Characteristics of patients at ART initiation.

Durability of the initially prescribed antiretroviral therapy (ART) regimen is critical for long-term virologic suppression among people living with human immunodeficiency virus (HIV). However, data on the incidence of regimen modification and its associated factors remain limited in China. We aim to quantify the incidence of initially prescribed ART modification and identify associated baseline factors in China. Treatment-naïve adults (≥18 years) who initiated ART with complete regimen records and have documented follow-up information in Beijing, China, from 2010 to 2020 were included. The primary outcome was initially prescribed ART regimen modification. A multivariable Cox proportional hazards model was applied to evaluate factors associated with modification risk. Of 18,911 participants included, 3,725 (19.7%) participants experienced ART modification over 472,565 person-months of follow-up (PMFU), a rate of 7.9 per 1,000 PMFU. The median follow-up was 32 (IQR 13–36) months. Modification rates peaked in months 0–6: TDF + AZT + NVP and LPV/r + 3TC + AZT exhibited the highest 6-month modification rates (59.4 and 57.7 per 1,000 PMFU, respectively), whereas TDF + 3TC + EFV, the most prescribed regimen, had the lowest early switch rate (8.8 per 1,000 PMFU). In multivariable analysis, baseline white blood cell (WBC) < 4.0 × 10⁹/L and WHO stage II–IV were associated with higher modification risk; missing baseline records of WBC, hepatitis B virus or hepatitis C virus coinfection, and later calendar year of ART initiation were associated with lower modification risk; compared to EFV + 3TC + TDF, LPV/r + 3TC + AZT had the highest modification risk. TDF + 3TC + EFV was the predominant initially prescribed regimen with the highest durability, while LPV/r + 3TC + AZT had the highest modification rate. These findings underscore the need for early ART initiation, comprehensive pretreatment screening, and enhanced early monitoring—especially during the first six months—to optimize regimen selection and minimize unnecessary modification.