Article Text
PDF
Abstract
Background The concept of concurrent partnerships, while theoretically appealing, has been challenged at many levels. However, non-monogamy may be an important risk factor for the acquisition and transmission of sexually transmitted infections (STI). One's own non-monogamy is a risk factor for transmitting STI to others, partners' non-monogamy is a risk factor for acquiring STI and, most importantly, mutual non-monogamy is a population level determinant of increased STI spread. This study describes the levels, distribution and correlates of non-monogamy, partners' non-monogamy and mutual non-monogamy among adult men and women in the USA.
Methods Data from the National Survey of Family Growth (NSFG) Cycle 6 were used. NSFG is a national household survey of subjects aged 15–44 years in the USA. Cochran–Mantel–Haenszel tests and χ2 tests were used in the analysis.
Results Among sexually active adults, 17.6% of women and 23.0% of men (an estimated 19 million) reported non-monogamy over the past 12 months in 2002. An estimated 11 million Americans (1 in 10) reported partners' non-monogamy and an estimated 8.4 million (7% of women and 10.5% of men) reported mutual non-monogamy. All three types of non-monogamy were reported more frequently by men than women. Younger age, lower education, formerly or never married status, living below the poverty level and having spent time in jail were associated with all three types of non-monogamy in general.
Conclusions The three types of non-monogamy may be helpful in tailoring prevention messages and targeting prevention efforts to subgroups most likely to spread infection.
- Sexual behaviour
- sexual networks
- sexual practices
Statistics from Altmetric.com
Introduction
Non-monogamy (ie, sexual intercourse with multiple sex partners) has long been recognised as a correlate of sexually transmitted infections (STI).1 The nature of this association was elaborated during later years in three distinct ways: (1) through the differentiation between risks for STI transmission versus STI acquisition2; (2) through recognition of the importance of temporal order, reflected in the differentiation between sequential versus concurrent partnerships3 4; and (3) through the differentiation between individual level risk factors for STI and population level determinants of STI spread.5 6 Particularly over the past decade, investigations of the role of concurrent partnerships in the prevalence and incidence of STI strongly suggested that, at the individual level, having concurrent partners may be a risk factor for transmitting STI to others.7 8 Conversely, having sex partners who themselves have concurrent partners has emerged as a risk factor for acquiring STI.9–11 While theoretically appealing, the concept of concurrency has proved challenging; it is difficult to measure, it may be correlated with other network characteristics which affect its net impact and it is difficult to distinguish from multiple monogamous partnerships accumulated over given time periods.12 Non-monogamy is not the same as having concurrent partners. Multiple partners may be concurrent or sequential. A recent formulation defines both concurrency and sequential partnerships in terms of the ‘gap’.13–16 The length of time (‘gap’) between the end of an individual's partnership with one sex partner and the start of their next partnership is an important factor influencing STI transmission. Gaps can be either positive, indicating that there is a non-zero interval between the two partnerships, or negative, indicating that the partnerships are concurrent (ie, that they overlap). Concurrent partnerships are associated with STI. However, even serially monogamous partnerships may be associated with STI if the gap length between them is shorter than the infectious period since there is then a non-zero risk of transmission.13–15
Collection and analysis of data used in assessing concurrent partnerships is complicated, time-consuming and potentially imprecise.17 It is also difficult to disentangle the effects of concurrency and multiplicity of partners.18 Consequently, while quite powerful and helpful at the conceptual and theoretical levels, concurrency of sex partnerships has not emerged as a frequently used parameter in the day-to-day prevention activities of STD prevention programmes. We propose that ‘non-monogamy over the past year’ may present a close approximation to concurrency. Non-monogamy does not require negative gaps or partnerships that overlap in time, it includes short gaps which are associated with non-zero risk of transmission and it is relatively easier to measure. It is important to differentiate between one's own non-monogamy (a risk factor for transmitting STI to partners), partners' non-monogamy (a risk factor for acquiring STI) and mutual non-monogamy (a population level determinant of increased STI spread). Mutual non-monogamy is the network pattern that would spread STI most effectively in the population. Concurrent partnerships with monogamous partners may even limit the spread of infection.19
In this paper, using nationally representative survey data, we describe the levels, distribution and correlates of non-monogamy, partners' non-monogamy and mutual non-monogamy among sexually active adult men and women living in the USA.
