This Web page has been archived on the Web.
2010 No R-196
Dianne Zakaria
Jennie Mae Thompson
&
Frederic Borgatta
Correctional Service of Canada
July 2010
The authors would like to thank the National Inmate Infectious Diseases and Risk Behaviours Survey Working Group for their guidance and insightful feedback throughout questionnaire design and report development (Jacqueline Arthur, Katherine Dinner, Marie-Line Gilbert, Emily Henry, Gayatri Jayaraman, Rhonda Kropp, Tammy Maheral, Marissa McGuire, Mary Beth Pongrac, Jonathan Smith, Greg Springer, and Jill Tarasuk). The support, cooperation and dedication of the National Senior Project Manager (Heather Lockwood), the Regional Survey Coordinators (Kimberley Andreassen, Michelle Beyko, Teresa Garrahan, David Lewis, and Hélène Racicot) and institutional survey coordinators (Tim Ankers, Cathy Ball, Bob Barkley, Louise Barriage, Gaston Bélanger, Réjean Bérard, Diane Bergeron, Sherry Blakeney, Laura Bodvarsen, Lori Boss-Greenhow, Michèle Boutin, Celeste Bowes-Koep, Randall Breaton, Pam Briar, Marsha Brown, Bev Bruce, Joan Christianson, Pénélope Cossette, Shelley Crawford, Solange Cyr, Johanne Demètre, Micheline Désilets, Penny Drury, Mary Ann Dundas, Annette Dupuis, France Duquet, Claire Erkan, Donna Fillmore, Linda Fumerton, Debbie Fury, Lyne Giroux, Carla Grace, Sue Groody, Derek Hutchings, Kevin Jean, Pat Jones, Virginia Jugo, Wally Klein, Dan Larocque, Luc Lavigne, Sylvain Lefebvre, Gail Lévesque, Cherie Maceachern, Daniel Major, Alison Martin, Penny Martin, Dorothy McGregor, Mark Noon-Ward, Tim O'Hara, Enid Oke, Viateur Perreault, Christian Rivest, Johanne Roy, Eva Sabir, Holly Samuels, Suzanne Scott, Dale Shackleford, Marg Smith, Natalie Soroka, Stéphanie Fournier, Jeff Strange, Paul Taylor, Tana Taylor, Brenda Tilander-Masse, Liza Trohan, Darlene Turk, Mario Veilleux, Vicki Vervynck, Katherine Visser, Lorena Watson, Marilyn Winters, and Carol Young) were integral to survey implementation and data collection. Finally, none of this work would have been possible without the participation of Correctional Service of Canada management, federal penitentiary staff and inmates.
Key words: risky sexual behaviour, exchange-sex, unprotected casual sex, sex with a partner of the same sex, sexually transmitted infections (STI), STI rates, inmate survey
Previous research indicates that Canadian federal inmates, particularly women, are a high-risk group for sexually transmitted infections (STIs), but information about incarceration characteristics and sexual behaviours most strongly associated with these infections in correctional systems is sparse. To address this and other deficiencies in the literature, in 2007 the Correctional Service of Canada (CSC) conducted the National Inmate Infectious Diseases and Risk Behaviours Survey (NIIDRBS), a self-administered paper questionnaire completed by a large sample of Canadian federal inmates (n = 3,370). This report presents NIIDRBS data on the rate of reported STIs since admission to Canadian federal prison. Further, it examines associations between reporting an STI since admission and several incarceration characteristics and sexual behaviours.
The overall rate of at least one reported STI since admission to federal prison was 158 per 10,000 person-years served. Comparisons across sub-groups revealed elevated rates in women and Aboriginal males. Specifically, the rate of reporting an STI since admission was 4.5 times greater in women than men (669 vs. 150 per 10,000 person-years) and 1.6 times greater in Aboriginal males than non-Aboriginal males (222 vs. 136 per 10,000 person-years).
Among men, several incarceration characteristics and in-prison sexual behaviours were associated with reporting an STI since admission to federal prison. First, the odds of reporting an STI were about two times greater among males in maximum security compared to males in minimum security. Second, after serving a total of eight years cumulatively, the odds of reporting an STI among males appeared to consistently increase over time, plateauing after 14 years. After serving more than 16 years, the odds of reporting an STI had increased by a factor of 1.8 relative to those serving two years or less. Last, the odds of reporting an STI were approximately five times greater among males reporting exchange-sex (i.e., a transaction involving the exchange of sex for money, works, rigs, drugs or goods) compared to those who did not. Only 1.8% of the male population, however, reported this high-risk activity during the past six months in prison. Among recently sexually active women, reporting an STI since admission was associated with only one in-prison sexual behaviour. Specifically, those reporting an STI since admission were 1.6 times more likely to have unprotected sex with a casual partner than those not reporting an STI (56% vs. 35%).
With respect to community sexual behaviours, only one remained significantly associated with reporting an STI since admission among men: the odds of reporting an STI were about 3.4 times greater among those reporting sex with males compared to those who did not. Among women, several sexual behaviours in the community were significantly associated with reporting an STI since admission. First, the odds of reporting an STI were approximately three times greater in those reporting sex with females compared to those who did not. Second, the odds of reporting an STI were approximately 2.4 times greater among those being paid for sex compared to those who were not. Last, the odds of reporting an STI were 85% lower among those reporting unprotected sex with a regular partner compared to those who did not. This last finding is counterintuitive. Unprotected sex is generally considered a risk-factor for STIs but our finding indicates it is protective. Although further research is necessary to validate this finding, it does reinforce the importance of distinguishing between unprotected sex with casual partners and regular partners.
Presently, all of the identified sexual risk-behaviours are screened for in the intake health status assessment at CSC. Healthcare professionals can use this assessment to identify individuals at elevated risk for STIs while incarcerated and thus more likely to benefit from regular STI testing at CSC. Since cumulative time served and security level were also associated with an increased risk of reporting an STI since admission among men, education, screening and testing needs to continue beyond admission, particularly among high-risk subgroups, and access to harm-reduction items (i.e., condoms, dental dams, and lubricant) needs to be monitored across security levels.
The primary limitations of this research arise from the cross-sectional design of the NIIDRBS and its reliance on self-report. To overcome these limitations, it may be useful in future research to test all inmates using biosamples (e.g., urine tests) at admission and regular follow-ups to maximize the accuracy of estimating the date of infection and recalling risk-behaviours.
STIs can cause serious health and reproductive complications, such as cancer, central nervous system and cardiovascular system damage, blood infections, internal organ damage, arthritis, pelvic inflammatory disease, epididymitis, infertility, ectopic pregnancy, and maternal-infant transmission of the STI. Because STIs are commonly asymptomatic, they frequently go undetected and untreated. Consequently, transmission can unknowingly continue through unprotected sexual activity (Public Health Agency of Canada, 2007b, 2009, n.d.c; Centers for Disease Control, 2005, 2006a, 2006b). Moreover, the risk of acquiring and transmitting human immunodeficiency virus (HIV) through sexual contact is increased in the presence of other STIs (Public Health Agency of Canada, 2007a, 2007b, 2009). Although HIV is an STI, the risk of self-reported HIV since admission to a federal penitentiary is addressed in another report (Zakaria, Thompson, & Borgatta, in pressa). Consequently, this report focuses on STIs apart from HIV.
