“MULTIPLE SEXUAL PARTNERS” is a factor that has long been associated with increased risk for transmission of sexually transmitted disease (STD) and HIV at the individual and population levels, 1,2 yet the term is broad and can be interpreted as either serial monogamy or concurrency (more than one partner at a time). Risk of STD acquisition depends less on the number of sex partners than on the probability of encountering an infected sex partner, which in turn depends on the prevalence of infection in the population. 3 Because STD prevalence varies in different subpopulations, sexual mixing between these various subpopulations is an important factor in transmission dynamics. When infected and uninfected individuals mix, transmission can occur through 1 serial monogamy, wherein the gap between partnerships is shorter than the duration of the infection, or 2 concurrent partnerships, wherein pathogens can be passed from one partner to the other.
In the early 1990s, the role of concurrent partnerships in escalating the spread of HIV infection was first identified by Watts and May 4 in mathematical models that accounted for duration and overlapping partnerships. Morris and Kretzschmar 5,6 subsequently developed stochastic simulation models and demonstrated that concurrent partnerships can amplify nascent HIV epidemics by as much as 10-fold. Other simulation models have shown that establishment of endemic infection is best predicted by the proportion and connectedness of concurrent partnerships in a population. 7 However, the net effect of concurrent partnerships on transmission dynamics also depends on the overall number of partners and the duration of sexual partnerships. 8 The importance of concurrency to transmission dynamics is highlighted by the observation that a high prevalence of concurrent partnerships in the sexually active population ensures the persistence of infection in low-risk groups. 9 Furthermore, Potterat 10 has shown that concurrency is a stronger risk factor for chlamydial infection than the number of partners, even when each of these factors is adjusted for the other.
Sex partner concurrency has been measured in a variety of ways. Information on the date of first and last sexual encounters with consecutive sex partners has been gathered and used to indirectly calculate the overlap in the National Health and Social Life Survey, 11 in a study assessing factors associated with transmission of chlamydial infection, 10 in a study on risk factors for HIV transmission, 12 and in a substudy of the National Survey of Family Growth. 13 Other studies have defined concurrency as having more than one sex partner during the previous 3 months, 14 multiple partners in the previous 30 days, 15 or evidence of at least two partnerships within a 2-week period. 16 Using coital diaries, Howard et al 17 defined concurrency as having two or more partners within the period covered by the diary, including a subsequent coital event with a previous partner. Morris coined the term bridging to describe the phenomenon in which individuals with concurrent partners link high- and low-risk groups, serving as a conduit for the spread of HIV. 18
Although empirical data have linked concurrent sex partnerships to increased risk for STD, 10,15,16 most of these studies have involved high-risk individuals (STD clinic attendees, infected with one or more STDs). The prevalence and correlates of concurrent partnerships in the general population remain largely undefined. Only one other study has evaluated the population prevalence of concurrency; that study used the National Survey of Family Growth and showed that approximately 25% of women aged 15 years to 44 years reported having concurrent sex partners, 13 defined by overlapping dates. There are no data on the prevalence of concurrency among men in the general population, and no empirical comparison of different measures of concurrency has been reported. In this article we examine different methods of measuring sex partner concurrency, estimate the frequency of concurrent partnerships within an urban United States population, and identify correlates of sex partner concurrency.
In February 1995, we conducted a random-digit-dial telephone survey of Seattle residents aged 18 years to 39 years (n = 544). Eligibility criteria included current residence in Seattle, age 18 years to 39 years, and fluency in the English language. Potential respondents were told the purpose of the study was “to learn how often people engage in behaviors that might put them at risk of acquiring an STD.” A previous report presented further details of the sampling and interviewing process. 19 The random-digit-dial survey sample included listed and unlisted numbers and was obtained from Survey Sampling, Inc., of Westport, Connecticut. A further sample of black respondents was added (n = 144) to provide sufficient numbers to explore reported differences in STD rates and sexual behavior by race. 11,20,21 The sampling frame for the black oversample was telephone directory listings in Seattle census block groups whose black population was 40% or greater. Up to six attempts were made to contact each number, at different times of the day.
