Study procedures
Data were collected between November 2013 and January 2014 through a nationwide, anonymous online-survey targeting MSM (the SMA 2013 survey). Participants were recruited for the survey through private messages and banners on several social networking and dating sites for gay men. By clicking on a link or banner the participant was referred to the survey’s entry site, which contained information about the goals and contents of the survey, terms of participation and data privacy. By clicking on a button “I have read and understood the information above” the participant gave his informed consent and was referred to the online questionnaire. At the end of the online questionnaire participants were offered a free test voucher for download. More details including a description of the CHERRIES criteria for the survey are provided in the Additional file 1.
The online survey protocol was evaluated and approved by the ethical review board of the Charité University Clinic in Berlin (EA1/266/13). Suggestions by the data protection office of the federal state of Berlin to improve data protection for survey participants were implemented. Participants had to be at least 16 years old. Although less than 18 years old are considered minors in Germany, the ethical review board accepted inclusion of this age group in the study without any additional specific requirements.
Measures
Testing sites, recency and frequency of testing, and reasons for testing and not testing for HIV
Data on HIV testing history (site of last HIV test, recency of last HIV test and frequency of HIV testing) as well as reasons for testing decisions were collected using closed ended questions. Private practices as testing sites were differentiated into practices of doctors known to be gay or with a large proportion of gay clients and other private practices. In addition to the response options provided about reasons for testing, the questions also contained an open ended option (“What other reason did you have?”). Answers to this open ended question were content analysed and either re-coded into existing response options or coded into new response categories. Reasons for not testing were assessed by two different questions: one direct question was asked to participants who reported never testing for HIV or not having been tested in the previous five years. Another question on reasons for refusal was asked to participants who declined the offer of a free test voucher.
Reported sexual behaviours and perceived risk of HIV infection
Sexual behaviour was assessed by questions on number, gender and type of partners (steady, non-steady); relationship status (single; monogamous relationship; open relationship); condom use for anal sex; risk management approaches (HIV serostatus communication with steady and non-steady partners); meeting places for partners; and the perceived risk of these behaviours. Perceived risk was assessed by an eleven-point scale from 0 to 10. For the analysis, the eleven point scale was reduced to four risk levels: no risk (0–1), low risk (2–4), moderate risk (5–7), and high risk (8–10).
Psychological scales
Internalized homonegativity
Internalized homonegativity is defined as the extent to which gay men agree with negative societal attitudes about homosexuality. To assess internalized homonegativity we used the established scale by Smolenski et al. [11]. This scale contains 8 items (e.g. “Even if I could change my sexual orientation, I wouldn’t”) with a seven-point Likert-type scale (totally agree – totally disagree). Reliability of this scale was good, with a Cronbach’s α of .78. For the analysis we condensed the seven point score to three levels, representing low, middle and high levels of internalized homonegativity.
HIV-related stigma
Stigmatizing attitudes were assessed using a self-constructed ad-hoc scale with six items (e.g. “HIV-positive people are irresponsible”, “I wouldn’t want to be in a relationship with an HIV-positive individual”). Responses used a four-point Likert-type scale (totally agree – totally disagree). The scale’s reliability was good, with a Cronbach’s α of .74. For the analysis we condensed the six point scale to three levels, representing low, middle, or high stigmatizing attitudes towards HIV.
Anticipated HIV stigma
Anticipated HIV stigma can be understood as expectations about the experience of stigma in case of being tested HIV positive. Anticipated HIV stigma was assessed with six self-constructed items (e.g. “My family would be disappointed with me”, “I would get trouble in my job”. Responses were recorded on a four-point Likert-type scale (very likely – very unlikely). Reliability of this scale was very good with a Cronbach’s α of .84. For the analysis we condensed the six point scale to three levels of low, middle, and high stigma anticipation in case of an HIV diagnosis.
More details on the items used in the psychological scales are provided in the Additional file 1.
Demographic and other variables
Demographic characteristics used in the analysis were age, education level (high school or less), type of work (blue collar worker; white collar worker/public official/self-employed; student/trainee), monthly equivalence income (<936€; 936–1,895€; >1,895€) [12], and settlement size (<100,000 inhabitants; 100,000-1,000,000; >1,000,000). Ethnicity or nationality of respondents was not queried.
Other variables used in the analysis were outness towards co-workers/classmates and towards primary health care provider about sexual orientation (less than half know; half or more know; not applicable); gay subculture involvement (frequently visiting social venues; frequently visiting sex venues; frequently visiting both social and sex venues; infrequently/never visiting gay venues – a more detailed description of this variable is given in the Additional file); and having been reached by the national HIV prevention campaign for MSM (IWWIT).
Statistical analysis
The analysed subsample included all men who had answered the question about ever testing for HIV and who had never received a positive HIV test result. For the purpose of this analysis three different HIV testing statuses were defined: recently tested, comprising men who reported testing for HIV within the previous 12 months; distantly tested, comprising men who reported testing for HIV more than 12 months ago; never tested, comprising men who reported never to have tested for HIV.
Associations of testing status with the variables described above were analysed using univariable and multivariable multinomial logistic regression. HIV-testing status was used as the outcome variable, with “Recently tested” set as reference. In a first step, we performed a univariable analysis with the variables age, educational level, occupational status, equivalent-income, settlement size, sexual attraction, outness towards co-workers/classmates, outness towards primary health care provider, gay subculture involvement, internalised homonegativity, HIV-related stigma, anticipated HIV stigma, familiarity with the IWWIT-campaign, relationship status, number of non-steady UAI-partners, and assessment of the own risk. Using a stepwise forward selection approach, where a p-value of <0.05 was considered to indicate statistical significance, a multivariable model was constructed. We also tested for interaction between the variables age and settlement size and the variables outness towards co-workers/classmates, readiness to HIV stigmatization, gay subculture involvement, and relationship status. Interaction terms found to improve the model significantly, using the likelihood-ratio test, were included in the final model.
All analyses were performed using the statistical software StataSE12.