A comparison of recent STI surveillance data for the Canadian general population (Public Health Agency of Canada, n.d.d) and Canadian federal inmates (Correctional Service of Canada, 2008) indicates a greater incidence rate of STIs among Canadian federal inmates (see Table 1). Table 1 also indicates that the inmate to general population differential is greater among females than males; and, that gender differentials are not consistent across the community and prison environments. Specifically, in Canadian federal prisons the rate of chlamydia, gonorrhoea and syphilis is consistently higher among women than men. In the Canadian general population, however, the rate of gonorrhoea and syphilis is greater among males than females.
Characteristics associated with an increased risk of STIs include:
Research involving inmates has found associations between incarceration characteristics and sexual behaviour. American research found that men in maximum security had more sexual encounters with other men, a higher overall interest in sexuality (i.e., higher overall disposition towards sexual stimuli), and a less homophobic attitude compared to men in minimum security (Garland, Morgan, & Beer, 2005). Similarly, Canadian research involving federal male inmates found that the proportion of men reporting sex with another inmate increased with security level from 3% in minimum to 7% in maximum. Further, among men reporting sex with another inmate, the proportion reporting condom use decreased from 51% in minimum security to 30% in maximum security (Price Waterhouse, 1996). Consequently, one might expect the risk of STIs to increase with security level.
The amount of time incarcerated may also be related to an inmate's sexual behaviour. If the acceptability of engaging in sex with fellow inmates increases with time incarcerated, one might expect sexual activity to increase with time served. Garland et al. (2005) found that having spent a greater total time incarcerated in combination with serving a longer current sentence or being in maximum security was associated with a greater sexual preference towards the same sex. Further, greater total time incarcerated was significantly associated with sexually identifying as homosexual. Total time served, however, was not associated with the number of sexual encounters with other men after adjusting for security level and length of current sentence (Garland et al., 2005). Similarly, in a maximum security correctional facility for men in the southern U.S., amount of time served was not associated with consensual sex with other men (i.e., kissing, touching, performing oral sex, receiving oral sex, performing anal sex or receiving anal sex) (Hensley, Tewksbury, & Wright, 2001). Thus, the impact of time incarcerated on the risk of STIs is less clear.
Previous research indicates that Canadian federal inmates, particularly women, are a high-risk group for STIs, but information about incarceration characteristics and sexual behaviours most strongly associated with these infections in correctional systems is sparse. To address these and other deficiencies in the literature, in 2007 the Correctional Service of Canada (CSC) conducted the National Inmate Infectious Diseases and Risk Behaviours Survey (NIIDRBS), a self-administered paper questionnaire completed by a large sample of Canadian federal inmates. The survey captured information on inmates' sociodemographics; incarceration characteristics; sexual behaviours in the community and while incarcerated; and, self-reported STIs since admission to CSC for their current sentence.
This report presents NIIDRBS data on the rate of self-reported STIs since admission to Canadian federal penitentiaries. Further, it examines associations between reporting an STI since admission and sociodemographics, sexual behaviour and incarceration characteristics. Such information informs future CSC health research, policy development and programming decisions by identifying high-risk sub-groups, and incarceration characteristics or sexual behaviours related to an increased risk of STIs.
To obtain the data to meet the study objectives, a project team drawn from several federal government departments1 opted to use a self-administered paper and pencil questionnaire (Zakaria, Thompson, & Borgatta, in pressb) as the data collection instrument. Questionnaire development included consultations with inmates in five different penitentiaries, including a women's facility and an Aboriginal inmate group, through focus groups. To maximize comprehension, the questions did not exceed a Grade 8 literacy level. Further, inmates could choose between the English or French version of the questionnaire.
The final questionnaire was 50 pages long and took inmates approximately 45 to 55 minutes to complete. The questionnaire captured information on risk-behaviours associated with the spread of blood-borne and sexually transmitted infections; inmate testing and treatment for HIV and hepatitis C virus (HCV) infections; inmate knowledge of HIV and HCV; and, inmate awareness and use of health education and harm reduction programs.
Prior to data collection, Health Canada's Research Ethics Board reviewed and approved the survey methodology.
NIIDRBS sections relevant to this report included reported STIs since admission to federal prison, sexual behaviours, and incarceration characteristics.
Inmates reported whether they were told they had each of several STIs (chlamydia, gonorrhoea, syphilis, genital herpes, genital warts, or other STI) since admission to prison for their current sentence. Response options included: yes, no, and don't know. To minimize misclassification, "don't know" was considered the equivalent of missing. Because inmates could not report multiple episodes of a specific STI (e.g., chlamydia), these variables indicate the presence or absence of at least one episode of the STI since admission.
An aggregate STI variable was derived from responses to the specific STI questions: inmates who responded "yes" to at least one of the STIs were assigned a value of "yes"; inmates responding "no" to each of the STIs were assigned a value of "no"; and, all other inmates were assigned a missing value.
Inmates reported on the following sexual behaviours during the past six months in prison: any sex (oral, vaginal, or anal); multiple sex partners; unprotected sex with male partners; unprotected sex with female partners; unprotected sex with regular partners; unprotected sex with casual partners (i.e., someone you don't know well); unprotected sex during private family visits; sex with a partner who has HIV, HCV, an STI or an unknown infection status; using a sex toy previously used by someone else; exchange-sex (i.e., a transaction involving the exchange of sex for money, works, rigs, drugs or goods); and, unwanted sex (i.e., forced oral, vaginal, or anal sex).
Inmates admitted within the past three years (n=1,985 or 59% of the full sample) also reported on most of the above sexual behaviours for the last six months in the community prior to starting their current sentence. Community questions were limited to inmates admitted within the past three years to optimize recall accuracy. Sexual behaviours not captured or irrelevant in the community included: unprotected sex with male partners; unprotected sex with female partners; unprotected sex during private family visits; and, using a sex toy previously used by someone else. For a detailed link between these sexual behaviours and the questionnaire, see Appendix A.
Incarceration characteristics examined included: security level, years served of current sentence, cumulative federal and provincial/territorial years served, and participation in unsupervised conditional releases during the past six months in prison.
Inmate security level was imputed using institutional security level. Consequently, security level is unknown for the majority of women residing in multi-level security institutions (99% of the female sample).
Each inmate's years served of current sentence was derived from his/her reported admission year and the survey completion year (2007). Inmates admitted within 2007 had a value of 0.25 years imputed, the approximate mid-point of potential values for a survey conducted in the middle of 2007.
Each inmate's cumulative federal and provincial/territorial years served was the sum of reported total years served in federal prison and provincial/territorial jail. Survey options for federal prison included "less than one year" or the actual total years and months (if one year or longer). Provincial/territorial jail options included "never been in provincial/territorial jail", "less than 1 year", or the actual total years and months (if one year or longer). For both federal prison and provincial/territorial jail, inmates reporting "less than 1 year" had a value of 0.5 years imputed (the mid-point of potential values).