Of the 3600 telephone numbers in the initial sampling frame, 2049 (56.9%) yielded successful contact; 1077 numbers (29.9%) were either nonresidential or disconnected, 245 (6.8%) connected only to answering machines, and 229 (6.4%) yielded no answer. Of the 2049 persons contacted, 1241 (60.6%) did not meet eligibility requirements, leaving 808 eligible individuals. Five hundred forty-four of the 808 eligible individuals contacted (67.3%) agreed to participate and completed the interview. In the black oversample, from the 2800 telephone numbers in the original sampling frame, we were able to contact 2122 persons (75.8%), of whom 510 (24.0%) met the eligibility requirements; 312 telephone numbers (11.1%) were nonresidential or disconnected, 160 (5.7%) connected only to answering machines, 206 (7.4%) yielded no answer, and 1612 (76.0%) yielded contact with ineligible persons. One hundred thirty-five of the eligible respondents (28.2%) agreed to participate and completed the interview.
The survey instrument included questions on lifetime sexual history, partner and partnership characteristics, STD history, and demographics. It was pretested on a sample of the study population and revised before initiation of data collection. The telephone survey required approximately 20 minutes to conduct and was administered with use of computer-assisted telephone interviewing software, which standardized the interview and minimized data entry errors. Respondents were given the option to select a male or female interviewer, among the 31 interviewers available.
This analysis was restricted to sexually active individuals whose most recent sexual partnership was heterosexual. Therefore, 51 (7.4%) of the 688 individuals in the original sample were excluded (43 [6.3%] had not yet initiated sexual activity and 8 [1.2%] had no opposite-sex relationships), resulting in a total of 637 respondents available for analysis.
Descriptive analyses of characteristics related to concurrency were performed with contingency tables and chi-square tests to assess significant differences between groups. An extension of the Kruskal-Wallis test was used to assess trend. 22 Multivariate logistic regression was used to describe correlates of concurrency. Because we hypothesized that characteristics of the individual that were associated with concurrency might differ from characteristics of the partnership unit, one set of multivariate models was developed to identify factors associated with concurrency at the individual level (stratified by sex), and another model was developed for partnership level factors. The Stata 22 software package was used in summarizing the data and conducting the analyses.
Definition of Concurrency
To assess prevalence of and compare different methods of defining concurrency, we collected three types of information (Table 1). Respondents were asked about (1) the number of people with whom they had had sexual contact while they were sexually involved with their most recent partner (concurrent partners); (2) start and end dates (month and year) of their most recent and next-to-last sexual relationships (overlapping dates); and (3) whether they believed their partner had other partners during the sexual relationship (see Table 2 for text of questions).
Individual concurrency describes the actions of the respondent alone, whereas partnership concurrency describes the actions of both the respondent and the partner. When asked directly if they had any other partners during their sexual relationship with their most recent partner (concurrent partners), more men than women were classified as reporting individual concurrency (27% versus 18%;P = 0.01). This direct question also was associated with the least amount of missing data (only 10%). Under the overlapping dates definition of concurrency (overlap), 21% of men reported concurrent partnerships, whereas only 12% of women did so. However, many respondents did not furnish information on start and end dates of sexual relationships, and 19% of responses to these questions were missing. A limitation to the use of this measure is the potential for misclassification. Since the “dates” for sexual contact were provided by month rather than actual calendar date, an individual whose last sexual encounter with partner A occurred early in the month and whose first sexual encounter with partner B occurred late in the same month would be classified as concurrent in this scheme, when in fact there could be a gap of almost 4 weeks between partners. Furthermore, the κ for comparison of the multiple concurrent partners measure to the overlapping dates measure was low (0.34), indicating there was little agreement between the two measures and that perhaps they were capturing different things. Although we also explored a combination of the overlap and concurrent partner variables by creating a composite measure, this included the biases inherent in the overlap variable and probably overestimated concurrency. Ultimately, we selected the concurrent partners question to define individual concurrency in these analyses because it captured the majority of individuals identified as concurrent by the other definitions, had the fewest missing data, and had the least likelihood of misclassification. With use of this definition, 22% of all respondents reported concurrent partners.
