Point-of-care HIV testing in primary care and early detection of HIV (RHIVA2): a cluster randomised controlled trial

Background: Early HIV detection in primary care is a key challenge. We completed a cluster randomised controlled trial in UK general practices to determine whether point-of-care testing improved early HIV detection. Methods: We randomised 40 of 45 general practices in Hackney, London, UK, to receive a programme promoting rapid HIV testing to newly registering patients, or to usual care. With use of minimisation, and maintaining allocation concealment, practices were randomly allocated to the intervention or control group using the following criteria: practice list size, indices of multiple deprivation score, and male HIV testing rate. We were unable to blind general practices and study participants to allocation. The study was done between April, 2010, and August, 2012. Participants were individuals who were registering at a general practice and who were aged 16 years and older. Those unable to communicate in English who presented without a suitable translator, and known HIV-positive patients, were excluded. General practice staff received education and training in rapid HIV testing (INSTI™ HIV-1/HIV-2 Rapid Antibody Test; bioLytical Laboratories, Canada) for patients at registration or first consultation in general practice. Safe diagnosis and transfer of newly diagnosed cases into specialist care was established with use of quality assurance and electronic data monitoring programmes. The primary outcome was mean CD4 cell count of all patients newly diagnosed as HIV positive in general practice. Primary analysis was by intention to treat, with pre-planned sensitivity analyses, using mixed effects regression analysis to adjust for stratification factors and clustering within practices. This trial is registered, number ISRCTN63473710. Findings: During the study 44 971 and 38 464 patients registered with intervention and control practices, respectively. Their mean age was 35·9 years (intervention group) versus 35·1 years (control group); and 45% in each group were men. Ethnicity was 49% white and 17% black (intervention) versus 57% and 15% (control). Intervention practices offered 11 180 rapid tests, with 4978 (44·5%) accepted (14 tests reactive, 11 confirmed HIV positive). Opportunistic serology testing identified 21 (intervention) and 14 (control) further cases. Of the 32 new diagnoses in the intervention group, 19 (59%) were men, 19 (59%) were black African, and five (16%) were men who have sex with men (MSM). Mean CD4 count was 356 cells per μL (SD 254) in the intervention group versus 270 cells per μL (SD 257) in the control group. The difference in CD4 cell count on a square root scale adjusted for stratification factors was 3·7 (95% CI 1·83–9·40; p=0·186). In a pre-planned sensitivity analysis, excluding seven patients diagnosed via the established NHS antenatal HIV screening programme, mean CD4 counts were 369 cells per μL (SD 262) in the intervention group versus 194 cells per μL (169) in the control group (adjusted difference: 7·5 [95% CI 0·94–14·07]; p=0·025). All patients diagnosed via rapid testing were successfully transferred into specialist care. Interpretation: We show that a programme offering rapid HIV testing at general practice registration led to earlier detection of HIV, and identified cases from non-MSM risk groups. Staff support and quality assurance are paramount for safe implementation of such an intervention. To reduce undiagnosed and late presentation of disease, we recommend that large-scale HIV testing be implemented in UK general practices in high-prevalence settings. Funding: NHS City and Hackney, Department of Health, UK. Contributors: CG had the original idea for the study. WL, HM, CG, SK, DM, SM, SC, KB, SB, JF, GH, and JA designed the study. WL, HM, DM, SC, JA, and CG contributed to general practice staff training and education. WL and HM undertook the quality assurance of the study. SK and NM did the statistical analysis. RA provided advice on ethical aspects of the trial, including data management and data protection. AM completed data quality assurance checks. WL and CG wrote the abstract with input from SK and JA. All authors have seen and approved the final version of the abstract for publication. Conflicts of interest: We declare that we have no conflicts of interest