The physiology of adolescent sexual behaviour: A systematic review

Abstract Objectives: To examine physiological influences of adolescent sexual behaviour, including associated psychosocial factors. Methods: Systematic review. Results: Thirteen studies met the inclusion criteria relating to adolescents, physiology and sexual behaviour. We excluded studies relating to abnormal development. Findings highlighted hormonal and gender differences. Females appear to be more influenced by psychosocial aspects, including the effects of peers, than males. Males may be more inclined to engage in unprotected sex with a greater number of partners. Early maturing adolescents are more likely to be sexually active at an early age. Conclusions: Hormonal, psychosocial context, and sexual preference need to be acknowledged in intervention development. Stage of readiness to receive information may differ according to gender and physiological maturity.

ABOUT THE AUTHORS Jan Pringle has worked as a research fellow and systematic reviewer for the Scottish Collaboration for Public Health Research and Policy (SCPHRP) for over two years. SCPHRP's vision is to develop Scotland as a leader in public health research through catalysing strong researcher/ research-user collaborations. The adolescent sexual behaviour paper published here is part of a larger project to examine all aspects of adolescent development, with a view to informing interventions and policy. The project as a whole was developed in partnership with key personnel at NHS Health Scotland, and collaborators from the Scottish Government.
Other aspects of physiological influence on adolescent health behaviours (e.g. sleep, substance use, eating behaviour, physical activity) are discussed in a wider report, as detailed in the paper. Our intention is to widen this focus, and examine psychological, sociological, environmental, and cultural influences in future work.

PUBLIC INTEREST STATEMENT
Sexual activity is a normal part of growing up. However, it may also cause harm, if sex takes place when children are too young, or do not know the risks. Risks may include sexually transmitted infections or unintended pregnancy.
Our work aimed to gather information about factors within the bodies of adolescents that affect their sexual behaviour. This information came from a large number of sources. We found that sexual behaviour is affected by many things, such as hormones and chemical changes. Family, friends, and social groups were also important in changing the way adolescents behaved. We found that young women were more influenced by their friends than young men. Those who matured early were more likely to have sex at a younger age. This could increase the risks for these adolescents. Understanding such things can help us to provide better support for adolescents and those who care for them.

Introduction
The sexual behaviour of adolescents is of importance due to the increasing number of sexually active adolescents globally (World Health Organisation [WHO], 2012). While initiation of sexual activity is a part of normal behaviour and development, it may also be associated with negative outcomes, if sexual behaviour involves engagement in sexual activity at too early an age, or without due attention to the risks involved (Maswikwa, Richter, Kaufman, & Nandi, 2015). Teenagers and young adults may face many sexual and reproductive health risks stemming from early, unprotected, or unwanted sexual activity (WHO, 2012). For example, early initiation of sexual activity increases the period of time adolescents are exposed to the risk of sexually transmitted infections or unintended pregnancy (WHO, 2011). Although there is no universal definition of early initiation of sexual activity, it is often classified as sexual intercourse during initial high school years (Johnson & Tyler, 2007) or sexual intercourse before the age of legal consent (Girma & Paton, 2015).
For many adolescents, sexual activity may start earlier than permitted by law (Klettke & Mellor, 2012;Yarrow et al., 2014); in the USA, for example, 62% of students were reported to have engaged in sexual activity before leaving school (Martinez, Copen, & Abma, 2011), and in many instances, young people may initiate sexual relations before the age of 14 years (WHO, 2011). There may be associated risks of early pregnancy, for example, if adolescents do not have a good understanding of contraception (WHO, 2012).
While rates of teenage pregnancy may be falling in some middle to high income countries (Girma & Paton, 2015), early sexual activity and pregnancy are still prevalent in other areas and cultures, and may be associated with increased morbidity and mortality (Maswikwa et al., 2015;WHO, 2012). Encouraging appropriate attitudes to sexual behaviour and activity during adolescence can help to ensure that contraception is understood, pregnancy is intended (Sher, 2016), and the cycle of deprivation associated with pregnancy in young people under the age of 18 is reduced (Scottish Government, 2016). Promoting healthy sexual behaviour can help reduce these risks, and further negative outcomes, such as sexually transmitted infections.
Sexual behaviour may be influenced by many physiological factors, in addition to the cultural and social pressures which can change rapidly from one generation to the next. Understanding the physiological influences that drive adolescent sexual activity, such as hormonal, chemical and neurological reactions and changes, can help inform interventions to support adolescents in making appropriate choices regarding their sexual behaviour. Exploring the nature of these physiological processes, and associated consequences, is essential for designing effective responses to meet the sexual and reproductive health needs of adolescents (WHO, 2011), and may help to protect their rights in relation to their choices. Improved understanding of factors that may predispose adolescents to certain behaviours can not only help to inform adolescents, but also those involved in their care.
This review aimed to examine the physiological factors that influence adolescent sexual behaviour. The review forms the first phase of a comprehensive project to examine all aspects of adolescent development and potential influences on health behaviour. This paper reports on the findings specifically relating to sexual behaviour. Other aspects of adolescent physiological development and health behaviour (e.g. substance use, sexual behaviour, physical activity) are discussed elsewhere, including our main report (McAteer et al., 2017).