Methods
We used data from the National Survey of Family Growth (NSFG) Cycle 6 for this analysis. NSFG is a national household survey of subjects aged 15–44 years in the USA. Details of the survey have been published elsewhere.20 21 Briefly, data were collected in 2002–3 and included oversampling of African Americans, Hispanics/Latinos and adolescents (aged 15–19 years). NSFG consists of an interviewer-administered survey and an audio computer-assisted self-interview (ACASI). For this analysis, data focusing on sexual risk behaviours were taken from the ACASI portion of the survey. Given the small sample sizes for several racial or ethnic groups, analyses were limited to respondents who were Hispanic/Latino, non-Hispanic white and non-Hispanic black (African American). Data analyses were limited to respondents who had at least one vaginal, oral or anal partner of the opposite sex in the past 12 months (referred to as ‘sexually active’). There were 5978 women and 3577 men who were sexually active and reported their race/ethnicity as Hispanic/Latino, white (non-Hispanic) or black (non-Hispanic).
For this analysis we used three behavioural outcome measures: (1) non-monogamy in the past 12 months; (2) partners' non-monogamy in the past 12 months; and (3) mutual non-monogamy in the past 12 months. Non-monogamy was defined as having two or more partners in the past 12 months based on responses to the question: ‘Thinking about the last 12 months, how many male sex partners have you had? Please count every partner, even those you had sex with only once’. For this question, sex was defined as vaginal, oral or anal sex. Partners' non-monogamy in the past 12 months was defined as having sex with someone of the opposite sex ‘who (was) also having sex with other people at around the same time’ and the variable was coded as ‘yes’ or ‘no.’ We created a mutual non-monogamy variable using the first two outcome variables. Respondents who reported their own non-monogamy in the past 12 months and who also reported a partner's non-monogamy in the past 12 months were classified as engaging in mutual non-monogamy; the remaining sexually active respondents who provided valid responses to these variables were classified as not engaging in mutual non-monogamy.
For each outcome variable we conducted separate Cochran–Mantel–Haenszel (CMH) tests to examine the relationship between the outcome and each demographic variable stratified by race/ethnicity. We used χ2 tests for all bivariate analyses and logistic regression to examine multiple models for each outcome. All analyses were conducted separately by sex. Demographic factors examined included age, highest level of education, current marital/cohabitation status and poverty level income. Poverty level income was calculated from family income divided by the appropriate average poverty level threshold (by family size) as determined by the US Census Bureau (http://www.cdc.gov/nchs/data/nsfg/App2_RecodeSpecs.pdf). For men we also conducted analyses to examine the outcome measures by ‘spent time in jail in the past 12 months’; this question was not included in the female survey. All analyses were weighted to represent the US population. To account for the complex sampling methods used by NSFG, the survey procedures in SAS Release 9.1 (SAS Institute) were used to calculate population total estimates and SUDAAN Release 9.0.1 was used to calculate the percentages, standard errors, CMH and χ2 analyses.
