Sexual Contact Patterns in High-income Countries A Comparative Analysis Using Data From Germany, The UK, and The US

Sexual contact patterns determine the spread of sexually transmitted infections, and are a central input parameter for mathematical models in this eld. We evaluated the importance of country-specic contact pattern parametrization for high-income countries with similar cultural backgrounds by deriving estimates for sexual contact patterns in Germany from two independent studies (HaBIDS and SBG), and comparing them to data from the National Survey of Sexual Attitudes and Lifestyles (Natsal) in the UK, and the National Survey of Family Growth (NSFG) in the US. UK and US participants reported a substantially higher number of lifetime opposite-sex sexual partners compared to both German studies. The difference in lifetime partners was caused by a higher proportion of individuals with many partners in the young age group (< 24 years) in the UK and the US. Partner acquisition in older age groups was similar. The number of same-sex partners was similar across the countries, while there was heterogeneity in the reported experience with partners from both sexes, consistent with the differences observed for opposite-sex sexual partners. These patterns can lead to substantially different dynamics of sexually transmitted infections across age, and have strong impact on the results of modelling studies. This study uses data from the Hygiene and Behaviour Infectious Diseases Survey (HaBIDS) panel which included individuals from four counties in the Federal State of Lower Saxony, Germany (Braunschweig, Salzgitter, Vechta, and Wolfenbüttel) and was run between 2014 and 2017 10 . HaBIDS was designed to study knowledge, attitudes, and practice towards infectious diseases from various perspectives 10–12 . Potential participants aged between 15 and 69 years were selected from the population registries of the respective counties by proportional stratied random sampling, and were invited by mail to participate initially in a mixed method (paper-pencil and online) and later on in an online only panel. A series of questionnaires regarding behaviour in the context of infectious diseases including sexual contact behaviour was sent to all participants. Data on sexual contacts were collected using paper-based (in July 2014) and online-based questionnaires (in June 2015) . Further details on the HaBIDS study and its methodology can be found in Rübsamen et al 10 . At the initial stage, the HaBIDS panel consisted of 2,379 individuals from a registration oce sample of 26,895 (9% participation rate). At the time of data collection on sexual contact behaviour, 123 individuals had already resigned from HaBIDS, 740 individuals did not send back the questionnaire and 318 individuals either found the questionnaire too personal, returned an empty questionnaire, or gave no information on the number of sexual partners, so that 1,198 participants were left for this analysis.


