Background: Routine, voluntary testing of hospital patients for the human immunodeficiency virus (HIV) has been proposed in order to identify those with early HIV infection in a setting where there is ready access to counseling, appropriate clinical referral, evaluation, and therapy. We studied the pattern of HIV infection among patients in 20 U.S. hospitals, in order to evaluate possible national strategies for the routine, voluntary HIV counseling and testing of hospital patients.
Methods: Blood specimens remaining after clinical use from a systematically selected sample of patients at 20 hospitals in 15 U.S. cities were tested anonymously for antibody to HIV type 1 (HIV-1). Multivariate regression was used to determine which variables best predicted HIV seroprevalence in individual hospitals. Using these data, we estimated the number of HIV-positive patients in all U.S. hospitals and considered the efficiency of routine counseling and testing in different subgroups of patients and hospitals.
Results: From September 1989 through October 1991, 9286 of 195,829 specimens (4.7 percent) were positive for HIV-1 in the 20 hospitals. The seroprevalence of HIV at these institutions ranged from 0.2 percent to 14.2 percent. Among HIV-positive patients, 32 percent had symptomatic HIV infection or the acquired immunodeficiency syndrome (AIDS) at the time of admission or evaluation. In the 20 hospitals, HIV seroprevalence was 10.4 times (95 percent confidence interval, 8.8 to 12.0) the AIDS-diagnosis rate (the annual number of patients with new diagnoses of AIDS per 1000 discharges in 1990). In a multivariate model that included 13 hospital-specific variables, only the AIDS-diagnosis rate was associated with the hospital-specific HIV-seroprevalence rate (P less than 0.001). Using these data and the AIDS-diagnosis rates for all U.S. acute care hospitals, we estimated that 225,000 HIV-positive persons were hospitalized (95 percent confidence interval, 190,000 to 260,000) in all 5558 such hospitals in 1990, including 163,000 persons presenting with conditions other than HIV or AIDS (95 percent confidence interval, 130,000 to 196,000). In 1990, in 593 U.S. hospitals with AIDS-diagnosis rates of 1.0 or more per 1000 discharges, HIV testing of patients 15 to 54 years old (3 million patients, or 12.0 percent of all patients in U.S. acute care hospitals) would have identified an estimated 68 percent of all HIV-positive patients (110,000 patients) who were admitted with conditions other than symptomatic HIV infection or AIDS.
Conclusions: We estimate that about 225,000 HIV-positive persons were hospitalized in 1990, of whom only one third were admitted for symptomatic HIV infection or AIDS. Routine, voluntary HIV testing of patients 15 to 54 years old in hospitals with 1 or more patients with newly diagnosed AIDS per 1000 discharges per year could potentially have identified as many as 110,000 patients with HIV infection that was previously unrecognized.