A measure of partnership concurrency was devised to combine information about the respondent (concurrent partners) with information about the partner (respondents’ belief their partner had other partners during the relationship). Slightly more women than men believed that their partners had other partners (18% versus 15%), consistent with the fact that more men than women reported concurrent partnerships (31% versus 26%). Although we explored a composite measure of partnership concurrency including information on overlapping dates, this was subject to the same limitations as the composite measure of individual concurrency, and we elected not to use it. In initial analyses we also separated partnership concurrency into two groups: (1) only one partner had other concurrent partners or (2) both partners had other concurrent partners. These two types of partnership concurrency were then compared with mutually monogamous partnerships. However, both individuals had other partners in only 53 partnerships (8%), and the factors associated with both categories of partnership concurrency were similar. Therefore, we defined partnership concurrency to reflect the minimum required for disease transmission (i.e., either of the partners had other sex partners during the relationship), and this was reported by 28% of respondents.
The relationship between individual characteristics and individual concurrency differed for men and women (Table 3) Divorced, widowed, or separated men were more likely than married men to report concurrent partners (37% versus 15%;P = 0.03), whereas marital status did not differentiate women with concurrent sex partners from monogamous women (15–19% concurrent, irrespective of marital status). Unemployed men reported concurrent partners slightly more often than employed men (36% versus 24%), whereas unemployed women reported concurrent sex partners slightly less often than employed women (12% versus 19%). Men who reported ever having spent a night in jail had concurrent partners more often than men who had never been in jail (44% versus 20%;P < 0.01), but this difference was less striking for women (27% versus 16%;P = 0.14). Young age at first sexual experience was strongly associated with concurrency in women (34% with first coitus before age 16 years, versus 11% after age 18 years;P < 0.01) but less so in men (31.3% before age 16 years, versus 22.2% after age 18 years). Both men and women who reported ever having had same-sex partners were about two times more likely to report concurrency (60.0% versus 25.4% [P = 0.015] for men; 28.6% versus 16.9% [P = 0.176] for women).
For both men and women, those with the fewest lifetime number of partners 1–3 were least likely to report concurrent partners (13% of men and 7% of women), with steady increases in the proportion concurrent up to 15 or more lifetime partners (52% of men and 29% of women). Men and women who engaged in a wider variety of sexual practices (oral and anal intercourse) were also more likely to have concurrent partners. A self-reported diagnosis of STD in the respondent during the most recent relationship was strongly associated with reports of concurrent partners among women (36%, versus 14% without an STD diagnosis;P = 0.001) but less so among men (36% versus 26%;P = 0.40). By contrast, history of STD before the most recent relationship was strongly associated with concurrency among men (41% versus 23%;P = 0.015) but had little effect on concurrency among women (16.9% versus 13.3%). Age, annual income, educational level, and race were not associated with individual reports of concurrent partners.
Further stratification by lifetime number of partners did not appreciably change the relations between individual characteristics and concurrency. Concurrency rates were highest for men with the greatest number of partners and increased as lifetime number of partners increased, in a stepwise fashion in most cases, irrespective of the individual factor of interest (range of 29.4–83.3% for those with 15 or more partners, versus 5.6–36.4% for those with 1–3 lifetime partners). A similar trend was observed among women, although the stepwise increase by lifetime number of partners was not as strong or as consistent as that observed among men (range of 16.7–50.0% for those with 15 or more partners, versus 2.8–33.3% for those with 1–3 lifetime partners).
