The validity of clinic-based samples in approximating out-of-care HIV-infected populations

Tremendous strides have been made in the diagnosis and treatment of human immunodeficiency virus (HIV); perhaps now the largest barrier to controlling HIV is retaining those diagnosed in care. Data on out-of-care populations are needed to develop effective retention methods, yet obtaining these remains methodologically challenging due to inherent difficulties in sampling. The purpose of this study was to evaluate whether individuals identified by two sampling methods commonly used to approximate out-of-care populations were significantly different from a sample of newly identified out-of-care persons. We compared medical records of 345 out-of-care persons identified by a novel population-based health information exchange who had not received CD4 or viral load monitoring in >1 year with: medical records from a randomly selected, time-matched sample of 488 HIV-infected persons with at least one HIV care visit in the past 5 years, and interviews with 382 participants from a time-matched clinic-based convenience sample. Newly identified out-of-care persons were significantly different from both proxies with respect to demographic, clinical, and utilization characteristics, suggesting that samples of in-care proxy persons are inadequate to describe those not engaged in care. Novel approaches to sampling out-of care populations are urgently needed in order to better understand these populations and ways to improve retention and slow the HIV/AIDS epidemic.