics

STD clinics serve large numbers of persons at increased risk for HIV infection due to unprotected sex and other high-risk behaviors, such as having multiple sex partners or using drugs. They are important sites for evaluating HIV prevention programs and for monitoring emerging patterns and trends in the epidemic. These STD clinic surveys provide comprehensive HIV prevalence data for MSM and high-risk heterosexuals in the United States. Persons at greatest risk for the sexual transmission of HIV are likely to be those also at risk of acquiring other STDs; thus, surveys in STD clinics provide useful information on the sexual transmission of HIV infection.

Patients are eligible for inclusion in the survey if they have not visited the clinic since initiation of the survey during the current calendar year and if blood is drawn for routine syphilis testing. The annual survey period in each clinic ranged from 3 months to 1 year, depending on clinic size. To ensure statistical validity, Table 1 includes only data from clinics that reported at least 500 eligible specimens collected and tested according to CDC protocol or the clinics that reported at least 200 MSM. Subgroups were analyzed for clinics that collected and tested at least 50 specimens per group. Median percentage positive is reported for each metropolitan area, and ranges are included for those areas with more than one clinic.

Persons attending STD clinics may not be representative of all persons with STDs. Because the prevalence of HIV is very high among MSM and IDUs, the misclassification of only a few of these patients could greatly elevate the measured prevalence among patients who reported heterosexual contact as their only risk. Therefore, prevalence rates among heterosexual persons who reported no other risk must be interpreted with caution.

Drug Treatment Centers

The HIV transmission associated with injection drug use occurs either directly through the sharing of drug injection equipment or indirectly through sexual and perinatal transmission from IDUs. Drug users entering treatment are a subgroup of all drug users that is easily accessible for unlinked prevalence surveys because blood specimens are routinely obtained as part of the medical assessment. Of the 21 drug treatment centers in this analysis, 18 were primarily methadone maintenance or methadone detoxification centers, and 3 were primarily drug-free programs.

Clients eligible for the unlinked prevalence surveys are those entering treatment for illicit drug use who have a routine blood test for purposes other than for HIV testing and who have used illicit drugs at least once during the past 12 months. The data in Table 2 are for eligible clients who reported having ever injected drugs. Only data from centers that reported at least 100 eligible specimens collected and tested according to the CDC protocol are included in the tables. Subgroups were analyzed for centers that collected and tested at least 25 specimens per group. Median percentage positive is reported for each metropolitan area, and ranges are included for those areas with more than one center.

Although the prevalence rates obtained from this survey may represent HIV infection rates within the population of IDUs in treatment, they may not reflect HIV prevalence among all IDUs. If IDUs who are not in treatment programs engage in high-risk behaviors more frequently than those who enter treatment, the prevalence rates obtained from this survey may be an underestimate of HIV in the entire population of IDUs. However, because in-treatment IDUs are in general older than IDUs who are not in treatment, the prevalence rates could be an overestimate of HIV prevalence among all drug users.

Adolescent Medicine Clinics

The HIV prevalence rates among adolescents likely reflect recent infections. Clinics specifically for adolescents and young adults are of special interest because they serve a population that may not be seen at other health-care facilities. These clinics offer a wide range of services, including general primary care, family planning, prenatal care, counseling, STD treatment, and sports physicals. Of the six adolescent clinics in four metropolitan areas that participated in the 1997 survey, three were hospital-based and three were community-based.

Eligible clients are those 13 through 24 years of age who initially visit the clinic during the survey year and from whom a blood specimen is drawn as part of routine clinic procedures. Clients who visit the clinic for HIV testing, for evaluation or treatment of HIV infection, or for follow-up visits are excluded from the survey.

Data from these surveys provide important prevalence information for selected clients in a given clinic. However, it is not possible to generalize those prevalence rates to all adolescents and young adults in the surrounding areas.

Job Corps Entrants

The Job Corps, administered by the U.S. Department of Labor, is an occupational training program for socially and economically disadvantaged youth from rural and urban areas of all 50 states and U.S. territories. Data sent to CDC from the Job Corps provide a system for monitoring the HIV epidemic in this population, which may be at increased risk for HIV infection.

The Job Corps recruits high school dropouts or high school graduates in need of additional education or training in order to obtain and hold meaningful jobs. The results in this report are for Job Corps entrants in the United States and Puerto Rico who were 16 through 21 years of age. All entrants residing at Job Corps centers during their training are counseled and tested for HIV within the first 2 days of residency. Those who test positive for HIV are allowed to continue as residential trainees and receive both medical care and social support. The entrant’s reported home state is the basis of the state-specific rates in this report.

Job Corps entrants may not fully represent the larger population of disadvantaged youth. Although sexual orientation or history of illicit drug use does not constitute a basis for exclusions, current illicit drug users, persons with severe medical or behavioral problems, and persons on supervised probation or parole or incarcerated are excluded. Because HIV testing is mandatory, self-selection for enrollment in the program could either decrease or increase the number of HIV-infected entrants.

Military Applicants

CDC monitors HIV prevalence data provided by the U.S. Department of Defense (DOD) from their screening program for military applicants. Because of the large number of male and female applicants from all areas of the country, this population provides valuable information about the HIV epidemic.

All persons applying for active duty or reserve military service, the service academies, or the Reserve Officer Training Corps must have high school diplomas or the equivalent. Applicants are screened for HIV infection as part of their entrance examinations. HIV-positive applicants are informed of their test results and counseled by a physician and are excluded from military service. On a quarterly basis, DOD sends CDC HIV test results and limited demographic information on all applicants tested. No information is available on behavioral risk factors for HIV infection. The applicant’s reported home state is the basis for the state-specific rates.

In July 1993, the military ended the screening exclusion of MSM. However, because HIV-positive applicants and drug users are not accepted into the military, associated self-selection bias among persons in high-risk categories may occur. Therefore, MSM, IDUs, and others who were already aware or suspected that they were infected with HIV are likely to be underrepresented in the population of military applicants.