Effect Of An HIV Risk Assessment Tool To Promote HIV Testing Among Men Who Have Sex With Men In Beijing,China:an Online,randomized,controlled Trial

Background:Men who have sex with men(MSM)are seriously affected by Human immunodeficiency virus(HIV)in the world.Current studies pointed out that low risk awareness is a key factor hindering HIV test among MSM.Researchers in the world developed and validated HIV risk assessment tools to provide MSM easy,quick and accurate equipments for self-assessment of HIV risk.In 2017,a HIV infection risk assessment tool for MSM was developed in China.Until now,there is no research to externally validate the tool’s prediction performance for HIV infection and to evlaute its ability to promote HIV testing frequency among MSM.Objectives:To externally validate the HIV infection risk assessment tool in an independent cohort study of MSM in Beijing,China;To evaluate the effect of the HIV infection risk assessment tool on promoting HIV testing frequency and reducing proportion of UAI by an online-based randomized,controlled trial.Methods:Part I:We validated the HIV risk assessment tool with data from an independent cohort study in Beijing Youan hospital from January 2009 to December 2016,the cohort recruited HIV negative adult MSM living in Beijing,China.Area under curve(AUC)of the receiver operating character curve(ROC)was used to quantify discrimination performance;calibration curve and Hosmer-Lemeshow statistic were used for calibration performance.Part Ⅱwas an online randomized,controlled trial.In October 2017,participants were recruited by a geosocial networking(GSN)application(app)if they satisfied the following inclusion criteria:men who aged>/＝18 years old,resided in Beijing,self-reported anal or oral intercourse with a man in past 6 months,and self-reported HIV negative/unknoxwn status.When participants finished the baseline survey,they were randomly assigned into 3 groups(1:1:1),namely Group l.Group2,and Control.Group 1,Group2,and Control received routine education.In addition,Groupl received HIV risk assessment+ automatic tailored feedback based on risk assessment items at baseline and month 6,and Group2 only received HIV risk assessment without feedback.Participants were followed at 1 month,3 month,and 6 month after randomization.HIV tests within one year after randomization in the app-based clinics were also recorded.The Primary outcomes in the trial were self-reported proportions of HIV self-testing,proportions of HIV facility-based testing and UAI proportion during follow ups,the secondary outcomes were change of HIV infection risk scores from baseline to month 6 after randomization,and mean number of HIV tests over 12 months in the app-based testing clinics.Multilevel logistic models were used to evaluate HIV self-testing proportion,HIV facility-based testing and proportion of UAI in the three groups;zero-inflated Poisson regression was used to evaluate the effect on number of HIV tests within I year of follow ups among the three gourps;and t test was used to evaluate changes of HIV infection risk scores fi-om baseline to month 6 after intervention between Group 1 and Group2.Results:Part I,external validation showed AUC was 0.63(95%CI:0.60-0.67,P<0.001),with sensitivity 0.71 and specificity 0.54.The goodness-of-fit Hosmer-Lemeshow test showed there was no statistical difference between observed and expected proportion of HIV sero-conversion(X2=4.55,P=0.80).Part Ⅱ,in October,2017,9280 participants were enrolled by a GSN app,and were randomly assigned into Group 1(3028),Group2(3065),and C.ontrol(3187).3130 participants who finished at least one follow up were included in the final analysis.At 6-month after randomization,the proportion of HIV self-testing were 37.07%in Groupl,34.07%in Group2,and 38.1%in Control;the proportion of HIV facility-based HIV testing were 16%in Group 1,17.03%in Group2,and 14.55%in Control;the proportion of UAI in Groupl was 26.67%,30.22%in Group2,and 30.95%in Control;multilevel logistic regression showed there was no statistical difference of self-reported HIV self-testing proportions,self-reported facility-based HIV testing,and self-reported UAI proportions among the three groups at follow ups of month 1,month 3 and month 6.At baseline,the mean HIV infection risk scores in Group 1 and Group2 were 4.03(95%CI:3.81-4.24),and 3.48(95%CI:3.26-3.71),and at 6 month after randomization,the mean risk score was 3.72(95%CI:3.49-3.95)in Group1 and 3.61(95%CI:3.38-3.84)in Group2,the effect size cohen’s d was-0.21(95%CI:-0.45,0.03),the risk score change between the two groups showed statistically difference(P=0.01).There were 391 tests in the app-based clinic over 12 months(mean 2.51 tests per person)in Groupl,352 tests(mean 2.01 tests per person)in Group2,and 295 tests(mean 1.72 tests per person)in Control;participants in Groupl had an increased mean number of HIV tests within 12 months(incident rate ratio,IRR= 1.32;95%CI:1.04-4.58,P=0.01).Conclusions:This study indicates that the prediction performance of the HIV infection risk assessment tool developed in China is acceptable.HIV risk assessment tool and tailored feedback increased numbers of HIV tests among MSM in Beijing,China over 12 months,and reduced HIV infection risk scores among MSM who received HIV infection risk assessment +automatic feedback.We should consider incorporate HIV infection risk assessment and tailored feedback into routine HIV preventions among MSM.