Abstract
This investigation compared circumcised and intact (uncircumcised) men attending sexually transmitted infection (STI) clinics on condom perceptions and frequencies of use. Men (N = 316) were recruited from public clinics in two US states. Circumcision status was self-reported through the aid of diagrams. Intact men were less likely to report unprotected vaginal sex (P < 0.001), infrequent condom use (P = 0.02) or lack of confidence to use condoms (P = 0.049). The bivariate association between circumcision status and unprotected sex was moderated by age (P < 0.001), recent
Introduction
Evidence strongly supports that condoms have protective value against sexually transmitted infections (STIs).1–4 Despite an expanding body of empirical literature on condom use issues among men5–10 and the potential importance of circumcision to preventing HIV acquisition,11,12 published studies have neglected to address the question of whether circumcised men differ from their intact (uncircumcised) counterparts regarding perceptions of condoms and frequency of their use. Because intact men must deal with foreskin while applying and using condoms, it is conceivable that differences between circumcised and intact men may exist regarding perceptions and frequency of use.
Men attending STI clinics are a particularly important population to study regarding the research question of differences based on circumcision status. 13 As a consequence of the synergistic relationship between STIs and human immunodeficiency virus (HIV),14–17 this population is inherently at an elevated risk of HIV infection. Thus, understanding factors related to optimizing condom use among men attending STI clinics is important to public health. Accordingly, the purpose of this investigation was to compare selected perceptions about condoms, and the frequencies of condom use, between circumcised and intact men (aged 18 and older) attending STI clinics.
Methods
Study sample
Data were collected from December 2007 through April 2011 as part of an National Institute for Health (NIH)-funded study of condom effectiveness. A convenience sample was enrolled. Established patients were recruited from two publicly-funded STI clinics: one in Louisville, KY and the other in Cincinnati, OH. Eligibility criteria included being 18 or older; English speaking; willing to be tested for Chlamydia, gonorrhea and trichomoniasis by providing a urine specimen; willing to provide contact information; reporting penile vaginal intercourse in the preceding three months; and providing written informed consent. Research assistants screened 1090 patients for eligibility. Of these 1042 were eligible and were thus invited to participate, with 661 agreeing to do so yielding a participation rate of 63.4%. Of these 661 volunteers, 316 were men and were therefore included in the present analysis. The study protocol was approved by the institutional review board at the University of Kentucky (USA).
Measures
Based on evidence suggesting the possibility of decreased reporting bias, 18 data were collected using audio-computer-assisted self-interviewing (A-CASI). An eloquent advantage of A-CASI for this study was the ability to provide men with two diagrams when they answered the key question, ‘Are you circumcised? (please use the diagram if needed)’. The diagrams clearly displayed and labelled a circumcised and an uncircumcised penis.
Regarding condom use perceptions and behaviours, the A-CASI clearly and repeatedly defined sex as putting the penis into the vagina. Using a three-month recall period, men reported how many times they had sex and, of these times, how many times a condom was used. Five items of the A-CASI also assessed men's perceptions of sensation pertaining to condom use. These five items (see Table 1) were derived from the sensation subscale of the Condom Barriers Scale, a previously validated assessment instrument. 19 In addition, a single item assessed men's confidence in their ability to use condoms.
Descriptive comparisons of selected condom use perceptions between circumcised and intact men
Response options ranged from 1 ‘strongly disagree’ to 5 ‘strongly agree’
Statistics
Bivariate assessments entailed χ2 tests to compare circumcised and intact men on condom sensation perceptions (dichotomized as ‘high’ or ‘low’ on a condom barriers subscale) and frequency of condom use (quantified by presence or absence of unprotected vaginal sex and, based on potential protective value of infrequent condom use, 20 by whether condoms were used in 50% or fewer of sex acts) as well as ascertain possible moderating effects of age (dichotomized as 24 years of age or younger versus 25 years or older), race (dichotomized as African American/Black or not African American/Black), recent history (past 3 months) of having an STI and men's confidence in their ability to use condoms (dichotomized as completely confident or not completely confident). Multivariate logistic regression models were then fit to relate frequency of condom use to circumcision status and any moderating variables identified in the bivariate assessments, with forward selection (using the Wald method with significance threshold 0.05) employed to cull the models for interpretability and generalizability beyond the present sample. Version 19 of SPSS (SPSS Inc, Chicago, IL, USA) was used for this analysis.