Finally, inmates reported any unsupervised conditional releases (unescorted temporary absences, day paroles, full parole, and statuary release) during the past six months in prison.
The sample frame was all inmates in federal penitentiaries, numbering approximately 13,749 just prior to the time of the survey (March, 2007). Excluded from the frame were inmates unable to understand, orally or in writing, English or French (less than 0.5% of the inmate population). Each penitentiary served as a stratum, the size of which varied from stratum to stratum. For each male penitentiary, a sample size was calculated to ensure estimated proportions had a small margin of error (±5%), 8 times out of 10 [a = 0.20 (two-tailed), σ2 = 0.25, finite population correction factor applied] (Cochran, 1977, p. 75). If the estimated sample size for a specific institution was 80% or more of the institution's population, the whole population of the institution was invited to participate. This occurred with small penitentiary populations so the extra survey cost was minimal. Given the small number (N = 479) of women inmates, all were invited to participate. The final sample size estimate for the entire federal population, including both men and women, was 4,981 inmates.
For each male penitentiary, simple random sampling without replacement from the sample frame generated a primary list. Two or more replacement lists (secondary lists) helped maintain required sample sizes in the event an inmate refused to participate in the study or was not in the institution. Lists sorted by Aboriginal self-identification, primary official language (English or French), and aggregate sentence length facilitated substitutions. If an inmate on the primary list declined to participate or was not in the penitentiary for any reason, another inmate from the secondary list with the same characteristics could substitute for the originally sampled inmate.
Regional (Atlantic, Quebec, Ontario, Prairies, and Pacific) survey coordinators were nominated by the Assistant Deputy Commissioners for Institutional Operations. In addition, each institution's warden nominated an institutional survey coordinator. Regional coordinators acted as liaisons with institutional coordinators and held weekly teleconferences with the Research Branch to resolve logistical issues during survey implementation. The Research Branch prepared an extensive survey training manual for the coordinators and conducted face-to-face training sessions to encourage survey ownership and standardize approaches and messaging.
Regional Management Committees, wardens, security staff and unions were briefed regarding the survey and indicated their support. To raise awareness in institutions about the survey, a general communication and frequently asked questions were sent to all CSC employees, and posters announcing the survey were posted in all institutions (Zakaria et al., in pressb). These posters emphasized the voluntary nature of the survey; guaranteed participants anonymity and confidentiality; and, reinforced that the overall purpose of the survey was to improve inmate health. Wardens also assisted by informing institutional management committees, inmate committees and local unions.
Institutional coordinators received lists of eligible inmates two to three weeks prior to the scheduled data collection period. Before inmates were approached, both primary and secondary lists were reviewed by an institution's Warden or his/her designate to identify security risks. Inmates deemed security risks were either excluded from further consideration or remained eligible to complete the questionnaire in their cell.
Institutional survey coordinators invited inmates on the sample list to participate in the study and to sign a consent form if they agreed (Zakaria et al., in pressb). For efficiency, group information sessions were organized with eligible inmates to describe the survey and review the consent form. Consent, however, was not obtained in a group setting but privately from each inmate. Inmates in segregation were recruited individually. Educational attainment information and experiences interacting with an inmate were used to decide whether to ask an inmate if he/she would like assistance completing the questionnaire. A small version (13.9 cm by 21.6 cm) of the survey poster was left with each inmate approached for participation (Zakaria et al., in pressb).
After scheduling was complete, CSC Security reviewed the list of inmates scheduled to complete the survey in a group setting to ensure compatibility among inmates scheduled for the same group session. Thereafter, each inmate was informed of when and where they were to complete the questionnaire and were reminded the day before. Recruitment activities continued, as necessary, until the end of the data collection period for a specific institution. This allowed replacement of inmates who were unable to complete the questionnaire for any reason.
From May 22 to July 6, 2007, a private firm administered the questionnaire in each institution to those inmates with a signed consent form. The survey coordinator was responsible for organizing inmates for the day and time the survey contractor arrived to distribute questionnaires. Since the contractor did not have the sample list and inmates were specifically instructed not to put their name or the name of anyone else on the questionnaire, it was impossible to link the consent form with the completed questionnaire. In this manner, inmates could be assured of their anonymity and confidentiality.
Each inmate completed a self-administered questionnaire: behind a privacy screen when completed in a group setting; in his/her cell if in segregation; or through private one-on-one interviews if an inmate requested assistance. All participating inmates received the answers to the questionnaire's HIV and HCV knowledge questions after data collection was complete within their institution (Zakaria et al., in pressb).
Several factors limit inmate recruitment and survey completion in the correctional environment including the transfer of inmates between institutions, the departure of inmates at warrant expiry, and inmates on conditional leave during the survey period. In total, 3,370 inmates (3,006 men, 351 women, 13 transgendered) completed a questionnaire. Operational issues limited the majority of facilities from maintaining detailed records of the total number of inmates asked to participate; however, 13 institutions, accounting for approximately 27% of the total federal inmate population at the time of the survey, provided adequate detail to estimate a survey consent and response rate. Across these 13 institutions, which included inmates residing in minimum to maximum security levels, 1,687 inmates were asked to participate, 996 consented (consent rate = 59%) and 811 completed a questionnaire (response rate = 48%). In comparison, the 1995 National Inmate Survey reported a response rate of 64.2% [response rate = number who completed a questionnaire/(number who completed a questionnaire + number who refused)]. If inmate illnesses, releases, and transfers are included in the denominator, however, the response rate declines to 59.7% (Price Waterhouse, 1996, derived from Exhibit 1.3 on p.12). The difference in the response rates across the two surveys could be due to several factors, such as a change in the inmate profile over time or the greater sensitive content of the NIIDRBS.
The contractor retained all completed questionnaires and provided a database of anonymous survey records in August 2007. Preliminary analyses to test the integrity of the data were conducted in the fall and winter of 2007/08. The contractor destroyed all completed questionnaires in June 2008 after all data integrity issues were resolved.
Typically, statistical procedures assume data were obtained through a simple random sample. Under such circumstances each inmate in the sample represents one inmate from the population and estimates derived from the sample relate to the population. In the NIIDRBS, inmates were randomly selected, but the sampling fraction was not consistent across institutions ranging from approximately 8% to 94%. Consequently, each inmate in the sample represented anywhere from about 1 to 13 inmates. Analyzing the NIIDRBS data as if it were obtained through simple random sampling (i.e., each inmate in the sample represents one inmate in the population) would produce incorrect population estimates and variances (Lee & Forthofer, 2006). All statistical estimates shown in this report acknowledge the NIIDRBS' complex sample design by incorporating weights that convey the number of inmates in the population represented by each inmate in the sample. The inverse of the institution's sampling fraction formed the weight for a record. Thus, estimates presented in this report relate to the Canadian federal inmate population. In addition, provision of estimated population sizes in the tables allows derivation of the number of inmates reporting a specific characteristic. Such information is of administrative value.
