Patterns of PrEP continuation and coverage in the first year of use: a latent class analysis of a programmatic PrEP trial in Kenya.

INTRODUCTION: Effective PrEP use is critical for impact, but data are limited on common patterns of continuation and coverage among persons using PrEP in real-world settings.

METHODS: Data are from the Partners Scale-Up Project, a programmatic stepped-wedge cluster-randomized trial to integrate PrEP delivery in 25 Kenyan public health facilities conducted between February 2017 and December 2021. We evaluated PrEP continuation using visit attendance and pharmacy refill records, and computed medication possession ratio to define coverage during the first year of use. Latent class mixture models were used to identify and characterize membership to different PrEP continuation patterns. Multinomial logistic regression was used to examine the association between group trajectories and demographic and behaviour characteristics.

RESULTS: Overall, 4898 persons initiated PrEP, 54% (2640) were female, mean age was 33 years (standard deviation 11) and 84% (4092) had partners living with HIV. PrEP continuation was 57%, 44%, and 34% at 1, 3, and 6 months, respectively. Four unique trajectories of PrEP coverage were identified: (1) one-fourth (1154) exhibited consistent high coverage throughout the year with 93%, 94%, 96%, and 67% continuing PrEP at months 1, 3, 6, and 12, respectively; (2) 13% (682) showed high coverage trajectory throughout 6 months but coverage rapidly declined thereafter (94%, 93%, 63%, and 10% continued at months 1, 3, 6, and 12, respectively); (3) 18.9% (918) exhibited moderate coverage trajectory with 91% of clients refilling PrEP at month 1 but nearly all dropped-off thereafter (37%, 5%, and 4% continued at months 3, 6, and 12, respectively); and (4) 43.8% (2144) exhibited immediate discontinuation trajectory, in which nearly all did not have any subsequent PrEP refill. Overall, being female, older age, having partners living with HIV or of unknown HIV status were statistically associated with better PrEP continuation trajectories compared to the immediate discontinuation trajectory (p <0.05 for all).

CONCLUSIONS: In this analysis of a real-world PrEP implementation programme in Kenya, we found four distinct patterns of PrEP continuation, with one-third of users exhibiting consistent high continuation throughout 12 months and two-fifths with immediate discontinuation patterns. These data may help guide tailored interventions to support PrEP continuation in this setting.