Methods
A systematic review was conducted from June 2015-January 2016, using a protocol aligned to PRISMA-R guidelines, with pre-specified inclusion/exclusion criteria. The review aimed to capture an extensive range of studies with different methodological approaches, and both published and unpublished literature, in order to produce an extensive understanding and avoid publication bias (de Souza, da Silva, & de Carvalho, 2010). Details of the overall review process are noted in the PRISMA flow chart (Figure 1).

Search strategy
A detailed review protocol and search strategy was developed and agreed between the project team and advisory group members (Pringle et al., 2016). In brief, we searched MEDLINE, Pubmed, PsycInfo, Embase, ERIC, ASSIA, Discovery, Cinahl, and Cochrane databases, using key words relating to adolescence and young adults, physiological development, and (sexual) health behaviours. We also searched webpages, scanned reference lists, and included papers identified by experts in the field.

Study selection and inclusion criteria
We set no search limits on study design, and scoped a wider age range (9-24 years) than the "second decade" definition (WHO, 2015) in order to include the initial pubescent years and subsequent early adulthood period.
Inclusion dates ranged from 1980-2016. Our focus was on studies relating the impact of physiological systems on health behaviour rather than examining the impact of health behaviour on physiology, since we were not seeking to conduct a review of intervention or behavioural effect. Our focus was on either physiological impact on health behaviour, or on research that indicated a bidirectional influence. We aimed to acknowledge associated psychosocial influences as they related to behavioural outcome findings. In order to include a broad range of potential outcomes, we did not put any predetermined definitions in place regarding sexual activity or behaviour; we were therefore able to include any such outcomes that had been examined in relation to physiological processes.  Two reviewers (JP and KM) screened papers independently using the inclusion and exclusion criteria (summarised in Table 1). A third reviewer from the project team examined papers where there was lack of agreement.

Data extraction, assessment of study validity, and analysis
Data relating to study aim, country, date, method, population, outcome measures, findings and conclusions were extracted and tabulated. We critically examined method, population and outcome measures to assess validity and limitations. Due to the heterogeneity of the study populations, methods, and outcome measures, meta-analysis and comparable quality appraisal were not feasible. A thematic narrative summation was therefore undertaken, to allow for a deductive approach to the development of key categories. Within identified key categories, we sought to extrapolate data relating to physiological influences (e.g. hormonal changes) and potential impact on sexual health behaviour. The resulting analysis provides a summary of three main topics of focus: hormonal influences, brain physiology, and associated psychosocial factors.

Results
The PRISMA flow chart ( Figure 1) summarises the review screening process, and results.
Studies relating to sexual behaviour formed one of six subdivisions within the larger review process. As mentioned above, other areas of interest (sleep, eating behaviour, genetic influences, physical activity, and substance use) are reported elsewhere.
The review identified 13 individual studies that focussed on adolescent sexual behaviour. Details of these studies are given in Table 2.
Of the 13 studies, all but two were conducted in the USA; one study involved Zimbabwean boys (Campbell, Prossinger, & Mbzivo, 2005) and the other was conducted by a Dutch research team (Baams, Dubas, Overbeek, & van Aken, 2015). There was therefore an overall lack of diversity in the populations studied. Dates of publication were broad, ranging from 1981-2015. There were three review papers; five of the remaining ten papers involved cross-sectional study designs, which limited any causality inferences. Follow-up periods for the 5 longitudinal studies ranged from 2-10 years. None of the studies involved data from participants under 12 years of age, and the upmost age of participants was 26 years of age. All papers examined factors affecting the onset of sexual activity, rather than other aspects of sexual behaviour. Generally onset of sexual activity was related to first coitus. Subset analysis divided the topics of focus into three main, albeit overlapping, areas: hormonal influences, brain physiology, and associated psychosocial factors.

