Analyses

Statistical analyses were performed using the procedures CORR, FREQ, and GENMOD of SAS Version 9.3 (SAS Institute Inc., Cary, NC). Fisher’s exact test for categorical variables and Wilcoxon Mann-Whitney rank-sum tests for continuous variables were used to test for differences between participants with and without foster care history. The relationship between foster care history, four associated syndemic vulnerabilities (group sex participation, being high on drugs/alcohol during sex, history of sexual victimisation, homelessness) and increased likelihood of having one or more HIV/STI diagnosis were examined in a logistic regression using the Poisson link function and Pearson correction for dispersion. Four interaction terms with foster care status were included to explore whether effects of syndemic vulnerabilities differed between foster care alumni and non-alumni. Characteristics that distinguished foster care status at p<.10 in univariate analyses were included as covariates, and age of first consensual sex was treated as an offset variable. Preliminary analyses suggested that linear modeling was adequate, and the logit model showed a good fit, X2=149.47, p=.64. Intra-correlations between explanatory variables did not exceed r=.2 except between mental distress and substance dependency symptoms (r=.3, p<.0001), and collinearity diagnostics indicated no problems. All p values were calculated with two-sided tests.