HIV screening practices and hospital characteristics in the US, 2009-2010

The Centers for Disease Control and Prevention recommends HIV screening in U.S. health-care settings unless providers document a yield of undiagnosed HIV infections of <1 per 1,000 population. However, implementation of this guidance has not been widespread and little is known of the characteristics of hospitals with screening practices in place. We assessed how screening practices vary with hospital characteristics.

We used a national hospital survey of HIV testing practices, linked to HIV prevalence for the county, parish, borough, or city where the hospital was located, to assess HIV screening of some or all patients by hospitals. We used multivariate logistic regression analysis to assess the association between screening practices and hospital characteristics that were significantly associated with screening in bivariate analyses.

Of 376 hospitals in areas of prevalence ≥0.1%, only 25 (6.6%) reported screening all patients for HIV and 131 (34.8%) reported screening some or all patients. Among 638 hospitals included, screening some or all patients was significantly (p<0.05) more common at teaching hospitals, hospitals with higher numbers of annual admissions, and hospitals with a high proportion of Medicaid admissions. In multivariable analysis, screening some or all patients was independently associated with admitting more than 15% of Medicaid patients and receiving resources or reimbursement for screening tests.

We found that few hospitals surveyed reported screening some or all patients, and failure to screen is common across all types of hospitals in all regions of the country. Expanded reimbursement for screening may increase compliance with the recommendations.