Little is known about characteristics of organizations that predict early adoption of highly active antiretroviral therapy
(HAART) for persons with HIV
To describe characteristics of sites where HIV
care is provided and to assess site characteristics that predict early adoption of HAART.
Cross-sectional analysis of survey data from patients, HIV
physicians, and medical directors.
Patients and Setting:
Participants in the HIV
Cost and Services Utilization Study, a national probability sample of persons with HIV
who received outpatient care in the continental United States during 1996.
Main Outcome Measure:
Rates of exposure to HAART by December 1996.
Nationally, 79% of patients were treated at sites specializing in HIV
sites). Over 90% of patients were cared for by physicians who were experts in HIV
care, either infectious disease specialists (46%) or general medicine experts (45%). Adjusted rates of exposure to HAART by December 1996 varied from 0.02 to 0.79 across sites (mean rate, 0.33). In multivariable models, HIV
specialization (odds ratio [OR], 3.6; P
< 0.001), total patient volume of more than 20,000 visits a year (OR, 2.1; P
< 0.01), and educational level of the zip code in which the site was located (OR, 1.2 for each 10% increase in college education) were associated with higher rates of exposure to HAART. These effects persisted after adjustment for physician HIV
expertise. Site effects were more important than physician effects in explaining rates of exposure to HAART.
In 1996 there were wide variations in rates of HAART use by site of care. Low-volume sites that do not specialize in HIV
care should take measures to ensure that HIV
expertise is available to their patients.