[Predictive ability of risk assessment tool regarding HIV infection for men who have sex with men]

Objective: To analyze the predictive ability of HIV infection risk assessment tool for men who have sex with men (MSM). Sentinel surveillance data of MSM in Taizhou prefecture of Zhejiang province was used. Methods: MSM involved in AIDS sentinel surveillance program in Taizhou from 2013 to 2017 were included in the study and items listed in the HIV infection risk assessment tool for MSM was revised. Related data on questions and options involved in sentinel surveillance was collected and individual risk scores were calculated. We determined the predictive ability of this tool by comprehensive analyzing the HIV infection status and individual risk scores. Results: A total of 1 944 MSM were included in the study, with an average age of (35.04±13.28)years old. Most of them were recruited from the venues (55.7%) and 48.2% were never married. Among these MSM, HIV infection rate was 12.6%(245/1 944) with the median of risk score as 23.99, versus 20.36 from the HIV negative ones. Significant differences appeared on the risk scores between the target populations that with different HIV status (Mann-Whitney test, P=0.007). According to the principle of decision tree, MSM were divided into two groups according to risk scores: ≤18.66 and >18.66. It appeared that the risk scores were in parallel with the rates of HIV infection (χ(2)=13.102, P<0.001). Results from the multivariate analysis showed that MSM with higher risk score were more likely to be infected with HIV (>18.66 vs. ≤18.66: aOR=1.72, 95%CI: 1.27-2.32, P<0.001). Area under the ROC curve (AUC) for HIV infection was 0.553 (95%CI: 0.516-0.590, P=0.007). At the point of risk score 19.01, Youden's index appeared the maximum, with sensitivity as 0.69 and specificity as 0.43, of this tool. Conclusions: The HIV infection risk assessment tool for MSM developed based on Delphi method can predict the risk of HIV infection in MSM to some extent. MSM with higher risk score seemed likely to be infected with HIV. Items of this tool need to be adjusted for the verification of the tool through cohort studies in the near future.