Results
Non-monogamy in past 12 months
Overall, of those who were sexually active, 17.6% of women and 23.0% of men reported non-monogamy (ie, having multiple (two or more) partners of the opposite sex) in the past 12 months; an estimated 19 million Americans aged 15–44 years engaged in non-monogamy in 2002 (table 1). For women and men, non-Hispanic blacks (26.8 and 38.9%, respectively) had significantly higher reports of non-monogamy in the past 12 months (χ2=18.79, df=2, p<0.0001 and χ2=25.83, df=2, p<0.0001, respectively). For all racial/ethnic groups, significantly higher percentages of female and male adolescents (15–19 years) and young adults (20–24 years) reported non-monogamy. More than one in three adolescents reported non-monogamy in the past 12 months. Education level was significantly associated with non-monogamy for sexually active women (CMH χ2=23.35, df=3, p<0.0001) and men (CMH χ2=9.17, df=3, p<0.0001); not surprisingly, those with less than a high school education (including those still in school) had higher reports of non-monogamy in the past 12 months. Women and men who had obtained a Bachelor's degree or higher (10.6% and 13.8%, respectively) had the lowest reports of non-monogamy. These findings held for non-Hispanic white and black women and non-Hispanic white men; however, education was not significantly associated with non-monogamy for Hispanic women and men and non-Hispanic black men.
Sexually active women† and men: non-monogamy in past 12 months (percentages with standard errors)
Generally, marital/cohabitation status, poverty and spending time in jail (men only) were significantly associated with non-monogamy (table 1). Specifically, marital/cohabitation status was a significant correlate of non-monogamy for all sexually active women (CMH χ2=105.16, df=3, p<0.0001) and men (CMH χ2=95.25, df=3, p<0.0001) and for each racial/ethnic group separately. More than one in three formerly married or never married women and men engaged in non-monogamy in the past 12 months; conversely, less than 5% of married women and men reported non-monogamy with the exception of Hispanic and non-Hispanic black men. For women, poverty level income was significantly related to non-monogamy (CMH χ2=21.54, df=2, p<0.0001), with those whose family income was <150% of the poverty level having the highest reports of non-monogamy (24.5% overall). These findings held when the analyses were conducted by race/ethnicity. For men, poverty level income was not associated with non-monogamy (CMH χ2=1.49, df=2, p=0.2315). Finally, for all sexually active men, having spent time in jail in the past 12 months was significantly associated with non-monogamy during the same time period (CMH χ2=41.29, df=1, p<0.0001), with 46.6% of men who spent time in jail reporting non-monogamy compared with only 21.2% of men who had not been in jail. These findings held for Hispanic and non-Hispanic white men; however, for non-Hispanic black men there were no significant differences (χ2=0.44, df=1, p=0.5078) in reports of non-monogamy between men who spent time in jail in the past 12 months (35.0%) and those who did not (39.5%).
When the demographic data were analysed in multiple logistic regression models (tables 4 and 5), several demographic characteristics were no longer significant predictors of non-monogamy. For women (table 4), those aged 15–19 years, 20–24 years and 25–34 years were 1.77 (95% CI 1.29 to 2.45), 1.78 (95% CI 1.28 to 2.48) and 1.32 (95% CI 1.02 to 1.71) times as likely as those aged 35–44 years to report non-monogamy; however, age was not significantly associated with non-monogamy in the logistic regression model for men (table 5). Conversely, while non-Hispanic black and Hispanic men were 2.02 (95% CI 1.52 to 2.70) and 1.76 (95% CI 1.28 to 2.42) times as likely as non-Hispanic white men to engage in non-monogamy, race/ethnicity did not significantly predict non-monogamy in the female model. In the male and female multiple models, education and poverty level income were not significantly associated with non-monogamy with one exception. Women with a high school degree or GED were 1.38 (95% CI 1.03 to 1.84) times more likely than women with a Bachelor's degree or higher to report non-monogamy. Current marital/cohabitation status remained a significant predictor of non-monogamy in the female and male models. Finally, men who spent time in jail in the past year were 2.5 (95% CI 1.84 to 3.40) times as likely as men who did not spend time in jail to report non-monogamy.