Partnership concurrency, as measured by self-report of concurrent partners and/or belief the partner also had other concurrent partners, was reported by 28% of this study population. Assuming that sexual behavior changes as individuals gain experience with each successive partnership, situating the current partnership in the context of previous ones (determining “partnership order”) is likely to be important in understanding determinants of concurrency. Because we had data on the last two partnerships and the lifetime number of partners, we were able to determine the partnership order of the most recent partnership. Given the potential importance of partnership order (represented here by lifetime number of sex partners), analyses of partnership concurrency were stratified by this factor to account for its effect. As partnership order increased, the proportion of concurrent partnerships also increased for almost every subset of partnership characteristics examined (Table 4). The exception to this is the time elapsed between meeting and engaging in sexual intercourse, which was short among concurrent partners in early relationships (36% of concurrent partnerships engaged in intercourse within 1 week) but longer in later relationships (60% of concurrent partnerships waited >6 months, whereas only 40% of monogamous partnerships waited >6 months [data not shown]). In addition, when couples were married or living together and in partnerships in which the partner got to know the respondent's family, concurrency occurred less often than in partnerships without these social ties, both in early relationships (9% versus 21% and 11% versus 25%, respectively) and (even more so) in later partnerships (44% versus 50% and 50% versus 39%, respectively).
Irrespective of partnership order, concurrency was reported more frequently in couples of discordant race (35% versus 23%;P = 0.002), among infrequent condom users (32% versus 23%;P = 0.025), among respondents who had an STD diagnosed during the relationship (50% versus 26%;P < 0.01), in partnerships where either partner had spent a night in jail (46% versus 21%;P < 0.01), and in partnerships of longer duration (approximately 33% in partnerships of >2 years versus approximately 20% in those of 1 year or less;P = 0.019). Concurrency was not significantly elevated in those partnerships characterized by educational or age differences or among those who met in an anonymous venue.
Three separate multivariate models were developed to identify the factors associated with individual and partnership concurrency (Table 5). The first two models were restricted to individual characteristics of respondents, by sex, and their association with reports of concurrent partners. Both models included factors associated with individual concurrency in either men or women to provide comparisons by sex. The third model describes the effect of partnership characteristics on the likelihood that at least one of the partners would have other concurrent partners (partnership concurrency); partnership variables associated with partnership concurrency were included in the model.
Individual Models of Concurrency
For men, the odds of having concurrent partners increased 15% for each partner accumulated, up to 15 (odds ratio [OR], 1.15; 95% CI, 1.07–1.23; descriptive analyses indicated that the proportion concurrent remained relatively constant for more than 15 partners. Men who had spent a night in jail were twice as likely (OR, 1.99; 95% CI, 1.03–3.82) to have concurrent partners as those who had not. Although the finding was not statistically significant, men who had ever had same-sex partners were more likely to have concurrent partners than those who reported only opposite-sex partners. Men who had first sexual intercourse before age 16 years were somewhat less likely than others to have concurrent partners, as were those employed at the time of the survey, in comparison with unemployed men. Neither age nor diagnosis of STD during the most recent sex partnership was significantly associated with concurrent partners in men.
For women, the strongest predictor of having concurrent partners was a self-reported STD diagnosis during the relationship. Such women were three and one half times more likely to have concurrent partners than those who had no STD diagnosed (OR, 3.53; 95% CI, 1.55–8.05). In contrast to men, women who initiated sexual activity before age 16 years were almost three times more likely to report concurrent partners than women whose sexual debut occurred at a later age (OR, 2.90; 95% CI, 1.38–6.10), and employed women were somewhat more likely than unemployed women to have concurrent sex partners. Similar to men, the odds of having concurrent partners increased by 9% with each lifetime partner acquired, up to 15 (OR, 1.09; 95% CI, 1.01–1.16). Having spent a night in jail, having had a same-sex partner, and age were not significantly associated with concurrency among women.
