Abstract Body

Background

Monitoring implementation, effectiveness, and longer term outcomes of long-acting injectable antiretroviral therapy (LAI ART) in health systems and diverse populations will be enhanced if critical data, including initiation, injection times, and discontinuation, can be reliably ascertained from electronic health records (EHR).

Methods

Among persons receiving HIV care at a large clinic in the southeastern US from 1/2021-5/2024, we identified those initiating LAI cabotegravir/rilpivirine, their first injection date, and LAI discontinuation from clinic records. We used EHR procedure (HCPCS code J0741), lab, and encounter data to characterize the frequency of injections, viral suppression (<50 cp/mL), and clinic visits after LAI initiation. We also validated use of EHR data to identify LAI initiators, first injection date, and discontinuation against clinic records.

Results

Among 1951 patients, 138 initiated LAI ART per clinic records. LAI ART initiators were 75.4% cisgender men, 64.5% non-Hispanic Black, and a median of 42 years old. At initiation, 84.8% were virally suppressed (VS) and median viral load (VL) of patients virally not-suppressed (VNS) was 16,474 cp/mL (IQR 78, 50,465). Median time from LAI prescription to injection was 26 days (IQR 16, 47). Procedure codes were 100% sensitive and 99.9% specific to identify LAI ART initiators, and 84.1% sensitive and 100% specific to correctly classify a first injection date (+/- 7 days). Outpatient prescription data alone were 92.8% sensitive and 98.1% specific to identify LAI ART initiators. In the 6 months after first injection (N=111 with sufficient follow-up), patients had a median of 4 clinic visits (IQR 3, 5). Of these 111, 96.8% of those VS at initiation remained VS, and 76.5% of those VNS at initiation became VS. Per clinic records, LAI ART was discontinued in 30 (21.7%) of all 138 initiators. A 90-day gap in injection procedure codes was 70.0% sensitive and 91.4% specific to identify persons who discontinued. Among persons with >90-days between injection procedure codes, 6.9% were VNS (median VL 2192 cp/mL) at next available assessment. 

Conclusions

Most patients remained on LAI ART with favorable clinical outcomes, regardless of suppression at initiation. EHR data identified LAI ART initiators with high accuracy; lower accuracy in identifying initiation date could affect analyses of LAI ART exposure, presumed clinical effectiveness, and implementation outcomes, and should be considered in future studies.