All analyses used SAS® 9.1 or 9.2 survey procedures (SAS Institute Inc., 2004, 2008) that take the complex sampling design into account. Inferences to the population use common decision criteria (e.g., two-tailed alpha of 0.05). To calculate the variance of an estimate, Taylor series (linearization)2 was used with the finite population correction factor. Each point estimate reported here comes with a two-sided 95% confidence interval using either the Student's t-distribution or standard normal distribution. During bivariate analyses, we used the Rao-Scott chi-square test3 for association if the data were categorical and the Wald F statistic4 for continuous data.
Question non-response is a limitation of most self-report surveys that probe personal or private matters such as sexual behaviour. Although sophisticated procedures exist for addressing low response rates on certain questions, this report used an approach similar to other studies found in the survey literature: on any given question we assume that non-responders and responders share similar characteristics. Tables shown in the report note those analyses using questions where the item non-response rate varied between 20% and 50% (based on the weighted distribution) to alert the reader to this issue. Furthermore, when item non-response exceeded 50%, we chose to suppress the reporting of estimates. For reasons of confidentiality and privacy, we do not report estimates where there are fewer than five inmates sharing a characteristic. Finally, due to their small number (n=13), results for the transgendered are not presented in this report.
To evaluate the extent to which the sample is representative of the inmate population, we compared sample estimates of sociodemographic and incarceration characteristics with estimates obtained from Canadian federal inmate administrative data.
For each specific STI and subgroup of interest, the rate is the weighted sum of reported cases relative to the weighted sum of years served of the current sentence or "person-years at risk". "Person-years at risk" is a measurement combining persons and time, used as a denominator in rates. It is the sum of the individual units of time that the persons in a population have been followed (Last, 1995). As examples, one person followed for one year contributes one person-year; 100 people followed for one year contribute 100 person-years; and, 100 people followed for 0.5 years contribute 50 person-years. Using person-years to derive rates allows more valid comparisons across groups followed for differing periods of time.
[D]
wi: weight assigned to an inmate's record
STIi: 0 indicates no STI since admission; 1 indicates at least one STI since admission
n: number of inmates in subgroup of interest
To contribute to a specific STI rate (i.e., chlamydia, gonorrhoea, syphilis, genital herpes, genital warts, other STI, or any STI), an inmate required complete information on both the presence/absence of the specific STI and the years served of current sentence. Subgroup-specific rates required additional information regarding gender and Aboriginal self-identification.
Estimates are presented for all inmates, separately for men and women, and by Aboriginal self-identification for each gender. An examination of reported STI rates by gender and Aboriginal self-identification allows CSC to respond to the unique needs of women and Aboriginal inmates. Since SAS® 9.2 is not presently capable of generating subgroup rates, these analyses were performed using a SAS macro (available at http://support.sas.com./kb/25/033.html). The t-test was used to assess differences in rates.
All rates are expressed per 10,000 person-years. These rates can be interpreted as the number of inmates reporting at least one STI episode among 10,000 inmates followed for one year.
To identify associations, incarceration characteristics and in-prison sexual behaviours were compared between those reporting an STI since admission and those who did not. These analyses are presented for all inmates and separately for men and women.
To obtain estimates of the relative importance of incarceration characteristics and in-prison sexual behaviours associated with reporting an STI since admission, logistic regression was used. All variables significantly associated with reporting an STI since admission in bivariate analyses were placed in the logistic regression model. If the overall likelihood ratio test for the full model was statistically significant, the linearity assumption for statistically significant continuous variables (e.g., years served of current sentence or cumulative federal and provincial/territorial years served) was assessed using graphing and statistical methods. Briefly, a second model was produced by replacing the continuous variable with a categorical variable; the widths of intervals being decided by the data distribution and interpretability of regression coefficients. If the second model was statistically significant, the regression coefficients of the categorical variable were graphed relative to the mid-points of the categories to visually assess linearity of the logit of STI risk. Statistically, the partial likelihood ratio test was used to determine if the two models significantly differed. If the graph suggested linearity and the models did not significantly differ, the model with the continuous variable was retained as the best model. Otherwise, the model incorporating the categorical variable was retained. This method of assessing linearity is detailed by Hosmer and Lemeshow (2000, p. 99-100).
The Wald chi-square test was used to assess the statistical significance of variables in the model, and the generalized coefficient of determination provided an estimate of the proportion of variation explained by the model (Nagelkerke, 1991). When interpreting generalized coefficients of determination, Hosmer and Lemeshow (2000, p.164) caution that low values are typical in logistic regression and do not necessarily reflect a bad model.
Because STIs diagnosed in prison may be acquired in the community, sexual behaviours during the last six months in the community, prior to starting the current sentence, were also examined. To identify associations, community sexual behaviours were compared between those reporting an STI since admission and those who did not. These analyses are presented for all inmates and separately for men and women.
To obtain estimates of the relative importance of community sexual behaviours associated with reporting an STI since admission, logistic regression was used as previously described.
Canadian federal inmate population characteristics were comparable across data sources indicating the sample was representative of the population (see Appendix B). Based on the NIIDRBS, the majority of inmates were English speaking (78%, 95% CI: 77, 79), non-Aboriginal people (79%, 95% CI: 77, 80), born in Canada (89%, 95% CI: 88, 91), who had a high school diploma or greater at the time of the survey (54%, 95% CI: 52, 56), and were not in committed relationships (69%, 95% CI: 68, 71). Gender differences existed. On average, males were older (38 vs. 34 years, F(1, 3192) = 106.64, p < 0.05), had served a longer duration of their current sentence (4.8 vs. 2.2 years, F(1, 2975)= 274.15, p < 0.05), and were less likely to be Aboriginal (21% vs. 36%, χ2(1, n = 3,234) = 94.37, p < 0.05) compared to women.
The overall rate of an STI since admission to federal prison was 158 per 10,000 person-years (95% CI: 135, 181) (see Table 2). Put differently, if 10,000 inmates were followed for one year after admission, approximately 158 or 1.6% would report at least one STI.
With respect to specific STIs, rates were highest for genital warts (53 per 10,000 person-years, 95% CI: 39, 67), chlamydia (44 per 10,000 person-years, 95% CI: 32, 55), and genital herpes (44 per 10,000 person-years, 95% CI: 33, 56) (see Table 2).
The rate of an STI since admission to federal prison was 4.5 times greater among women than men (669 vs. 150 per 10,000 person-years, t(3,315)=7.89, p<0.05) (see Table 2). Similarly, for specific STIs, the female rates ranged from 2.7 to 9.6 times greater than the male rates.
Among men, the rate of an STI since admission to federal prison was approximately 1.6 times greater in Aboriginal than non-Aboriginal inmates (222 vs. 136 per 10,000 person-years, t(3,315)=2.3, p<0.05) (see Table 3). With respect to specific STIs, only the chlamydia rate significantly differed by Aboriginal self-identification: Aboriginal men were 4.4 times more likely than non-Aboriginal men to report at least one episode of chlamydia since admission to prison (106 vs. 24 per 10,000 person-years, t(3,315) = 3.25, p <0.05).