Introduction
Sexually Transmitted Infections (STIs) are caused by over 30 different bacteria, viruses and parasites 1 , and are a major threat to global public health with more than 350 million curable STIs cases 1 diagnosed each year. Behavioural factors associated with the number of sexual partners, the concurrency of sexual partnerships, and the use of preventive barrier devices like condoms are crucial for the transmission dynamics of STIs 2 . The so-called sexual contact patterns are subject to change, and can depend on cultural habits, religious beliefs, and societal norms 2,3 .
Several studies have assessed the relationship between sexual contact patterns and various sexually transmitted infections 2,4−6 , and were e.g. crucial in the understanding of the role of concurrent partnerships for the heterosexual spread of HIV in Sub-Saharan Africa 6 . Information about sexual contact patterns is important for the parametrization of mathematical models studying STIs transmission, which are used to guide prevention efforts. However, there is little systematic evidence for population-based estimates on the frequency of various types of sexual contact patterns in many geographic and cultural settings. The scarcity of research is often attributed to the fact that surveys on sexual contacts are generally sensitive and privy topics for discussion.
However, there are few large-scale projects like Natsal (National Survey of Sexual Attitudes and Lifestyles, Natsal 7 ) in the United Kingdom and NSFG in the United States (National Survey of Family Growth 8, 9 ), which provide long-term high-quality data for selected high-income countries. Sexual contact patterns reported in these two projects are commonly used for modelling STIs transmission in countries without systematic data collection, like e.g. Germany. The main objective of this study is to provide populationlevel estimates for sexual contact patterns in Germany, and to compare them to data obtained from Natsal and NSFG for the UK and US. Page  Salzgitter, Vechta, and Wolfenbüttel) and was run between 2014 and 2017 10 . HaBIDS was designed to study knowledge, attitudes, and practice towards infectious diseases from various perspectives [10][11][12] . Potential participants aged between 15 and 69 years were selected from the population registries of the respective counties by proportional strati ed random sampling, and were invited by mail to participate initially in a mixed method (paper-pencil and online) and later on in an online only panel. A series of questionnaires regarding behaviour in the context of infectious diseases including sexual contact behaviour was sent to all participants. Data on sexual contacts were collected using paper-based (in July 2014) and online-based questionnaires (in June 2015) . Further details on the HaBIDS study and its methodology can be found in Rübsamen et al 10 . At the initial stage, the HaBIDS panel consisted of 2,379 individuals from a registration o ce sample of 26,895 (9% participation rate). At the time of data collection on sexual contact behaviour, 123 individuals had already resigned from HaBIDS, 740 individuals did not send back the questionnaire and 318 individuals either found the questionnaire too personal, returned an empty questionnaire, or gave no information on the number of sexual partners, so that 1,198 participants were left for this analysis.
In the questionnaire, participants were asked if they had a sexual contact ever in their life, at which age they had their rst sexual contact, how many sexual partners (opposite-and same-sex) they had during their entire life, during the last 12 months and how many new sexual partners they had during the last 12 months. Moreover, participants were asked to report the age and sex of their last three sexual partners which we refer to as last/current partner, penultimate partner, and pre-penultimate partner. We compared HaBIDS data to sexual contact patterns from a recent nationwide survey on sexual behaviour in Germany (SBG) 13 (Table 1). Sexual contact patterns (in contrast to other sexual behaviour data) from this survey were not published before but were provided to us by the SBG authors. SBG participants were aged between 14 and 99 years at data collection in 2016. The included households were identi ed by subdividing the regional areas using the random route method; potential participants were selected using the kish selection grid 13 . We also compared HaBIDS and SBG to two long-term projects from the UK (Natsal) 7 and the US (NSFG) 8,9 (Table 1). Natsal data used in our analysis were collected between 2010 and 2012 (Natsal 3 wave). Natsal used a multistage, clustered and strati ed probability sample design and enrolled participants between the ages of 16 and 74 years 7 . Data used in this publication from the NSFG study were sampled in the 2006 to 2010 survey. NSFG included participants aged 15 to 44 years based on an independent, national probability sampling 8, 14 . For all studies, individual anonymized primary data were provided and used for further analyses.
Sexual contact patterns were compared between the four studies; the questions to collect these data were similarly phrased across the studies. Data on the number of lifetime sexual partners were available in all four studies. Data on the number of sexual partners in the last 12 months were available for HaBIDS, Natsal, and NSFG; data on the number of new partners in the last 12 months were available for HaBIDS and Natsal. No data were available for NSFG participants above the age of 44 years.
Participants who reported at least one opposite-sex sexual partner and at least one same-sex sexual partner are referred to as participants with "both-sex experience".

Data analysis
Absolute and relative frequencies and medians/means were used for descriptive analyses. Differences in the distribution of reported number of opposite-sex partners, same-sex partners and both-sex partners were investigated using negative binomial regression adjusted for age and sex (as well as strati ed by age). Local regression methods were used to visualize the relationship between the number of sexual partners and the corresponding age of participants in a smoothed way. All analyses were performed using R-3.6.1 15 with the packages ggplot2 16 and MASS 17 .

Characteristics of the samples
Of the 1,198 HaBIDS participants included in this study, 748 were female (62%) and 450 were male (38%). The median age was 47 years (interquartile range [IQR] 35-57 years); 36 participants (3%) were below the age of 20 years and 213 participants (18%) were over 60 years old. SBG participants had a median age of 47 years (IQR 32-58), while it was 34 years (IQR 24-52) for Natsal participants and 28 years (IQR 21-36) for NSFG participants. Further information on the study populations can be found in Table 1.