Partnership Models of Concurrency
In this multivariate analysis, partnerships in which the respondent was diagnosed with an STD were more than two and one half times as likely to be concurrent as those in which no infections were diagnosed (OR, 2.68; 95% CI, 1.42–5.07). Similarly, partnerships in which either of the partners ever spent a night in jail were twice as likely to be concurrent as those in which neither partner had been in jail (OR, 2.04; 95% CI, 1.32–3.17). Couples married or living together were approximately 40% less likely than other partnerships to be concurrent, and partnerships in which the partner got to know the respondent's family were almost 30% less likely than others to be concurrent. Partnerships whose duration was greater than 6 months were nearly two and one half times more likely to be concurrent than those of shorter duration (OR, 2.43; 95% CI, 1.41–4.19). Adjusting for the rate of partner change (lifetime number of partners/total years of sexual activity) did not appreciably change this estimate. Partnerships of persons of different races were 70% more likely to be concurrent than those in which the partners were of the same race.
In this random-digit-dial telephone survey of the Seattle general population aged 18 years to 39 years, we studied the measurement, frequency, and correlates of individual and partnership concurrency. We observed that direct reports from concurrent partners provided the most complete data with the lowest potential for misclassification as a measure of individual concurrency. The disparities between the overlapping dates and the concurrent partner measures may be due to respondents’ thinking in terms of established relationships when asked about overlapping dates and including “one-night stand” activity when asked directly about concurrent sex partners, although this merits further study.
We did not explore problems in recall of concurrent partners or overlapping dates of partnerships. However, Brewer et al 23 recently reported on the utility of using repeated prompting and reiterating subjects’ answers to improve recall of past sex partners. Such techniques may reduce recall bias and should be considered in future studies of concurrent sex partners. Daily diaries, such as those used by Howard et al, 17 provide a more optimal measure of individual concurrency but are burdensome and not feasible in many contexts. An ideal measure of concurrency would involve ascertainment of complete sexual networks. Concurrency could then be estimated from the mean and variance of the degree in the network. 5,6 However, this requires that some boundary for the population of interest be established and that all sexual partnerships within that boundary be identified. To date, only one published report has approached this level of detail, 10 and the relative completeness of the data were possible only because of official efforts by the health department to trace contacts. Most survey studies will not be able to command similar authority or resources; thus, in most cases, complete sexual network assessment remains largely a theoretical proposition. In a questionnaire setting, therefore, a more accurate measure of concurrency might include both direct questions (concurrent partners) and indirect questions (overlapping dates), as well as a programmed system to identify and track discrepancies in the two responses during the interview. An interviewer could then systematically resolve any disparities between the two measures with the respondent at the time of data collection.
The epidemiologic relevance of definitions of concurrency will vary according to the pathogen of interest. Infections of very short duration will require a more stringent operational definition of concurrency (e.g., a very short time between partners for the infection to be transmitted) than infections of a long duration. Because our objective was to estimate the proportion of concurrent partnerships in the population as a measure of established transmission pathways, we defined concurrency globally, irrespective of any specific pathogen.
More than one fifth (22%) of respondents reported individual concurrency, and 28% reported that either they or their partner had concurrent partnerships (partnership concurrency). Characteristics associated with individual concurrency differed somewhat from those associated with partnership concurrency, but both measures are relevant and can be used to identify potential concurrent partnerships.
Individual and partnership concurrency was independently associated with an STD diagnosis in the most recent partnership, even after adjustment for lifetime number of sex partners, linking concurrency to individual STD risk. Although this is not a unique finding, 10,15,16 previous studies were conducted in high-risk STD clinic populations, made adjustments for fewer factors, and were designed to assess predictors of STD infection rather than evaluate correlates of concurrency itself.
The frequency and correlates of concurrency varied strikingly by sex. Men more often reported concurrent partners than did women, consistent with previous observations 9,16 and with parallel evidence that factors associated with STD risk differ significantly between men and women. 11,24,25 Most significant predictors of individual concurrency were different in men and women (factors associated with individual concurrency for men were lifetime number of partners, having spent a night in jail, and ever having had same-sex partners; factors associated with individual concurrency for women were lifetime number of partners, age at first sex <16 years, and STD diagnosis during the relationship), and some effects were in opposite directions. These differences may be due to real behavioral differences between men and women or to reporting differences, which may arise from differential perception of the context. We speculate that men may focus more on individual sex acts and thus recall more concurrent partners, whereas women may focus more on a relationship and neglect to report intermittent sexual encounters outside the context of an established relationship. Alternatively, differential social desirability bias may influence reporting accuracy and motivate men to report more concurrent sex partners, while at the same time discouraging women from reporting them.
