Multivariable logistic regression: UAI | ||||||||||
---|---|---|---|---|---|---|---|---|---|---|
Prevalence of UAI | Overall sample | Subgroup analysis: MSM outside of North America and western Europe | ||||||||
Number of syndemic conditions* | No | (%) | Yes | (%) | aOR†‡§ | 95%CI¶ | p Value¶ | aOR††† | 95% CI | p Value |
None (n=1305) | 869 | (66.6) | 436 | (33.4) | 1.00 | – | – | 1.00 | – | – |
One (n=1661) | 960 | (57.8) | 701 | (42.2) | 1.44 | 1.23 to 1.68 | <0.001 | 1.45 | 1.23 to 1.71 | <0.001 |
Two (n=783) | 402 | (51.3) | 381 | (48.7) | 1.89 | 1.51 to 2.36 | <0.001 | 1.91 | 1.53 to 2.38 | <0.001 |
Three or more (n=185) | 90 | (48.6) | 95 | (51.4) | 2.03 | 1.43 to 2.89 | <0.001 | 2.32 | 1.76 to 3.05 | <0.001 |
Multivariable logistic regression: self-reported HIV infection | ||||||||||
---|---|---|---|---|---|---|---|---|---|---|
Prevalence of self-reported HIV infection | Overall sample | Subgroup analysis: MSM outside of North America and western Europe | ||||||||
Number of syndemic conditions* | No | (%) | Yes | (%) | aOR†** | 95%CI¶ | p Value¶ | aOR†‡‡ | 95% CI | p Value |
None (n=1305) | 1130 | (86.6) | 175 | (13.4) | 1.00 | – | – | 1.00 | – | – |
One (n=1661) | 1340 | (80.7) | 321 | (19.3) | 1.67 | 1.24 to 2.26 | <0.001 | 1.39 | 1.01 to 1.90 | 0.042 |
Two (n=783) | 615 | (78.5) | 168 | (21.5) | 2.02 | 1.44 to 2.85 | <0.001 | 1.70 | 1.14 to 2.53 | 0.010 |
Three or more (n=185) | 146 | (78.9) | 39 | (21.1) | 2.35 | 1.31 to 4.21 | 0.001 | 1.84 | 1.13 to 2.98 | 0.014 |
*Conditions include: illicit substance use, homeless/unstable housing, violence due to sexual orientation, sexual stigma (ie, external homophobia) and depressive symptoms. Substance use was defined as any injection or non-injection substance use within 12 months; sensitivity analyses excluding injection drug use (prevalence=1.29%) in definition of substance use within syndemic measure do not significantly change results of models. Depression was measured using the validated patient health questionnaire (PHQ-2) scale, which was dichotomised using the established cut-off of 3 for symptomatic depression.11 Sexual stigma was measured with a seven-item scale that evaluated perceptions of homophobia,10 which had high internal validity (Cronbach α=0.85) and featured questions with Likert-like response sets ranging from 1 to 5 (corresponding with lowest to highest perceived homophobia, respectively). Responses across items were averaged to calculate the mean score for sexual stigma, then dichotomised as ‘high’ for participants with mean score ≥4. Current housing status was dichotomised as being homeless versus otherwise. Violence for being perceived as MSM, which assessed whether participants or someone they personally know have been physically assaulted, threatened or blackmailed for being perceived as MSM in the past 12 months, was dichotomised as any violence versus otherwise.
†aOR=adjusted OR from multivariable model, holding age, declared sexual orientation, education, racial/ethnic minority status and region of participants constant, while accounting for clustering within country.
‡Evidence of linearly increasing relationship (test of log-linear trend p value <0.001); no evidence of departures from log-linear trend (test for quadratic trend p value=0.14).
§Sensitive analyses including HIV-related stigma from healthcare providers as a syndemic condition produced similar results showing increasing odds of UAI associated with greater number of syndemics: aOR of UAI among those with one, two, and three or more syndemics were 1.31 (95% CI 1.16 to 1.48), 1.77 (1.50 to 2.08) and 2.27 (1.85 to 2.78), respectively; test of log-linear p value <0.001; test for quadratic trend p value=0.88.
¶Bonferroni-adjusted.
**Evidence of linearly increasing relationship (test of log-linear p value < 0.001); no evidence of departures from log-linear trend (test for quadratic trend p value=0.14).
††Evidence of linearly increasing relationship (test of log-linear trend p value <0.001); no evidence of departures from log-linear trend (test for quadratic trend p value=0.42).
‡‡Evidence of linearly increasing relationship (test of log-linear p value=0.011); no evidence of departures from log-linear trend (test for quadratic trend p value=0.46).
MSM, men who have sex with men; UAI, unprotected anal intercourse.