Results
The mean age of the sample was 31.5 years (standard deviation [SD] = 11.0). The majority (80.7%) identified as African American/Black. The median monthly income (including income from government assistance programs) was between $500 and $1000. More than one of every five men (22.8%) reported being diagnosed with an STI (other than the day of study enrollment) in the past three months. Most men (75.9%) indicated they had been circumcised, with the remainder (24.1%) indicating they had not been circumcised. Circumcised men did not differ significantly from intact men on age (P = 0.49) or race (P = 0.37). However, circumcision status was significantly associated (P = 0.01) with recent STI history. Among those who had been circumcised, 25.4% indicated having an STD in the past three months versus 11.8% among intact men. In the past three months, 77.5% of the men reported engaging in unprotected vaginal sex (UVS) and 49.4% reported using condoms for 50% or fewer of all penile-vaginal sex acts occurring in that time period.
Bivariate associations
The five-item scale assessing men's condom-associated sensation perceptions achieved adequate inter-item reliability (Cronbach's alpha = 0.70). Scores on the composite variable ranged from 5 to 25, with higher scores representing greater levels of sensation-related issues pertaining to condom use. The mean was 11.7, with a standard deviation of 4.5. The distribution was markedly non-normal (skewness = 3.0); thus the composite variable was dichotomized by an approximate median split which left 46.8% classified ‘high’ on issues related to condom sensation. Circumcision status was not significantly (P < 0.23) associated with this dichotomous outcome. Among those circumcised, 48.8% were classified as ‘high’ compared with 40.8% among intact men (percent relative difference = 19.4). Table 1 displays descriptive statistics for each of the five condom sensation items by circumcision status.
In addition to condom sensation issues, men's perceptions about their ability to use condoms were evaluated for association with circumcision status. This single item had a five-point response scale (ranging from 0 ‘very confident’ to 4 ‘not confident’). The mean score was 0.51, with a standard deviation of 0.83. The distribution was markedly non-normal (skewness = 14.8); thus the ability variable was dichotomized by an approximate median split which left 35.8% classified as ‘less than completely confident’ regarding their ability to use condoms. Circumcision status was significantly (P = 0.049) associated with this dichotomous outcome. Among those circumcised, 38.7% were not completely confident compared with 26.3% among intact men (percent relative difference = 47.1).
Circumcision status was significantly (P < 0.001) associated with the outcome of having any UVS in the past three months, with circumcised men being more likely to report UVS. Among those circumcised, 82.9% reported UVS compared with 60.5% among intact men (percent relative difference = 37.0). Circumcision status was also significantly (P = 0.02) associated with the outcome of using condoms for 50% or fewer of the sex acts occurring in the past three months, with circumcised men being more likely to report this risk behaviour. Among those who were circumcised, 52.9% reported 50% use or less compared with 38.1% among intact men (percent relative difference = 38.8).
Tests for moderation
The bivariate association between circumcision status and UVS was moderated by age, with a significant relationship among those less than 25 years old (P < 0.001) but not among those at least 25 years old (P = 0.12). The association was also moderated by recent history of STI acquisition, with significance among those not diagnosed in the past three months (P < 0.001) but not among those recently diagnosed (P = 0.87), and by men's confidence to use condoms, with significance among those indicating complete confidence (P < 0.001) but not for those lacking complete confidence (P = 0.33). The association was not moderated by race.
The bivariate association between circumcision status and using condoms for 50% or fewer of recent sex events was similarly moderated by age (P = 0.002 for younger men versus P = 0.94 for older men). The association was also moderated by recent history of STI acquisition (P = 0.02 for those not having a recent history versus P = 0.86 for those reporting a recent diagnosis) and by men's confidence to use condoms (P = 0.01 for those indicating complete confidence versus P = 0.76 for those lacking complete confidence). The association was not moderated by race.
Regression findings
In the multivariate logistic regression model for UVS, only two main effects were included by forward selection (circumcision status and confidence to use condoms), and no interaction effects were added. Men who had been circumcised were estimated to have almost three times the odds (estimated odds ratio [EOR]2.96; 95% confidence interval [CI]1.66–5.27, P < 0.001) to report UVS compared with intact but otherwise similar (i.e. comparably confident) counterparts. Men lacking complete confidence in their ability to use condoms were estimated to have more than double the odds (EOR 2.28; 95% CI 1.21–4.31, P < 0.01) to report UVS compared with completely confident but otherwise similar (i.e. same on circumcision) counterparts.
