For HIV incidence, total person-years (PY) were calculated as the time from study entry to the estimated date of HIV seroconversion. With the exception of circumcision, all risk factors were examined as time-dependent covariates. Potential risk factors that defined subgroups of the HIM cohort in which the annual HIV incidence was ≥2 per 100 PY, as a starting point, were identified. It should be noted that the subgroups were not mutually exclusive, as participants could potentially report more than one risk factor in a 6-month period, so that a person could fall into multiple risk factor groups.
Risk factors examined that had been identified in univariate analysis and reported in previously published research from the BGJ398 molecular weight HIM study included any previous nonoccupational post-exposure prophylaxis (NPEP) use , circumcision status , and any of the following in the past 6 months: reported unprotected anal intercourse (UAI) with a known HIV-positive partner,
Selleckchem Bortezomib receptive UAI with a casual partner , having an HIV-positive regular partner , anal sexually transmitted infections (STIs) , use of oral erectile dysfunction medications and recreational drug use . In some cases [27,30], the definition of risk behaviours used in published reports was slightly modified and the data reanalysed to provide a more pragmatic definition of risk groups. For example, in a previous publication, the risk
associated with receptive UAI was stratified by whether or not ejaculation had occurred , but for the current analysis, Janus kinase (JAK) any report of receptive UAI with a casual partner was included. Demographic factors such as age, income and education level were also examined . Other risk factors for higher HIV incidence were identified from other published sources and examined within the HIM cohort. These included number of sexual partners in the past 6 months [32,33,34], alcohol consumption , previous hepatitis B virus (HBV) infection  and injecting drug use . For each risk factor, the HIV incidence was calculated as the number of HIV seroconversions divided by the total PY of follow-up for participants who reported/experienced that risk factor. A stepwise procedure was used to rank the factors described above according to the incidence in subgroups defined by the factor. First, the factor associated with the highest incidence was identified, the HIV incidence was recorded and then participants in the cohort who reported the factor were excluded from further analysis. The incidence in each of the remaining risk factor subgroups was then recalculated in the remaining individuals. On each occasion when HIV incidence in the remaining risk factor subgroups was recalculated, the incidence in that stratum changed after the participants above were excluded.