Partners' non-monogamy in past 12 months
Overall, approximately one in 10 sexually active women and men reported a partner's non-monogamy in the past 12 months (table 2). This finding suggests that an estimated 11 million women and men may be at risk for STIs because of their partners' non-monogamy. There were statistically significant differences in partner's non-monogamy by race/ethnicity for women (χ2=23.68, df=2, p<0.0001) and men (χ2=10.53, df=2, p=0.0001), with non-Hispanic blacks having the highest reports (18.5% and 21.3%, respectively). Similar to the analyses for respondent's non-monogamy, there were significant differences in reports of partners' non-monogamy in the past 12 months by age for men (CMH χ2=18.83, df=3, p<0.0001) and women (CMH χ2=23.31, df=3, p<0.0001) with adolescents (21.1% and 20.0%, respectively) and young adults (21.2% and 16.5%, respectively) having the highest reports of a non-monogamous partner. Although these findings held for non-Hispanic white women (χ2=24.35, df=3, p<0.0001) and men (χ2=15.88, df=3, p<0.0001) and Hispanic women (χ2=4.42, df=3, p=0.0062), there were no significant differences in partners' non-monogamy for non-Hispanic black women and men and Hispanic men. Similar patterns emerged when examining the relationship between education and partners' non-monogamy; there were significant differences for all sexually active women (CMH χ2=5.53, df=3, p=0.0016) and men (CMH χ2=6.54, df=3, p=0.0005), for non-Hispanic women and men and for Hispanic women. Those with the highest education level (Bachelor's degree or higher) had the lowest reports (∼7%) of partners' non-monogamy.
Sexually active women and men: partners' non-monogamy in past 12 months
For partners' non-monogamy, findings for marital/cohabitation status, poverty and time spent in jail (men only) were similar to those for respondents' non-monogamy. Marital/cohabitation status was a significant correlate of partners' non-monogamy for women (CMH χ2=64.86, df=3, p<0.0001) and men (CMH χ2=52.43, df=3, p<0.0001) overall and by racial/ethnic group (p<0.0001). Approximately one in four sexually active women and men who were formerly married or never married reported a partner's non-monogamy. Only 2.4% of married women and 2.7% of married men reported a partner's non-monogamy. Poverty level income was a significant correlate of partners' non-monogamy for women only (CMH χ2=7.98, df=2, p=0.0007). When examined separately by race/ethnicity, this finding held for non-Hispanic white women only (χ2=6.45, df=3, p=0.0025). Non-Hispanic white women whose family income was <150% of the poverty level had significantly higher reports of a non-monogamous partner (13.6%) than women whose family income level was 150–299% of the poverty level (8.1%) or those whose family income was ≥300% of the poverty level (8.5%). Finally, time spent in jail was significantly associated with partners' non-monogamy for all sexually active men (CMH χ2=27.69, df=1, p<0.0001) and for Hispanic (χ2=7.02, df=1, p=0.0096) and non-Hispanic white men (χ2=18.21, df=1, p<0.0001). Approximately 30% of Hispanic and non-Hispanic white men who had spent time in jail in the past 12 months reported partners' non-monogamy compared with approximately 11% of their counterparts who did not spend time in jail.
Similar to findings from the multiple models for non-monogamy, education and poverty level were not significant predictors of partners' non-monogamy in the female and male models (tables 4 and 5). Additionally, current marital/cohabitation status was significantly related to partners' non-monogamy in both female and male models. Although age and race/ethnicity were not significantly associated to partners' non-monogamy for men, in the female model those aged 20–24 years were 1.45 (95% CI 1.10 to 1.92) times as likely as those aged 35–44 years and non-Hispanic black women were 1.37 (95% CI 1.06 to 1.78) times as likely as non-Hispanic white women to report partners' non-monogamy. Finally, men who spent time in jail in the past year were 2.5 (95% CI 1.60 to 3.05) times as likely as men who did not spend time in jail to report partners' non-monogamy.