It is somewhat surprising that sociodemographic factors strongly associated with STD risk did not predict individual concurrency (e.g., age, race, socioeconomic status, and education). Similarly, on the partnership level, despite the fact that partnerships discordant on race, age, education, and sexual activity level have been associated with increased rates of STD, 25–27 only race-discordant partnerships were predictive of concurrent partnerships. Thus, some determinants of concurrency likely differ from determinants of STD, again highlighting the importance of underlying population prevalence as well as mixing and sexual network patterns in understanding disease transmission.
The increased likelihood of concurrency associated with either partner ever having spent a night in jail may be due to seeking out a replacement during those periods when the partner is unavailable, a phenomenon previously noted in qualitative analyses of concurrency. Gorbach et al have identified six distinct types of concurrency: experimental, separational, transitional, reciprocal, reactive, and compensatory, with the most common type being experimental (short-term overlapping partnerships occurring as individuals explore new partnerships). In many of these concurrency types, the common denominator is one of social context: a shortage of monogamous main partners. 28 Alternatively, spending time in jail may also be a marker of higher risk behavior in general.
Partnership concurrency incorporates the actions of both partners and was associated with race discordance, relationship duration of >6 months, married/living together, either partner having spent a night in jail, and an STD diagnosis during the relationship. Increasing partnership order (or its proxy, lifetime number of partners) was strongly associated with both individual and partnership concurrency, yet partnership characteristics associated with concurrency varied by partnership order, consistent with earlier observations of a correlation between partnership order and partnership characteristics. 29
The observation that longer-term partnerships were associated with increased concurrency may be partially explained by the increased opportunity to have concurrent partners over time. Alternatively, concurrent partnerships may typically occur in the context of long-term relationships, whereas shorter-term partnerships might be more often characterized by serial monogamy. That neither age nor a difference in age between partners was associated with concurrency may be a function of the restricted age group (18–39 years) in this study population. While race was not associated with concurrent partnerships, race difference between the two partners was. Cross-racial partnerships may be exploratory in nature, although there was no difference in the duration of race-discordant partnerships in comparison with that of race-concordant partnerships.
There are several limitations to this study. Although response rates were within the norms for survey research (68%), approximately one third of individuals contacted did not agree to an interview. In addition, although the sampling frame can be considered representative of the Seattle population aged 18 years to 39 years (96.0% of Seattle residents speak English and 97.4% of households have telephones, per 1990 U.S. Census data), individuals we were unable to contact may have been different from those we reached, a circumstance potentially introducing bias. Furthermore, it is likely that those we succeeded in contacting and who were eligible and agreed to participate were somewhat different from those who were eligible yet refused to participate, which may have introduced a degree of volunteer bias into the study. For example, if those who chose to participate had more liberal views toward sexuality, they may also have been more likely to report concurrent partners than those who declined to participate. This would lead to overestimation of the true prevalence of concurrent partnerships in the population, and to the extent that participation is also associated with correlates of concurrency, it may bias the estimates of these correlates.
Although these data provided the opportunity to compare several different measures of individual concurrency, the analysis is largely descriptive, and the study was not designed to measure the relative efficacy of these different measures. Further studies, expressly designed to determine the validity and reliability of a broader range of measures of individual concurrency, will be necessary to determine the optimal measure.
