A Prediction Risk Score for HIV among Adolescent Girls and Young women in South Africa: Identifying those in need of HIV Pre-Exposure Prophylaxis

Abstract

Abstract Background In sub-Saharan Africa, adolescent girls, and young women (AGYW) are among those at the highest risk of acquiring HIV. Risk factors for HIV in AGYM are well studied and known in the literature. However, there is need to combine these factors into a single summary measure that could be used in the identification of the AGYW who are more likely to acquire HIV. This study aimed at developing and validating an HIV risk prediction tool for AGYW. Methods We analyzed existing HIV-related data on 4,399 AGYW from South Africa. The HIV risk scores were computed from summing predictor coefficients of the resulting logistic regression model. The performance of the final model at discriminating between HIV infected and non-HIV infected AGYM was assessed using area under the receiver-operating curve (AUC) and measures of discriminative abilities such as predictive values, sensitivity, and specificity. The optimal cut-point of the risk score was determined using youden index. Results The weighted HIV prevalence was estimated at 12.4% (11.7–14.0). Our risk scores ranged from − 1.26 to 3.80 with a mean score of 1.38 and a standard deviation of 0.86. The optimal cut-point was estimated at 1.80 with sensitivity of 62% and specificity of 70%. The prediction model’s sensitivity was 15.19% and specificity of 98.92%. The model’s positive predictive value was 67.42% while the negative predictive value was 88.79%. Our model performed well at predicting HIV positivity with training AUC of 0.770 and a testing AUC of 0.751. Conclusion Our risk score tool has shown good discrimination and calibration at predicting undiagnosed HIV. This tool could provide a simple and low-cost strategy for screening AGYW in primary health care clinics or community-based settings. The risk assessment tool will help service providers identify and link AGYW to PreP services.