Mutual non-monogamy in past 12 months
An estimated 8.4 million Americans aged 15–44 years engaged in mutual non-monogamy in the past 12 months; 7.2% of sexually active women and 10.5% of sexually active men reported mutual non-monogamy (table 3). Similar to findings for respondents' non-monogamy and partners' non-monogamy, mutual non-monogamy was significantly associated with age (women: CMH χ2=19.81, df=3, p<0.0001; men: CMH χ2=23.58, df=3, p<0.0001) and marital/cohabitation status (women: CMH χ2=45.36, df=3, p<0.0001; men: CMH χ2=39.51, df=3, p<0.0001). For adolescents, 15.0% of women and 16.3% of men reported mutual non-monogamy. Significant differences for age remained when analysing the data separately by race/ethnicity except for Hispanic men. Formerly and never married women and men had significantly higher reports of mutual non-monogamy—nearly one in five reported mutual non-monogamy. Overall, for women (CMH χ2=6.44, df=3, p=0.000) and men (CMH χ2=3.94, df=3, p=0.0110) there were significant differences in mutual non-monogamy by education level; as education level increased, reports of mutual non-monogamy appeared to decrease. These findings held for non-Hispanic white men and non-Hispanic white and black women.
Sexually active women and men: mutual non-monogamy in past 12 months
Finally, for mutual non-monogamy, the results for poverty level income and time spent in jail (men only) were somewhat similar to those for respondents' and partners' non-monogamy. Poverty income level was not significantly associated with mutual non-monogamy for men (CMH χ2=0.05, df=2, p=0.9468); however, there were significant findings for all sexually active women (CMH χ2=5.88, df=2, p=0.0041) and non-Hispanic white women (χ2=5.68, df=2, p=0.0049). For time spent in jail the findings were nearly identical to those for respondents' and partners' non-monogamy. Overall, sexually active men who had spent time in jail in the past 12 months had significantly higher reports of mutual non-monogamy than their counterparts who did not spend time in jail (CMH χ2=21.82, df=1, p<0.0001). Nearly one in four men who spent time in jail reported mutual non-monogamy compared with only one in 10 men who had not spent time in jail. These findings were similar for Hispanic (χ2=5.75, df=1, p=0.0188) and non-Hispanic white men (χ2=15.10, df=1, p=0.0002); however, time spent in jail was not significantly associated with mutual non-monogamy in non-Hispanic black men (CMH χ2=0.05, df=1, p=0.8168).
Findings from the female and male multiple logistic regression models for mutual non-monogamy were somewhat different from the models for non-monogamy and partners' non-monogamy (tables 4 and 5). Although the younger age groups (15–19 years, 20–24 years, 25–34 years) were 1.5–2.0 times as likely as those aged 35–44 years to report mutual non-monogamy in the female model, only those aged 25–34 years (adjusted OR 1.51, 95% CI 1.10 to 2.09) differed significantly from 35–44-year-olds in the male model for mutual non-monogamy. Education and poverty level income were not significantly associated with mutual non-monogamy in the female and male models, with the exception that women with some college education are 1.86 as likely (95% CI 1.07 to 3.23) to report the behaviour as those with a Bachelor's degree or higher. Once again, current marital/cohabitation status was significantly associated with mutual non-monogamy in both the female and male models. Finally, men who spent time in jail in the past year were 2.32 (95% CI 1.64 to 3.28) times as likely as men who did not spend time in jail to report mutual non-monogamy.
Sexually active women: multiple logistic regression models for each measure of non-monogamy
Sexually active men: multiple logistic regression models for each measure of non-monogamy
Discussion
We analysed patterns and correlates of non-monogamy, partners' non-monogamy and mutual non-monogamy in the USA using nationally representative data. Given the difficulties in the definition and measurement of concurrent partnerships, including problems with reporting of start and end dates of partnerships and the correlation between numbers of partners and concurrency12 and the evidence that short gaps between serially monogamous partnerships may contribute to the spread of STI,15 we believe non-monogamy may be a conceptually clearer parameter which is easier to measure. The three types of non-monogamy may also be easier to understand and use in communications with populations at risk.
Our findings suggest significant patterns in the distribution of non-monogamy, partners' non-monogamy and mutual non-monogamy by demographic and socioeconomic characteristics. Overall, all three types of non-monogamy are reported more frequently by men than women. In line with expectations, for both men and women the most frequently reported type of non-monogamy is the respondent's own non-monogamy followed by partners' non-monogamy. Mutual non-monogamy was least frequently reported, although over 7% of sexually active women of reproductive age and 10.5% of sexually active men of reproductive age reported mutual non-monogamy.