Among women, the rate of an STI since admission to federal prison did not significantly differ by Aboriginal self-identification (691 vs. 673 per 10,000 person-years, t(3,315) = 0.13, p > 0.05) (see Table 4). Non-Aboriginal women, however, had an "other STI" rate which was 1.9 times greater than that among Aboriginal women (332 vs. 172 per 10,000 person-years, t(3,315) = 2.21, p<0.05).
Among men, security level (χ2(3, n = 2,515) = 14.00, p < 0.05) and cumulative federal and provincial/territorial years served (F(1, 2758) = 6.52, p < 0.05) were significantly associated with reporting an STI since admission (see Table 5). With respect to security level, males reporting an STI were more likely to reside in maximum security facilities (29% vs. 19%) and less likely to reside in minimum security facilities (14% vs. 19%) compared to males not reporting an STI. In regards to years served, males reporting an STI since admission had served, on average, a greater number of years in federal and provincial/territorial institutions compared to males not reporting an STI (11.4 years vs. 9.6 years).
Since security level was unknown for the majority of women who resided in multi-level security institutions, the relationship between security level and reporting an STI since admission could not be examined. None of the remaining incarceration characteristics, however, were significantly associated with reporting an STI since admission among women.
Among men who were sexually active during the past six months in prison, reporting an STI since admission was associated with three in-prison sexual behaviours (see Table 6). Those reporting an STI since admission were less likely to report unprotected sex during private family visits (29% vs. 53%, χ2(1, n = 374) = 7.22, p < 0.05), more likely to report using someone else's sex-toy (16% vs. 5%, χ2(1, n=313) = 4.15, p < 0.05), and more likely to report exchange-sex (27% vs. 8%, χ2 (1, n = 337) = 8.64, p < 0.05) than those not reporting an STI. As previously mentioned, exchange-sex is a transaction involving the exchange of sex for money, works, rigs, drugs or goods. This includes both "being paid for sex" and "paying for sex." These two risk-behaviours were combined because of small sample sizes.
Among women who were sexually active during the past six months in prison, reporting an STI since admission was associated with one in-prison sexual behaviour (see Table 6). Specifically, those reporting an STI since admission were 1.6 times more likely to report unprotected sex with a casual partner compared to those not reporting an STI (56% vs. 35%, χ2(1, n = 77) = 5.11, p
Since women had only one in-prison sexual behaviour associated with reporting an STI since admission (i.e., unprotected sex with casual partners), logistic regression was limited to the men's data (see Table 7). Both statistical testing (partial likelihood ratio test (df = 7, n = 2,296) = 45.87, p < 0.05) and graphing (see Figure 1) indicated that cumulative federal and provincial/territorial years served should be modelled as a categorical variable rather than a continuous variable. In the logistic regression, three variables continued to be significantly associated with reporting an STI since admission: security level (Wald χ2 (3, n = 2,296) = 12.19, p < 0.05), cumulative federal and provincial/territorial years served (Wald χ2 (8, n = 2,296) = 15.74, p < 0.05), and exchange-sex during the past six months in prison (Wald χ2 (1, n = 2,296) = 15.66, p < 0.05).
The odds of reporting an STI since admission were about two times greater among inmates residing in maximum security prisons compared to those residing in minimum security prisons, after adjusting for the other variables in the model (OR = 1.97, 95% CI: 1.29, 3.02, Wald χ2(1, n = 2,296) = 9.75, p < 0.05).
Relative to inmates serving two years or less, the odds of reporting an STI since admission fluctuated over the next 8 years served. Thereafter, the odds consistently increased, plateauing after 14 years. Only one of the "years served" categories significantly differed from the reference group: the odds of reporting an STI since admission were 1.8 times greater among those serving more than 16 years compared to those serving 2 years or less, after adjusting for the remaining variables in the model (OR = 1.77, 95% CI: 1.03, 3.03, Wald χ2(1, n = 2,296) = 4.33, p < 0.05).
The odds of reporting an STI since admission were approximately five times greater among inmates reporting exchange-sex compared to those not reporting exchange-sex, after adjusting for the remaining variables in the model (OR = 4.99, 95% CI: 2.25, 11.07, Wald χ2(1, n = 2,296) = 15.66, p < 0.05).
Because Aboriginal males had a higher rate of reported STIs since admission compared to non-Aboriginal males (see Table 3), Aboriginal self-identification was tested in the final model. Aboriginal self-identification was not independently associated with reporting an STI since admission, after adjusting for the remaining variables in the model (OR = 1.36, 95% CI: 0.89, 2.06, Wald χ2(1, n = 2,248) = 2.03, p > 0.05).
Figure 1. Logit of Reported STI Risk since Admission by Cumulative Federal/Provincial Time Served for Canadian Federal Male Inmates
Note. STI = sexually transmitted infection.
[D]
Among men, reporting an STI since admission was associated with three community sexual behaviours (see Table 8). Those reporting an STI since admission were more likely to report unwanted sex (7% vs. 2%, χ2(1, n =1,234) = 6.40, p < 0.05) and any sex (94% vs. 83%, χ2 (1, n = 1,266) = 5.10, p < 0.05) compared to those not reporting an STI. Further, among men who were sexually active during the last six months in the community, those reporting an STI since admission were three times more likely to report having sex with males compared to those not reporting an STI (15% vs. 5%, χ2(1, n = 874) = 12.23, p < 0.05).
Among women, reporting an STI since admission was associated with five community sexual behaviours (see Table 8). Those reporting an STI since admission were two times more likely to report unwanted sex compared to those not reporting an STI (29% vs. 14%, χ2(1, n = 214) = 11.68, p < 0.05). Further, among women who were sexually active during the last six months in the community, those reporting an STI since admission were less likely to report unprotected sex with regular partners (70% vs. 90%, χ2(1, n = 167) = 18.52, p < 0.05) and any unprotected sex (83% vs. 93%, χ2(1, n=167) = 6.02, p < 0.05) compared to those not reporting an STI. Further, those reporting an STI were more likely to report being paid for sex (51% vs. 29%, χ2(1, n = 167) = 14.09, p < 0.05) and having sex with females (41% vs. 19%, χ2(1, n = 163) = 16.41, p < 0.05) compared to those not reporting an STI.
In the logistic regression for men, one community sexual behaviour remained significantly associated with reporting an STI since admission: having sex with male partners during the last six months in the community (see Table 9). The odds of reporting an STI since admission were 3.4 times greater among those reporting sex with males compared to those who did not, after adjusting for the remaining variables in the model (95% CI: 1.63, 7.26, Wald χ2 = (1, n=7942) = 10.50, p < 0.05). Because Aboriginal males had a higher rate of reported STIs since admission compared to non-Aboriginal males (see Table 3), Aboriginal self-identification was tested in the final model. Aboriginal self-identification was not independently associated with reporting an STI since admission, after adjusting for the remaining variables in the model (OR = 1.38, 95% CI: 0.75, 2.57, Wald χ2(1, n=1,050) = 1.06, p > 0.05).
In the logistic regression for women, three community sexual behaviours remained significantly associated with reporting an STI since admission: unprotected sex with a regular partner, being paid for sex, and having sex with female partners (see Table 10).