The partnership analysis relies on surrogate reports of partner characteristics, the accuracy of which is likely to vary, depending on the characteristic and the respondent. Males have demonstrated better agreement than females in their reports of partners’ education level, and male partners often disagree on whether their female partner is a casual or regular partner. 30 Agreement between respondents and their partners is good for race/ethnicity (90%) and age (81%); fair for education (67%) and estimated lifetime number of partners (65%); and poor for the partner's number of other partners over the previous 3 months (only 35% agreement). 25 Therefore, respondents’ assessments of whether their partners had other, concurrent partners may have been incorrect, resulting in misclassification, which would bias our estimates.
We did not collect information on qualitative aspects of partnerships (e.g., casual versus main partners or measures of intimacy) and therefore cannot assess the degree to which these factors may be associated with concurrency. Other studies have shown that an individual's sexual behavior and condom use vary by partner type, 31–33 suggesting that partner type may also be related to concurrency and should be explored in other studies.
At the population level, concurrent sex partnerships accelerate the spread of STDs. At the individual level, the impact of concurrent sex partners probably depends on the network to which an individual becomes connected and the extent to which condoms are used. A partner's unsuspected concurrency could put an individual at higher risk for STD than if the individual had concurrent sex partners of his/her own, because in such cases the individual might take protective precautions less often. Concurrency is therefore an important factor in transmission dynamics but interdependent with other essential factors (prevalence of infection in the population, duration of specific infections, an individual's awareness of partnership concurrency, total number of partners). Future work should include expanding data collection from the local network (egocentric) model to include direct data collection from partners. By increasing the accuracy of measures of both concurrent sex partnerships and the factors correlated with concurrency, we will be better equipped to tease out the role of concurrency in STD transmission dynamics. The strongest correlates of concurrency observed in this population were behavioral rather than individual demographic factors. This suggests that forces above and beyond individual demography influence individual behavior and that both population demography and social structural contexts may be required to explain concurrency and assess its impact on STD risk.
1. Seidman SN, Mosher WD, Aral SO. Women with multiple sexual partners: United States, 1988. Am J Public Health 1992; 82: 1388–1394.
2. Anderson RM, May RM. Epidemiological parameters of HIV transmission. Nature 1988; 333: 514–519.
3. Aral SO, Holmes KK. Social and behavioral determinants of the epidemiology of STDs: industrialized and developing countries. In: Holmes KK, Sparling PF, Mardh P-A, et al, eds. Sexually Transmitted Diseases. New York: McGraw-Hill, 1999:39–76.
4. Watts CH, May RM. The influence of concurrent partnerships on the dynamics of HIV/AIDS. Math Biosci 1992; 108: 89–104.
5. Morris M, Kretzschmar M. Concurrent partnerships and the spread of HIV. AIDS 1997; 11: 641–648.
6. Morris M, Kretzschmar M. Concurrent partnerships and transmission dynamics in networks. Social Networks 1995; 17: 299–318.
7. Ghani AC, Swinton J, Garnett GP. The role of sexual partnership networks in the epidemiology of gonorrhea. Sex Transm Dis 1997; 24: 45–56.
8. Garnett GP, Johnson AM. Coining a new term in epidemiology: concurrency and HIV. AIDS 1997; 11: 681–683.
9. Anderson RM. Transmission dynamics of sexually transmitted infections. In: Holmes KK, Sparling PF, Mardh PA, et al, eds. Sexually Transmitted Diseases. New York: McGraw-Hill, 1999:25–38.
10. Potterat JJ, Zimmerman-Rogers H, Muth SQ, et al. Chlamydia transmission: concurrency, reproduction number, and the epidemic trajectory. Am J Epidemiol 1999; 150: 1331–1339.
11. Laumann EO, Gagnon JH, Michael RT, Michaels S. The Social Organization of Sexuality: Sexual Practices in the United States. Chicago: The University of Chicago Press, 1994.
12. Serwadda D, Wawer MJ, Musgrave SD, Sewankambo NK, Kaplan JE, Gray RH. HIV risk factors in three geographic strata of rural Rakai District, Uganda. AIDS 1992; 6: 983–989.
