The questionnaire contained:
1. Standard demographic questions: on age, sex, ethnicity, civil status and current occupation.
2. A question on participants' sexual orientation using Kinsey ratings based on sexual experiences [17
]. The respondent was asked to circle any number from 1 to 7 that corresponded to the statement that best described their sexual experiences. These were as follows: 1 entirely heterosexual; 2 largely heterosexual, but with some homosexual experience; 3 largely heterosexual, but considerable homosexual experience; 4 equally heterosexual and homosexual; 5 largely homosexual, but with considerable heterosexual experience; 6 largely homosexual, but with some heterosexual experience and 7 entirely homosexual.
3. Short Form 12 (SF-12): This is a well-validated quality of life questionnaire. We used the 12-item version of this questionnaire that produces separate physical and psychological well-being scores [18
4. The General Health Questionnaire (GHQ): a measure of current mental health problems. Since its development, it has been extensively used in different settings and cultures [19
]. The questionnaire was originally developed as a 60-item instrument but the GHQ-12, a shortened version of the questionnaire, has since been developed. The GHQ-12 is easy to complete and was designed to screen for psychological symptoms in the community. It is not a diagnostic instrument. There is evidence that the GHQ-12 is a consistent and reliable instrument when used in general population samples [24
]. The GHQ asks whether the respondent has experienced a particular psychological symptom or behaviour recently. Each item is rated on a four-point scale (less than usual, no more than usual, rather more than usual, or much more than usual).
5. The four-item CAGE questionnaire [25
], which is a known screening instrument for possible alcohol misuse. We also included one question on current consumption of cigarettes.
6. Questions about numbers of sexual partners in the preceding four weeks in order to assess risk of sexually transmitted infection.
7. Four screening questions about sexual experiences under the age of 16 years [26
]. These concerned a) someone trying to or succeeding in having sexual intercourse with them b) touching, grabbing, kissing or rubbing up against them in public or private c) taking photographs of them naked or exhibiting parts of their body to them or performing a sex act in their presence and d) perpetrating oral sex or anal intercourse on them.
8. The brief sexual function questionnaire for men [27
] and a modified version for women [29
]. This questionnaire collected information on sexual activity (i.e. masturbation, oral sex and sexual intercourse) in the last month and also enabled us to make a sexual dysfunction diagnosis according to the 10th
Edition of the International Classification of Diseases [30
] (lack/loss of sexual desire, sexual aversion, and dyspareunia in both sexes; arousal and orgasmic disorders and vaginismus in women and erectile and premature and retarded ejaculatory problems in men) and assessed their satisfaction with their sex life.
9. Total family practice consultations over the preceding two years were collected for those participants who allowed us access to their clinical records.
Definition of sexual orientation and physical, psychological and sexual problems
a) Sexual Orientation:
There is still no universal agreement on how to define sexual orientation [31
]. Thus we approached our analysis on the assumption that society is more accepting of people who report largely heterosexual, rather than largely homosexual, experiences. Hence we classified people responding to rating 1 on our sexual experiences scale as unequivocally heterosexual. Similarly, we assumed that participants who indicated their experiences had been largely heterosexual or equally heterosexual and homosexual (ratings 2 to 4) were primarily
identifying with heterosexuality (despite sexual experiences with both sexes) and classified them as bisexual. Finally we assumed that those who indicated that homosexual experiences were a large or entire part of their sexual lives (ratings 5 to 7) were likely to be primarily gay or lesbian and were classified accordingly.
b) Poor physical function (quality of life) was defined as those scoring below the 25th centile of the physical function subscale score of the short form12 questionnaire.
c) Psychological distress
The GHQ-12 provides a total score of 36, based on the Likert scoring of 0-1-2-3 or 12 based on a bi-modal (0-0-1-1). We used the latter as it is a more common scoring for psychological symptoms and defined those scoring three or more as likely to have significant psychological distress [20
d) Problems with alcohol use were defined as those scoring two or more on the CAGE questionnaire.
e) Current smokers were defined as those that admitted to being smokers at the time of interview.
f) Sexual experiences in childhood were defined as unequivocal in participants who gave affirmative answers to at least three of the four screening questions.
g) Sexual function problems
: were defined using responses from the brief sexual function questionnaires. In order to meet conservative criteria for a clinical ICD-10 diagnosis (F52.0 to F52.6) the problem needed to be present all or almost all of the time for each diagnosis [16
]. However, to account for variation in sexual practices in same sex relationships, we widened the sexual situations in which arousal and orgasmic disorders might occur to include masturbation and oral sex. As previously described [16
], we examined any sexual problem reported without requiring that the participant have a sexual partner or report sexual intercourse in the preceding four weeks.
h) Consultations: people were classed as high consulters if their total consultation rate over the previous two years exceeded the 75th centile for the study population.
We examined the data using descriptive statistics. Differences in the demographic, sexual activity, and health outcome data between gay/lesbian, bisexual and heterosexual classified people were examined using the Chi squared statistic for dichotomous and analysis of variance for continuous data. We then explored the influence of age, civil status, ethnicity and current employment on sexual orientation and the primary outcomes of interest. These were physical quality of life (as measured on SF-12) and psychological distress (on the GHQ-12), sexual function, number of sexual partners in the previous month, childhood sexual experiences, CAGE scores, smoking status and consultation rates. Only civil status (married or living with a partner versus the remainder) and ethnicity (white versus non white) were found to be associated both with sexuality and the health outcomes at a significance level of 10% or less. Thus, we adjusted the analysis of each health factor against our classification of sexual orientation for each of these confounders by fitting logistic models and comparing their fit using the likelihood ratio. We report the p values to indicate whether the adjustments made a significant difference to the model. People in the heterosexual group were used as the index population on the grounds of their overwhelming majority. Each multivariate analysis was conducted separately for men and women. We analysed the data using SPSS version 10 and Stata version 7.