The odds of reporting an STI since admission were 85% lower among those having unprotected sex with a regular partner compared to those who did not, after adjusting for the remaining variables in the model (OR = 0.15, 95% CI: 0.04, 0.60, Wald χ2(1, n = 193) = 7.23, p < 0.05).
The odds of reporting an STI since admission were 2.4 times greater among those who were paid for sex compared to those who were not, after adjusting for the remaining variables in the model (OR = 2.44, 95% CI: 1.21, 4.92, Wald χ2(1, n = 193) = 6.16, p < 0.05).
The odds of reporting an STI since admission were 2.9 times greater among those having sex with females compared to those who did not, after adjusting for the remaining variables in the model (OR = 2.94, 95% CI: 1.43, 6.04, Wald χ2(1, n=193) = 8.56, p < 0.05).
The overall rate of at least one reported STI since admission to Canadian federal prison was 158 per 10,000 person-years served (see Table 2). Comparisons across sub-groups revealed differences by gender and Aboriginal self-identification. With respect to gender differences, the rate of reporting an STI since admission was about 4.5 times greater in women than men (669 vs. 150 per 10,000 person-years) (see Table 2). Similarly, for all specific STIs, rates among women were consistently greater than rates among men. Conversely, in the Canadian general population, gonorrhoea and syphilis incidence rates are actually greater in males than females. This gender difference among Canadian federal inmates and the disparity with the Canadian general population have been previously documented (see Table 1). In regards to differences by Aboriginal self-identification, the rate of reporting an STI since admission was approximately 1.6 times greater in Aboriginal males than non-Aboriginal males (222 vs. 136 per 10,000 person-years) (see Table 3). Examination of more specific STI rates among men revealed that chlamydia predominantly accounted for this disparity. Specifically, the rate of reported chlamydia since admission was more than 4 times greater among Aboriginal males than non-Aboriginal males (106 vs. 24 per 10,000 person-years). Caution should be exercised when interpreting these results. Although the suggestion is that women are a high-risk population for reporting an STI diagnosed since admission, other factors could account for the observed elevated risk. In particular, since the questionnaire did not enquire if inmates had been tested for STIs since admission, rates of STIs were calculated among all inmates (i.e., tested and not). If women are more rigorously targeted for testing or are more likely to seek testing at CSC, then they are more likely to have STIs diagnosed compared to men.
Comparisons of STI rates derived from the NIIDRBS with CSC surveillance rates and Canadian general population rates (see Table 1 and PHAC, 2009) are limited because of methodological differences. First, unlike previously published estimates, in this study, the number of episodes of a specific STI is limited to one for each inmate, and each inmate's actual time at risk is estimated as the years served of their current sentence. Although using "time-served" in the denominators of rates provides more accurate estimates of "time at risk," limiting the number of specific STI episodes reported by an inmate will have the impact of biasing rates downward. Second, previously published estimates are based on the number of STIs diagnosed and reported while the NIIDRBS relied on self-report. Estimates based on self-report are vulnerable to recall and social desirability biases. The last methodological difference applies to comparisons between the Canadian general population and federal prison population. There are differences in the rigor of STI screening in these two populations. All inmates are offered voluntary screening and risk-based testing for STIs on admission. Moreover, testing is available upon request by an inmate at any time throughout his/her sentence (Correctional Service of Canada, 2008a, 2008b). Conversely, the Canadian general population must seek testing. These testing differences may be contributing to a spuriously large differential in STI rates between the Canadian general public and Canadian federal inmates.
Among recently sexually active women, reporting an STI since admission was associated with only one in-prison sexual behaviour: unprotected sex with casual partners during the past six months in prison. Those reporting an STI since admission were 1.6 times more likely to have unprotected sex with a casual partner than those not reporting an STI (56% vs. 35%) (see Table 6).
Among men, several incarceration characteristics and in-prison sexual behaviours were associated with reporting an STI since admission to federal prison (see Tables 6 and 7). Hence, logistic regression was used to determine their relative importance. Security level, cumulative federal and provincial/territorial years served, and exchange-sex during the past six months in prison all remained significantly associated with reporting an STI since admission (see Table 7).
Among men residing in maximum security, the odds of reporting an STI since admission were about twice that of men in minimum security. Previous American and Canadian research has found that men are more likely to report sex with other men in maximum security than in minimum security (Garland et al., 2005; Price Waterhouse, 1996). Furthermore, the 1995 Canadian federal inmate survey found that, among men having sex with other men, the proportion reporting condom use decreased from 51% in minimum security to 30% in maximum security (Price Waterhouse, 1996). Unfortunately, sex with male partners in prison could not be thoroughly evaluated using the NIIDRBS because of non-response.
After serving approximately eight years, the odds of reporting an STI among men appeared to consistently increase over time, plateauing after 14 years. Only those inmates serving more than 16 years, however, had odds of an STI which were significantly greater than those serving two years or less (OR = 1.77). Serving time is not a causal risk-factor for STIs, but it may be a surrogate for likelihood of exposure or magnitude of exposure to sexual risk-behaviours. More precisely, as an inmate serves more time, they may become more accepting of same-sex relationships and with additional time comes additional opportunity to engage in risky sexual behaviours.
Finally, the odds of reporting an STI were approximately five times greater among men reporting exchange-sex compared to those who did not. Only 1.8% of the male population, however, reported this high-risk activity during the past six months in prison.
Although Aboriginal self-identification was associated with the rate of reporting an STI since admission among men (see Table 3), it was not associated with reporting an STI in the male logistic regression. That is, the odds of reporting an STI were not significantly greater among Aboriginal men compared to non-Aboriginal men (OR = 1.36) after adjusting for incarceration characteristics and in-prison sexual behaviours.
Because STIs diagnosed in prison may be acquired in the community, the relationships between reporting an STI since admission to federal prison and sexual behaviours in the community were examined. In the logistic regression for men, only one community sexual behaviour remained significantly associated with reporting an STI since admission: sex with male partners during the last six months in the community (see Table 9). The odds of reporting an STI were about 3.4 times greater among those reporting sex with males compared to those who did not. Again, Aboriginal self-identification was not independently associated with reporting an STI since admission after adjusting for community sexual behaviours (OR = 1.38).
In the logistic regression for women, several sexual behaviours during the last six months in the community were significantly associated with reporting an STI since admission (see Table 10). First, the odds of reporting an STI were approximately three times greater in those reporting sex with females compared to those who did not. Second, the odds of reporting an STI were approximately 2.4 times greater among those being paid for sex compared to those who were not. Last, the odds of reporting an STI were 85% lower among those reporting unprotected sex with a regular partner compared to those who did not. This last finding is counterintuitive. Unprotected sex is generally considered a risk-factor for STIs but our finding indicates it is protective. Although further research is necessary to validate this finding, it does reinforce the importance of distinguishing between unprotected sex with casual partners and unprotected sex with regular partners.
It is interesting that for both men and women, sex with a partner of the same sex was associated with greater odds of reporting an STI since admission compared to those not reporting this sexual behaviour. Additional research is necessary to determine the underlying reasons for this observed association. Potential reasons include an increased likelihood of continuing sex in the same-sex environment of prison and/or riskier sexual behaviour among those engaging in sex with a partner of the same sex.