Current partnership and age at sexual debut
There were no relevant differences in sexual contact patterns between HaBIDS participants using a paperbased questionnaire and those using an online questionnaire when taking age, sex, and education into account (Supplementary Table S1). At the time of data collection, 83% of the participants had a sexual partner and 60% of those thought that this sexual relationship will also continue for at least three months. Thirty participants (2.5%; 20 men, 10 women; median age 18, IQR 17-26) reported that they have never had a sexual contact in their lifetime.
Men were older at their sexual debut than women in HaBIDS (median ages 18 years vs 17 years; Table 2, Fig. 1) as well as in the other studies. Participants with a university quali cation or university degree were generally older at sexual debut when compared to those without a university quali cation (Fig. 1).
Women reported more often a sexual relationship with older opposite sex partners while men reported frequently to be engaged in sexual relationships with younger opposite sex partners (Fig. 2). Lifetime opposite-sex partners In HaBIDS, men and women had a median of 4 (IQR 2-8, and 2-7, respectively) lifetime opposite-sex sexual partners. In the group of 15 to 19 years old participants, 17 participants (46%) reported to have never had an opposite-sex sexual partner; this proportion decreased strongly for older age groups. Fortythree percent of women and 13% of men reported to have had just a single partner in their entire life. Men had lower numbers of lifetime opposite-sex sexual partners in the younger age groups, and higher numbers in older age groups compared to women (Fig. 3a). Apart from the younger age groups (below 25 years for women and 30 years for men), the proportions of both men and women with a given number of lifetime partners were stable across age.
The number of reported lifetime partners was highest in Natsal followed by NSFG and both German studies (Fig. 4a, left panel, Table 3). The difference in lifetime partners can be linked to differences in the number of partners in younger age groups (14-24 years), while in the older age groups the age-related increase in the number of partners was similar. Consequently, the acquisition of new partners (partners in the last 12 months) differed more in younger age groups across countries (Fig. 4a, right panel). These patterns were generally similar in men and women, although absolute numbers of partners were consistently higher in men. The higher acquisition rate of new partners was not evenly distributed across all participants but caused by a higher proportion of persons in the groups with the highest partner changes (10-19, 20-29, 30 or more) in Natsal and, to a lesser extent, in NSFG (Fig. 4c). Those who reported more lifetime sexual partners had also more sexual partners in the last 12 months, supporting individual stability of behaviour as used in the concept of stable sexual activity groups (Fig. 4b). Note: Opposite/same/both-sex partners in the last 12 months are based on participants who had at least one lifetime opposite/same/both-sex sexual partner. New opposite/same/both-sex sexual partners in the last 12 months are based on participants who had at least one opposite/same/both-sex sexual partner in the last 12 months. CI, con dence interval; IRR, incidence rate ratio; OR, odds ratio Opposite-sex partners in the last 12 months Among HaBIDS participants who had at least one lifetime opposite-sex sexual partner, 14% reported to have had no opposite-sex sexual partner in the last 12 months, while 79% reported one and 7% reported two or more opposite-sex sexual partners in the last 12 months. In the 15 to 19 years old HaBIDS participants, among those with at least one lifetime opposite-sex partner, all women had at least one opposite-sex partner, whereas 25% of men had no opposite-sex sexual partner in the last 12 months (Fig. 3b)

New opposite-sex partners in the last 12 months
Of those participants in HaBIDS with at least one opposite-sex sexual partner in the last 12 months, only 19% reported one or more new opposite-sex sexual partners in the last 12 months. The highest proportion of participants who reported a new opposite-sex sexual partner occurred in the age group 15 to 19 years (88%) (Fig. 3c). Natsal reported a slightly higher number of new opposite-sex sexual partners (IRR 1.36, 95%CI 1.14-1.62) among those who had at least one lifetime partner compared to HaBIDS (Table 3,  Supplementary Table S4, Figure S4b).
Fifty percent of HaBIDS participants with at least one lifetime same-sex sexual partner, reported one or more same-sex sexual partner in the last 12 months (Supplementary Table S6). NSFG participants reported lower numbers of same-sex sexual partners in the last 12 months compared to HaBIDS (IRR 0.61, 95%CI 0.43-0.87); this was not the case for Natsal (IRR 0.90, 95%CI 0.63-1.29). The number of new same-sex partners among those participants with a same-sex sexual partner in the past 12 months did not differ between Natsal and HaBIDS (IRR 1.38, 95%CI 0.73-2.54; Table 3; Supplementary Table S7).
Participants with both-sex sexual experience during lifetime and in the last 12 months