Inclusion criteria Exclusion criteria
Studies/reports involving: Studies/reports involving: • Adolescents (nominal age range 9-24 years*) and • Physiological development, including developmental influencers (e.g. biochemical influences, hormonal activity, enzyme activity, neurotransmitters etc.), with related outcomes and • Sexual health behaviour, including lifestyle or behavioural factors/outcomes, and decision-making • Children (e.g. under 9 years*) as sole focus and/or • Adults (e.g. >24 years of age*) as sole focus and/or • Atypical or pathological physical /psychological/ physiological developmental focus (e.g. eating disorders, addictions, gambling, or major mental health disorders), and their treatment interventions • Adolescent pregnancy (as sole focus) • Incidence, prevalence or trend papers (as sole focus) • Protocols for studies yet to be undertaken Abbreviations: SR = self report; fMRI = functional Magnetic Resonance Imaging; BOLD = blood oxygen level dependent (contrast imaging).

Hormonal influences
With regard to the onset of sexual activity, several papers examined the hormonal influence of testosterone. Although a longitudinal study by Halpern, Udry, Campbell, and Suchindran (1993) did not provide support for the idea of a direct causal link between testosterone levels and change in sexual motivation or behaviour, two later longitudinal studies, with more frequent data collection, were able to demonstrate such links (Halpern, Udry, & Suchindran, 1997, 1998. The first, reported that higher levels of salivary testosterone were associated with more frequent sexual activity in adolescent males, with the second reporting similar associations in adolescent females, and the timing of their first coitus. For adolescent females, frequency of attendance at religious services was also examined as a variable (Halpern et al., 1997), and found to moderate the effects of testosterone on sexual transition. Psychosocial influences are discussed further below.
In contrast, Udry, Billy, Morris, Groff, and Raj (1985) examined hormonal and social effects on adolescent male sexual behaviour, and concluded that testosterone appeared to affect sexual motivation directly in boys, and did not operate through social influences, such as peer pressure, in the same way that it may do for girls. In addition to the effects of testosterone, Campbell et al. (2005) reported that boys' first spontaneous nocturnal emission was a stronger predictor of initiation of sexual activity than secondary sexual characteristics, such as the development of genital hair. Both of these studies were cross-sectional in design, so no causal links could be established. However, taken in conjunction with the previous longitudinal studies, they add weight to evidence of the influence of hormonal and developmental physiological processes on sexual activity (coitus) initiation.
Alongside testosterone, other hormonal changes (not withstanding peer influences) may also account for gender variations. For example, Smith, Udry, and Morris (1985) found that the sexual behaviour of female adolescents was positively affected by increasing levels of the sex hormones, testosterone and oestrogen. In contrast, for males, greater pubertal/physiological development in terms of Tanner staging (Marshall & Tanner, 1970), regardless of age, was associated with greater sexual involvement, which was unrelated to friends' sexual behaviour. Although the study by Smith et al. (1985) was cross-sectional in method, such gender differences were also found in a review of studies by Graber and Sontag (2006) who concluded that, under the influence of gonadal hormones, sexuality begins to develop more fully during puberty, develops extensively over adolescence, and is interconnected with changes in self and social context for girls during this period. Longitudinal research by Moore, Harden, and Mendle (2014) also found that girls' perceived pubertal timing, in addition to actual age at menarche, predicted age of first sexual intercourse. This is concurrent with the large review by Baams et al. (2015), who were in agreement that early physiological development (i.e. early onset of puberty) was associated with earlier sexual activity, especially in girls.

Brain physiology
Although there was limited data relating to brain physiology and sexual behaviour in the review, in a sample of high-risk adolescents, one study correlated how an individual's brain activity and associated inhibiting behaviour related to his/her frequency of unprotected sexual intercourse and substance use (Feldstein, Houck, & Bryan, 2015). These high-risk adolescents were recruited from a court ordered diversion program in the USA. The study examined the relationship between brain responses during a response inhibition task and past month health-risking behaviour. They found a negative correlation between substance use and brain responses (specifically in the left inferior frontal gyrus and right insula), and a positive correlation between risky sex and brain responses in the right inferior frontal gyrus and left middle occipital gyrus during response inhibition. The authors suggest that these findings indicate that engaging in risky sexual behaviour or substance use reflects more than the individual's cognitive control capacities, but also their ability to assess the relevant socioemotional factors of that decision. This fits with the frameworks considering how adolescents incorporate social factors when assessing the value of engaging in a health risk (Blakemore & Mills, 2014).