The NIIDRBS indicates Canadian federal inmates report sexual risk-behaviours in both the community and, to a lesser extent, prison. Further, inmates report STIs while in prison. Consistent with previously established associations, our research found that involvement in exchange-sex and unprotected sex with casual partners were associated with an increased likelihood of reporting an STI since admission. Further, sex with a partner of the same sex in the community; and, among men, being in maximum security and cumulative time served were also associated with an increased likelihood of reporting an STI since admission.
These findings have implications for CSC. Presently, CSC's intake health status assessment screens for the following sexual behaviours: sex with a partner of the same sex; unprotected sex with regular and casual partners; ever being a sex trade worker; and, ever being a client of a sex trade worker. In addition to the usual education and recommendations for testing and treatment at admission, healthcare professionals can use affirmative responses to these questions to identify individuals at elevated risk for STIs while incarcerated and thus more likely to benefit from regular STI testing at CSC. In addition, the reasons why offenders choose to have unprotected sex can be explored in a non-judgemental fashion while educating inmates about the harm-reduction items (i.e., condoms, dental dams, and lubricant) CSC provides to prevent the transmission of STIs (CSC, 2004).
Among women, sex with female partners in the community and unprotected sex with casual partners while incarcerated were associated with an increased likelihood of reporting an STI since admission. Although rates of STIs tend to be higher among heterosexual and bisexual women, STIs can be transmitted among women having sex exclusively with women (PHAC, 2008). Thus, STI testing recommendations for women who have sex with women should be based on a detailed risk assessment rather than assumptions of low-risk sexual behaviour. In addition, healthcare professionals need to dispel misconceptions regarding STI risk among women having sex with other women.
Among men, cumulative time served and security level were associated with an increased risk of reporting an STI since admission. Although additional research is needed to identify the underlying reasons these factors are related to STIs, the findings suggest that education, screening and testing needs to continue beyond admission, particularly among high-risk sub-groups, and that access to harm-reduction items needs to be monitored across security levels.
The NIIDRBS has several important limitations. First, since the NIIDRBS was a self-report survey, it is vulnerable to recall and social desirability bias. CSC, however, employed several strategies to optimize accuracy (e.g., an external private firm administered and retained the anonymous self-administered questionnaires). Second, since STIs may not have any signs or symptoms, some inmates in the "no STI" group may have undiagnosed STIs. This type of misclassification would tend to bias associations with reporting an STI towards the null; that is, true associations would be masked. Third, the questionnaire did not allow a distinction between STIs contracted in the community and prison. Hence, the actual rate of STIs resulting from in-prison sexual behaviour cannot be determined. Fourth, as previously mentioned, bias may have resulted from differences in the rigour with which subpopulations are tested for STIs, or from differences in the seeking of STI testing across subpopulations. Finally, a temporality issue exists when examining the associations between reporting an STI since admission and in-prison sexual behaviour. Specifically, behaviours were reported for the past six months in prison, not for the period prior to diagnosis with an STI. Thus, if inmates altered their behaviour because of an STI diagnosis, behaviours captured during the past six months in prison would not reflect pre-diagnosis behaviours. Again, this would contribute to biased estimates of association. Notwithstanding these limitations, which are common to self-report cross-sectional surveys, most of the factors identified as increasing the likelihood of reporting an STI since admission are consistent with previous research.
To overcome methodological limitations, it may be useful in future research to: test all inmates using biosamples (e.g. urine tests) instead of relying on self-report; test inmates at admission and regular follow-ups (e.g., every six months) to maximize accuracy in estimating the time of infection and recalling risk-behaviours; and, ask about risk-behaviours occurring since the last follow-up to minimize temporality issues. Finally, additional research is needed to delineate the factors underlying the observed associations between cumulative time served, security level, and STIs.
Centers for Disease Control (CDC) (2005). Syphilis-CDC Fact Sheet. Retrieved June 15, 2008, from http://www.cdc.gov/std/syphilis/STDFact-Syphilis.htm.
Centers for Disease Control (CDC) (2006a). Chlamydia-CDC Fact Sheet. Retrieved June 15, 2008, from http://www.cdc.gov/std/chlamydia/STDFact-Chlamydia.htm.
Centers for Disease Control (CDC) (2006b). Ghonorrhea-CDC Fact Sheet. Retrieved June 15, 2008, from http://www.cdc.gov/std/Gonorrhea/STDFact-gonorrhea.htm.
Cochran, W.G. (1977). Sampling Techniques, 3rd edition. New York: John Wiley & Sons, Inc.
Correctional Service of Canada (2001). Commissioner's directive 770: visiting. Ottawa, Ontario: Correctional Service of Canada.
Correctional Service of Canada (2004). Commissioner's directive 821: management of infectious diseases. Ottawa, Ontario: Correctional Service of Canada.
Correctional Service of Canada (2008a). Commissioner's Directive 800: Health Services. Ottawa, Ontario: Correctional Service of Canada.
Correctional Service of Canada (2008b). Infectious disease surveillance in Canadian federal penitentiaries, 2002-2004. Ottawa, Ontario: Correctional Service of Canada.
Ford, P.M., Pearson, M., Sankar-Mistry, P., Stevenson, T., Bell, D., & Austin, J. (2000). HIV, hepatitis C and risk behaviour in a Canadian medium-security federal penitentiary. Queen's University HIV Prison Study Group. QJM: An International Journal of Medicine, 93(2), 113-9.
Garland, J., Morgan, R., & Beer, A. (2005). Impact of time in prison and security level on inmates' sexual attitude, behavior, and identity. Psychological Services, 2(2), 151-162.
Hensley, C., Tewksbury, R., & Wright, J. (2001). Exploring the dynamics of masturbation and consensual same-sex activity within a male maximum security prison. The Journal of Men's Studies, 10, 59-71.
Hosmer, D., & Lemeshow, S. (2000). Applied logistic regression, 2nd edition. New York: John Wiley & Sons Inc.
Last, J. (1995). A dictionary of epidemiology (3rd edition). New York: Oxford University Press.
Nagelkerke, N.J.D. (1991). A note on a general definition of the coefficient of determination. Biometrika, 78(3), 691-2.
Price Waterhouse (1996). 1995 National inmate survey: final report (No. SR-02). Ottawa, Ontario: Correctional Service Canada.
Public Health Agency of Canada (2007a). HIV/AIDS Epi Updates. Surveillance and Risk Assessment Division, Centre for Infectious Disease Prevention and Control, Public Health Agency of Canada.
Public Health Agency of Canada (2007b). 2004 Canadian sexually transmitted infections surveillance report. Canada Communicable Disease Report, 33S1, 1-69.
Public Health Agency of Canada (2008). Canadian guidelines on sexually transmitted infections. Retrieved May 20, 2010, from http://www.phac-aspc.gc.ca/std-mts/sti-its/guide-lignesdir-eng.php.
Public Health Agency of Canada (2009). Brief report on sexually transmitted infections in Canada: 2007. Retrieved May 21, 2010, from http://www.phac-aspc.gc.ca/publicat/2009/sti-its/pdf/sti_brief-its_bref_2009-eng.pdf.