13. Kraut JR, Aral SO. Sexually transmitted diseases and the duration between partnerships (“gap”) (abstract no. 033). Presented at the STIs at the Millenium: Past, Present, and Future. Baltimore, MD. May 3–7, 2000.
14. Colvin M, Abdool Karim SS, Connolly C, Hoosen AA, Ntuli N. HIV infection and asymptomatic sexually transmitted infections in a rural South African community. Int J STD AIDS 1998; 9: 548–550.
15. Daker-White G, Barlow D. Heterosexual gonorrhoea at St Thomas’—II: sexual behaviour and sources of infection. Int J STD AIDS 1997; 8: 102–108.
16. Rosenberg MD, Gurvey JE, Adler N, Dunlop MB, Ellen JM. Concurrent sex partners and risk for sexually transmitted diseases among adolescents. Sex Transm Dis 1999; 26: 208–212.
17. Howard MM, Fortenberry JD, Blythe MJ, Zimet GD, Orr DP. Patterns of sexual partnerships among adolescent females. J Adolesc Health 1999; 24: 300–303.
18. Morris M, Podhisita C, Wawer MJ, Handcock MS. Bridge populations in the spread of HIV/AIDS in Thailand. AIDS 1996; 10: 1265–1271.
19. Foxman B, Aral SO, Holmes KK. Interrelationships among douching practices, risky sexual practices, and history of self-reported sexually transmitted diseases in an urban population. Sex Transm Dis 1998; 25: 90–99.
20. Division of STD Prevention, U.S. Department of Health and Human Services, Public Health Service. Sexually Transmitted Disease Surveillance, 1997. Atlanta: Centers for Disease Control and Prevention, 1998.
21. Rice RJ, Roberts PL, Handsfield HH, Holmes KK. Sociodemographic distribution of gonorrhea incidence: implications for prevention and behavioral research. Am J Public Health 1991; 81: 1252–1258.
22. STATA, version 6.0. College Station, TX: Stata Corporation, 1999.
23. Brewer DD, Garrett SB, Kulasingam S. Forgetting as a cause of incomplete reporting of sexual and drug injection partners. Sex Transm Dis 1999; 26: 166–176.
24. Aral SO, Soskoline V, Joesoef RM, O'Reilly KR. Sex partner recruitment as risk factor for STD: clustering of risky modes. Sex Transm Dis 1991; 18: 10–17.
25. Aral SO, Hughes JP, Stoner B, et al. Sexual mixing patterns in the spread of gonococcal and chlamydial infections. Am J Public Health 1999; 89: 825–833.
26. Garnett GP, Hughes JP, Anderson RM, et al. Sexual mixing patterns of patients attending sexually transmitted diseases clinics. Sex Transm Dis 1996; 23: 248–257.
27. Rothenberg RB, Potterat JJ. Temporal and social aspects of gonorrhea transmission: the force of infectivity. Sex Transm Dis 1988; 15: 88–92.
28. Gorbach PM, Stoner BP, Aral SO, Coronado N, Connor S, Holmes KK. It takes a village: understanding concurrent partnerships in Seattle, WA. Sex Transm Dis. In press.
29. Lindblade K, Foxman B, Koopman JS. Heterosexual partnership characteristics of university women. Int J STD AIDS 1994; 5: 37–40.
30. Ellish NJ, Weisman CS, Celentano D, Zenilman JM. Reliability of partner reports of sexual history in a heterosexual population at a sexually transmitted diseases clinic. Sex Transm Dis 1996; 23: 446–452.
31. Lansky A, Thomas JC, Earp JA. Partner-specific sexual behaviors among persons with both main and other partners. Fam Plann Perspect 1998; 30: 93–96.
32. Macaluso M, Demand MJ, Artz LM, Hook EW 3rd. Partner type and condom use. AIDS 2000; 14: 537–546.
33. Plichta SB, Weisman CS, Nathanson CA, Ensminger ME, Robinson JC. Partner-specific condom use among adolescent women clients of a family planning clinic. J Adolesc Health 1992; 13: 506–511.