Association between Drug Co-use Networks and HIV Infection:A Latent Profile Analysis in Chinese Mainland

Persons who use drugs are a high-risk subpopulation for HIV infection because more than one drug user often take drugs together,which increases the risk of HIV transmission.For the PWUDs who had tested for HIV,we extracted data on demographic characteristics,drug types,injection drug use behavior,encrypted IDs of co-users,and drug use location from the National Dynamic Management and Control Database for Persons Who Use Drugs between January 2008 and July 2016.We mapped and classified Co-UDs’networks and estimated the risk of HIV infection.Drug co-use networks were formulated on the basis of connections between PWUDs at the same time and location.Latent profile analysis(LPA)and logistic regression models were used to cluster the classes and to estimate the HIV infection risk among them,respectively.A total of 470967 PWUDs were included,among whom 69529 Co-UDs formed 7001 networks,which were categorized into three classes:Class A(n=211,3.0%)had large networks,with more members(median 10.0(IQR:8.0-14.5)),and comprised depressant and stimulant users located in central and eastern China;Class B(n=3770,53.8%)was dominated by depressant users(median 3.0(IQR:3.0-4.0))across the country;Class C(n=3020,43.1%)was dominated by stimulant users(median 3.0(IQR:3.0-4.0))located in central and eastern China.The HIV infection rates were 9.71%,3.33%,and 0.33%in Classes A,B,and C,respectively.Compared with Class A,Classes B and C were associated with decreased odds of HIV infection.Our findings suggest that more attention should be paid to larger networks,with a relatively high HIV infection risk,to enhance prevention and intervention strategies.