Public Health Agency of Canada (n.d.c). Infectious diseases A-Z index. Retrieved Dec 15, 2006, from http://www.phac-aspc.gc.ca/id-mi/index-eng.php.
Public Health Agency of Canada (n.d.d). Notifiable Diseases On-Line. Retrieved Sep 24, 2008, from http://dsol-smed.phac-aspc.gc.ca/dsol-smed/ndis/index_e.html.
SAS Institute Inc. (2004). SAS/STAT ® 9.1 User's Guide. Cary, NC: SAS Institute Inc.
SAS Institute Inc. (2008). SAS/STAT ® 9.2 User's Guide. Cary, NC: SAS Institute Inc.
UNAIDS (2006). 2006 Report on the global AIDS epidemic. UNAIDS.
Zakaria, D., Thompson, J., & Borgatta, F. (in pressa). Rates of reported HIV and HCV infections since admission to Canadian federal prison and associated incarceration characteristics and drug-related risk-behaviours. Ottawa: Correctional Service Canada.
Zakaria, D., Thompson, J., & Borgatta, F. (in pressb). Study materials for the 2007 National Inmate Infectious Diseases and Risk-Behaviours Survey. Ottawa: Correctional Service Canada.
Sexual Behaviours | NIIDRBS Questions | |
---|---|---|
Prison | Community | |
Sexual Behaviours during the Past Six Months | ||
Any sex (oral, vaginal, or anal) | Q55 | Q89 |
Sex with multiple partners | Q55a | Q90, Q90a, Q91, Q91a |
Unprotected sex with male partners | Q57 | - |
Unprotected sex with female partners | Q56 | - |
Unprotected sex with regular partners | Q59 | Q93 |
Unprotected sex with casual partners | Q60 | Q94 |
Unprotected sex during private family visits | Q62, Q63 | - |
Having sex with a partner who has HIV, HCV, an STI, or an unknown infection statusa | Q61 | Q95 |
Using someone else's sex toy after they used it | Q58 | - |
Exchange-sex (i.e., a transaction involving the exchange of sex for money, works, rigs, drugs or goods) | Q65, Q66, Q67, Q68 | Q96, Q97 |
Unwanted sex (i.e., forced oral, vaginal, or anal sex) | Q69 | Q98 |
NIIDRBS (n=3,357) (N=13,701) |
CSC Administrative Data (N=13,041) |
||||||||
---|---|---|---|---|---|---|---|---|---|
Men (n=3,006) (N=13,222) |
Women (n=351) (N=479) |
X2(df) or F(v1,v2) |
Men (N=12,574) |
Women (N=467) |
|||||
Characteristics | n | Mean or % (95% CI ) | n | Mean or % (95% CI ) | N | Mean or % | N | Mean or % | |
Age (years) | 2,899 | 38 (38, 39) | 335 | 34 (34, 35) | 106.64* (1,3192) | 12,554 | 38 | 466 | 35 |
Highest level of education at time of survey (%) | |||||||||
Less than highschool diploma | 1,252 | 46 (44, 48) | 156 | 48 (45, 51) | 0.68 (1) | - | - | - | - |
Highschool diploma or greater | 1,533 | 54 (52, 56) | 176 | 52 (49, 55) | - | - | - | - | |
Marital status (%) | |||||||||
Married/common law | 884 | 31 (29, 32) | 121 | 35 (32, 38) | 4.90* (1) | 4,839 | 39 | 165 | 36 |
Single/separated/divorced/widowed | 2,043 | 69 (68, 71) | 224 | 65 (62, 68) | 7,654 | 61 | 297 | 64 | |
Country of birth (%) | |||||||||
Canada | 2,622 | 89 (88, 90) | 320 | 92 (91, 94) | 5.87* (1) | 11,175 | 89 | 412 | 89 |
Other | 305 | 11 (10, 12) | 26 | 8 (6, 9) | 1,386 | 11 | 53 | 11 | |
Aboriginal self-identification (%) | |||||||||
Aboriginal | 612 | 21 (19, 22) | 129 | 36 (33, 38) | 94.37* (1) | 2,466 | 20 | 147 | 32 |
Non-Aboriginal | 2,281 | 79 (78, 81) | 212 | 65 (62, 67) | 10,023 | 80 | 310 | 68 | |
Race (%) | |||||||||
White/caucasian | 1,852 | 65 (63, 67) | 179 | 55 (52, 58) | 82.52* (2) | 8,482 | 68 | 258 | 56 |
Aboriginal | 612 | 21 (20, 23) | 129 | 36 (34, 38) | 2,466 | 20 | 147 | 32 | |
Other visible minority | 356 | 14 (13, 15) | 28 | 9 (7, 11) | 1,541 | 12 | 52 | 11 | |
Language most comfortable speaking (%) | |||||||||
English | 2,154 | 78 (77, 79) | 302 | 84 (83, 86) | 32.90* (2) | 8,425 | 74 | 317 | 79 |
French | 719 | 20 (20, 21) | 37 | 14 (13, 15) | 2,342 | 21 | 62 | 15 | |
Other | 54 | 2 (1, 2) | 6 | 2 (1, 2) | 642 | 6 | 22 | 5 | |
Years served of present sentence | 2,702 | 4.8 (4.6, 5.1) | 318 | 2.2 (2.0, 2.4) | 274.15* (1, 2975) | 12,554 | 4.4 | 466 | 3.0 |
Region (%) | |||||||||
Atlantic | 317 | 10 (10, 10) | 50 | 13 (13, 13) | - | 1,297 | 10 | 62 | 13 |
Quebec | 868 | 24 (24, 24) | 42 | 16 (16, 16) | 2,990 | 24 | 73 | 16 | |
Ontario | 627 | 27 (27, 27) | 84 | 26 (26, 26) | 3,344 | 27 | 123 | 26 | |
Prairie | 847 | 25 (25, 25) | 137 | 33 (33, 33) | 3,168 | 25 | 151 | 32 | |
Pacific | 347 | 15 (14, 15) | 38 | 13 (12, 13) | 1,772 | 14 | 58 | 12 | |
Security level (%) | |||||||||
Maximum | 581 | 21 (21, 21) | 0 | - | 3,199 | 25 | 102 | 22 | |
Medium | 1,488 | 60 (60, 60) | 0 | 6,934 | 55 | 196 | 42 | ||
Minimum | 869 | 18 (18, 18) | 4 | 1 (1, 1) | 1,907 | 15 | 161 | 34 | |
Unknown | 68 | 1 (1, 1) | 347 | 99 (99, 99) | 534 | 4 | 8 | 2 |
1 CSC Research Branch, CSC Public Health Branch, and the Public Health Agency of Canada HIV/AIDS Policy, Coordination and Programs Division and Community Acquired Infections Division.
2 See SAS Institute Inc. (2004, p. 166) for details and related references.
3 See SAS Institute Inc. (2004, p. 4216) for details and related references.
4 See SAS Institute Inc. (2008, p. 6558) for details.