Background: Pre-exposure prophylaxis (PrEP) has shown high efficacy in clinical trials, but few observational studies have confirmed its effectiveness when prescribed in real life to users with diverse profiles. This study aimed to assess real-world PrEP effectiveness. Methods: We did a matched, nested case-control study among adult men at high risk of HIV infection between Jan 1, 2016, and June 30, 2020, using data from the French national health data system. Men who were newly diagnosed with HIV infection up to Dec 31, 2020, were individually matched with up to five controls for age, socioeconomic status, place of residence, calendar year, and follow-up duration. PrEP use was characterised on the basis of tenofovir disoproxil fumarate plus emtricitabine dispensing over time. Conditional logistic regression was used to calculate the adjusted odds ratios (ORs) of PrEP use associated with HIV infection. PrEP effectiveness (computed as 1–adjusted OR), was estimated overall, by mode of PrEP use, and by individuals’ sociodemographic characteristics. Findings: Among a total of 46 706 individuals, 256 patients with HIV infection were identified and matched with 1213 controls. PrEP users accounted for 29% of cases and 49% of controls. PrEP effectiveness was 60% (95% CI 46 to 71) overall, reaching 93% (84 to 97) for a high amount of PrEP consumption, and 86% (78 to 92) if excluding periods after PrEP discontinuation. PrEP effectiveness was significantly reduced in people younger than 30 years (26% [–21 to 54]) and in those who were socioeconomically deprived (–64% [–392 to 45]), both of which groups showed low amounts of PrEP consumption and high rates of PrEP discontinuation. Interpretation: PrEP effectiveness appears to be lower in real-world conditions than is reported in clinical trials. Strengthening efforts to improve the monitoring of PrEP compliance will be essential to ensure PrEP effectiveness, especially among young and socioeconomically deprived recipients. Funding: None.
The Lancet Public Health, Volume 7, June 2022,