A Four-Item Risk Score to Target Acute HIV Infection Testing Among Men Who Have Sex With Men in Indonesia: Development and Validation in the INTERACT Prospective Cohort.

Abstract

Introduction: Indonesia has an escalated HIV epidemic among key populations, especially men who have sex with men (MSM). Diagnosis and immediate treatment of acute HIV infection (AHI), the earliest phase with the highest transmission risk, is beneficial for individual health and can reduce onward transmission. To inform whom to test for possible AHI using targeted, risk-stratified HIV-PCR testing, this study evaluated the performance of the validated, seven-item Amsterdam AHI risk score among Indonesian MSM, and developed a locally optimized score.

Methods: We used the INTERACT prospective cohort of MSM (≥16 years) attending sexual health clinics in Jakarta and Bali (May 2023-February 2025) who were tested with add-on Xpert HIV-PCR (Cepheid) if their HIV antibody rapid testing was negative or inconclusive. We used generalized estimating equation models to generate risk scores, combining symptoms, risk factors and socio-demographics. The optimized risk score was internally validated using bootstrap resampling. We calculated area under the curve (AUC), sensitivity and specificity (ISRCTN41396071).

Results: Among 1887 individuals, 20 were diagnosed with AHI, and 1867 tested AHI negative across 3446 test visits. The Amsterdam score yielded an AUC of 0.82 (95% CI 0.75-0.90) with a sensitivity of 85.0% (64.0%-94.8%) and a specificity of 59.2% (57.5-60.8). The optimized risk score included one symptom (fever <2 weeks), one risk factor (condomless receptive anal intercourse <6 months) and two socio-demographic characteristics (age 16-30 years, not having received higher education), and achieved an AUC of 0.91 (0.87-0.96) with a sensitivity of 100% (83.9-100) and a specificity of 65.3% (63.6%-66.8%). Internal validation yielded an AUC of 0.86 (0.67-0.97). Applying this risk score would classify 35.1% of MSM as eligible for add-on HIV-PCR testing, identifying 83.9%-100% of individuals who have AHI.

Conclusions: This four-item risk score of easily collected variables can facilitate efficient AHI detection in high-yield clinic settings, enhancing opportunities for HIV prevention. In the Indonesian context, younger MSM with lower educational attainment were particularly vulnerable to AHI.