In the multivariate logistic regression model for using condoms 50% of the times sex occurred or less, three main effects were included by forward selection (circumcision status, age and confidence to use condoms), and one interaction effect (involving circumcision status and age) was added. The nature of the interaction effect was that, controlling for confidence, circumcision considerably amplified the odds of 50% or less condom use among young men in particular (EOR 4.28; 95% CI 1.61–11.44, P = 0.04). On the other hand, men lacking complete confidence were estimated to have more than double the odds (EOR 2.10; 95% CI 1.29–3.39, P = 0.003) of reporting 50% or less condom use compared with completely confident but otherwise similar (i.e. same on circumcision and age) counterparts.
Discussion
To the best of our knowledge, this is the first published investigation of differential perceptions and behaviours relative to condom use between circumcised and intact men attending STI clinics in the USA. The investigation is important because it sheds light on a previously unexplored area of inquiry: whether intact men have greater issues with condom use than circumcised men. While intuitively appealing, the suppositions that intact men may have poorer perceptions and behaviours regarding condom use were not supported by the present findings. Indeed, these findings suggest quite the opposite in that intact men were less likely to report UVS and less likely to report infrequent condom use compared with their circumcised counterparts. These two observations are quite consistent with the descriptive finding that showed a greater than two-fold past history of STIs for circumcised men compared with intact men. Biological reasons for this difference in STI history are not currently warranted because circumcision has only been shown to protect against HIV 12 Future research should determine possible reasons why intact men may practice safer sex.
Although the scale measure of sensation perceptions did not vary significantly by circumcision status, intact men were more likely to have complete confidence in their ability to use condoms than circumcised men. This observation suggests that intact men may, for some reason, have gained information or skills associated with condom use that their circumcised counterparts lacked.
The regression findings were particularly informative given their parsimony. That, for example, the model for UVS included only circumcision status and confidence in use of condoms is rather surprising since age, race and recent STI history (as well as interaction terms based on circumcision status) were also eligible for inclusion via forward selection. Regarding the outcome of infrequent condom use (50% or fewer of penile-vaginal sex acts), circumcision status significantly predicted the outcome only for those younger than 25 years of age. This suggests that differences in frequency of condom use between circumcised and intact men may diminish with age after the mid-20s. Moreover, the significant multivariate association between confidence in condom use and this outcome suggests the importance of self-efficacy to condom use frequency, especially since significance was maintained despite the eligibility of other variables for inclusion via forward selection.
Limitations
As is true for most sexuality research, findings are limited by the validity of self-report. However, the use of diagrams to help men determine their circumcision status may have been an important buffer against possible misclassification bias. Also, the use of a convenience sample limits the generalizability of the findings to other populations of STI clinic attendees. Additionally, the sample size was fairly modest at just over 300 men thereby creating a potential for type II hypothesis testing errors. Also, although our scale measure achieved an alpha of 0.70 it should be noted that this is the lower limit of a reliable construct. It should also be noted that we did specifically design the assessment instrument to measure problems or issues with condom use that may be unique to intact men, for example, pulling back the foreskin during application of condoms. Finally, measurement of the outcome variables is a potential limitation; numerous methods exist for assessing condom use behaviours and only two were applied in the current analysis.
Conclusions
This novel investigation of men attending STI clinics explored whether circumcision status was associated with condom-related perceptions and behaviours. Despite having to retract their foreskins to apply condoms, intact men may have more confidence in their ability to use condoms and they may be more likely to do so than their circumcised counterparts. However, even after controlling for circumcision status and age, confidence predicted condom use, suggesting that intervention programmes should focus on building men's confidence in their ability to use condoms. On the other hand, differences in condom use frequency between intact and circumcised men may dissipate after age 25, and sensation-related aspects of condom use may not differ between intact and circumcised men attending STI clinics. Collectively, these findings suggest that STI clinic counselling and education protocols may benefit men, regardless of race, age or recent STI history, by emphasizing strategies to enhance self-efficacy for condom use and by targeting young circumcised men for added intervention. Future studies should build on this investigation to determine why intact men may be more likely to use condoms.
Footnotes
Acknowledgement
Support for this project was provided by a grant to the first author from the National Institutes of Allergies and Infectious Diseases, grant # 5 R01 AI068119.