In general, younger age, lower education, formerly or never married status, living below the poverty level and having spent time in jail are associated with all three types of non-monogamy. All three types of non-monogamy are most frequently reported by black women (26.8%, 18.5% and 11.3%, respectively) followed by white women (16.1%, 9.4% and 7.9%, respectively) and Hispanic women (15.6%, 7.9% and 5%, respectively). For men the pattern is somewhat different. While black men report all three types of non-monogamy most frequently (38.9%, 21.3% and 18.7%, respectively), this is followed by Hispanic men (27.6%, 14.0% and 11.7%, respectively) and white men (19%, 11.4% and 9%, respectively).
The study has limitations. Our data are based on self-reports and may be subject to bias. In addition, people may find it difficult to remember whether partnerships happened within the time period covered. Reported non-monogamy on the part of partners may be an underestimate or an overestimate since it may be based on attributions and assumptions as well as knowledge. However, it may be the acknowledged and reported non-monogamy (one's own as well as that of partners) that is most relevant to the spread of STI since it may reflect normalised, openly and frequently practised non-monogamy.
Despite its limitations, our study suggests new ways of approaching partnerships. While the concepts of own, partners' and mutual non-monogamy have not appeared in the peer-reviewed literature, they were included in a PhD dissertation completed at Imperial College, London in the early 1990s.22 This conceptualisation lends itself to the differentiation between the risks for acquisition and transmission and the determination of spread in the population. In addition, these measures may be comparable across time and space, thus facilitating monitoring of temporal trends and cross-country patterns through surveillance data.
The concept of mutual non-monogamy is particularly interesting in that it can help identify sexual network patterns that are most effective in facilitating spread of STI. However, as a concept it is relatively easy to comprehend and may even be easy to measure. In a recent study among individuals in HIV-serodiscordant heterosexual couples, 40% of individuals with concurrent partners had non-monogamous sex partners.23
The patterns of determinants and correlates of non-monogamy suggested in our findings are in line with expectations based on earlier findings regarding levels and distribution of risk behaviour parameters. Analyses of additional independent variables may reveal other results. Contextual factors may play an important role in the determination of patterns of non-monogamy. Parameters such as the sex ratio, the proportion of men who have spent time in jail living in a particular geographical area and the proportion of men and women living in a particular geographical area who are not originally from that area may be interesting to explore as they relate to the prevalence of own monogamy, partners' monogamy and mutual non-monogamy.
The three types of non-monogamy described here may be used in tailoring prevention messages. Risk assessment questions can directly include the three types of non-monogamy. People who report own non-monogamy can be counselled about the risks to which they may expose their sex partners and the importance of using other methods of prevention such as condoms. People who report partners' non-monogamy may be counselled about their own risk of acquiring STI and the importance of using prevention measures. Such a specific and concrete description of their risk situation as related to non-monogamy may increase motivation to practise prevention.
Non-monogamy, particularly mutual non-monogamy, may also be helpful in targeting prevention efforts and resources to subgroups that are most likely to spread STI and HIV in the population. Epidemiological and mathematical modelling work on STI consistently points out the relative importance of core groups in the spread and maintenance of STI in populations.24 Mutual non-monogamy may provide a relatively easy-to-use operational definition for the core group concept. Future research can further illuminate epidemiological and programmatic uses, including strengths and weaknesses, of the three types of non-monogamy discussed here.
Acknowledgments
The authors thank Patricia Jackson for her outstanding support in the preparation of this article.
References
Footnotes
Competing interests None.
Provenance and peer review Not commissioned; externally peer reviewed.
Disclaimer The findings and conclusions in this paper are those of the author(s) and do not necessarily represent the views of the Centers for Disease Control and Prevention.