Discussion
In this manuscript we provide for the rst time systematically collected data on sexual contact patterns in the German population from two independent studies; these data can be used for the parametrization of mathematical models in the eld of sexually transmissible infections. Our analysis found that sexual contact patterns in Germany were similar to those reported from the UK and US. However, there were important differences with respect to the proportion of participants with a high number of partners in younger age groups or the proportion of individuals with both-sex sexual experience. The results of both German studies agreed in the majority of analyses.
Median age of participants' rst sexual contact in HaBIDS was 17 years in women and 18 years in men which is consistent with the median age of sexual debut in Natsal and NSFG 7,8 . Across all age groups, over three-fourths of participants in HaBIDS and Natsal maintained a stable opposite-sex sexual relationship in the last 12 months. The only exceptions were those below 25 years where almost half of the study population in both studies reported one or more new opposite-sex sexual partners. This is not unexpected as research has previously demonstrated that adolescents are more likely than adults to report multiple sex partners in the recent past 18,19 . German studies showed a lower number of lifetime sexual partners than Natsal and NSFG. This pattern was driven by a higher proportion of individuals with a high number of partners in the younger age groups in the UK and US. The observed differences have potentially huge implications for the way infectious diseases spread in the respective societies, since high activity groups are at the core of transmission.
Less than 10% of all participants reported to have ever had a same-sex sexual relationship in all studies across all age groups; among those with experience with sexual partners from both sexes the number of lifetime sexual partners was considerably lower in HaBIDS than in the other studies. This is especially important because this subpopulation serves as a bridging population in diseases where transmission dynamics are different among opposite-sex and same-sex partnerships.

Strengths and limitations
Even though the four surveys were conducted independently with different setups, they sought the same information. Nevertheless, questions used in the four surveys differed slightly from each other opening the chance for misclassi cation.
Some of the differences observed between studies might have occurred because of differences in sampling techniques, survey aims, or the longevity of projects associated with surveys. More sexual partners were reported in the UK which could be attributed to Natsal's long-term research on sexual contacts, so that study participants were aware and familiar with Natsal and less likely to report lower numbers of partner because of societal norms. On the contrary, HaBIDS was not purely designed to focus on sexual behaviour but evaluated in more general terms knowledge, attitudes, and practice related to infectious diseases so that participants might have felt less comfortable and might have underreported the true number of sexual partners.
The data were cross-sectional, so that comparisons across age are also comparisons of different cohorts. Nevertheless, the observed patterns across age appeared consistent with stable partner change behaviour in the last few decades.
The long-term panel nature of HaBIDS resulted in a much lower response proportion (9%) compared with SBG (52%), Natsal (58%) and NSFG (77%); while willingness to participate in studies generally decreases, the lower response proportion could be a potential source of bias e.g. because individuals with university degrees were more likely to participate resulting in an overestimation in HaBIDS, and tend to differ in their sexual contact patterns from individuals without university degrees. In all the four studies, participants lled in questionnaires by themselves without interviewers. While this provided more privacy, it is still possible that respondents deliberately chose not to divulge correct information. Moreover, respondents might not have recalled information correctly, particularly if the numbers were high. In case of di culties to recall, partnerships in the past 12 months may be the most reliable measure 20 . Given that these points affected all studies equally, it is unlikely that they were responsible for observed differences between studies.

Conclusion
In both German studies, there was a lower proportion of participants with a high number of partners in younger age groups when compared to Natsal and NSFG. This resulted in a lower number of lifetime partners also at older ages, even if the partner acquisition above 24 years was similar in all studied populations. Dependent on the characteristics of the studied sexually transmitted infections, these patterns can lead to substantially different prevalences and incidences of sexually transmitted infections across age, and have strong impact on modelling studies of potential interventions (e.g. vaccination or screening effects).     Comparison between studies on (a) the mean number of reported opposite-sex sexual partners by age in lifetime and in the last 12 months: (b) mean number of opposite-sex sexual partners in the last 12 months by age and lifetime partners; and (c) proportion of lifetime partners across age groups