Associated psychosocial factors
While the main focus of the review was on physiological aspects of adolescence, there were other factors that were important to take into account from the findings. For example, in addition to physiological maturation, a review identified psychosocial factors that preceded the onset of adolescent sexual intercourse, which included alcohol use, delinquency, and school problems (Zimmer-Gembeck & Helfand, 2008). Zimmer-Gembeck and Helfand (2008) also found that the onset of adolescent sexual intercourse was correlated with depressive symptoms (especially for girls engaging in early sexual activity). With regard to other consequences, apart from early physiological development being associated with earlier and more (risky) sexual behaviour, especially in girls, early sexual activity may also result in increased incidence of infection or unplanned pregnancy (Baams et al., 2015). Zimmer-Gembeck and Collins (2008) found that adolescents accumulated a higher number of sexual partners by age 16 years if they were more physiologically advanced (e.g. looked older), drank alcohol more frequently, and were more involved with dating in early to mid-adolescence (13-16 years of age). Male gender (and thereby male physiological hormonal profile), in particular, was associated with accumulation of sexual partners more rapidly between ages 16 and 26 years. There was little indication that the accumulation of different sexual partners had begun to slow by age 26 in this study. These findings indicate that interventions targeted at teens to reduce sexual risk behaviour may be just as necessary in the later teen years/emerging adulthood, and may need to be gender sensitive in terms of timing and content.
As discussed above, frequency of attendance at religious services was found to temper the effects of testosterone on sexual transition (Halpern et al., 1997), which is consistent with a biosocial model suggesting that physiological influences can be moderated by relevant social variables.

Discussion
This review reports on adolescent physiological development and its impact on sexual behaviour. The papers relating to adolescent sexual behaviour highlight the effects of differing hormonal profiles, and associated gender differences. In addition to the effects of hormonal factors, females appear to be more influenced than males by psychosocial factors, including the effect of peers. An associated paper (Skinner et al., 2015), which did not focus on physiological factors, reported that behavioural issues (e.g. anger, antisocial behaviour) from as early as 5 years of age in boys and 10 years of age in girls were a significant risk factor for earlier age at first sexual intercourse. This is something that adolescent sexual health programmes may therefore want to consider when developing interventions.
In further relation to the exact specifics of sexual activity, most of the studies defined this as engagement in vaginal coitus, either implicitly or explicitly. However, an early study by Dornbusch et al. (1981) reported on data limited to whether participants had ever been on a date. Conclusions regarding the timing of dating behaviour drawn from this study may therefore not be comparable with the timing of more specific sexual activity (e.g. coitus) data from other studies. This perhaps reflects the less open attitudes that prevailed during earlier eras, and possibly the acceptability of researching such personal topics. It does, however, provide a historical view of past research approaches and findings, to help situate current research findings.
Within the sexual behaviours that were examined in the included studies, and the method(s) of assessing pubertal status, the studies varied in terms of outcome measures, which made direct comparisons difficult. For example, pubertal status measures ranged from Tanner staging assessed by a professional, to parental or self-assessment. Although these latter measures may be considered reliable, there may also be elements of self-assessed psychological and behavioural maturity incorporated in such considerations (Dorn & Biro, 2011), which may not be acknowledged as confounding factors, thereby limiting the findings.
The review papers highlighted links between early sexual maturity and early sexual activity. With regard to the measurement of developmental changes, Halpern, Kaestle, and Hallfors (2007) used "perceived physical maturity" (i.e. the individual's perception of their maturity) as a proxy for pubertal status, and concluded that, although advanced physical maturity was associated with increased risk (in terms of alcohol, substance use, sexual risk-taking etc.), it was also associated with greater likelihood of having a romantic partner, which further increased risk, especially for girls with an older (≥2 years) partner. However, as Halpern et al. (2007) noted themselves, although self-perceived measures are meaningful and may be valid, they are not the same as objective measures. It is possible that adolescents with older partners may see themselves as more mature by association. Past sexual experience may also impact on an individual's maturity self-perception. In addition, girls on average enter puberty two years in advance of boys (Kail & Cavanuagh, 2015), so an age gap of this range would indicate a physical maturity match, and potentially a match with regard to readiness to engage in sexual activity; however, as the review has highlighted, this latter consideration may be mediated by culture, society or religion. Whilst these factors may be acknowledged as influencing issues, the impact on individual behaviours is difficult to ascertain, unless these are specific elements under study.
With further regard to sexual behaviours, these were often not specified in sufficient detail for comparison between studies; for example, intercourse may or may not be explicitly specified as vaginal, and there is often little acknowledgement that sexual activity may involve actions other this. Anal intercourse, for example, may be increasingly likely regardless of sexual orientation, possibly due to changing social norms (Marston & Lewis, 2014). Similarly, homosexual sexual activity, either as a behaviour or in relation to greater risk, is not acknowledged. Baams et al. (2015), for example, noted only one study (out of 50 studies in their review) as specifically relating to homosexual activity, but this was not discussed in their analysis. Similarly, Halpern et al. (1997), although noting such behaviour as increasing risk, did not discuss data from within this category.
Likewise, adolescent males may be generally less positive about condom use (Rich, Mullan, Sainsbury, & Kuczmierczyk, 2014), with subsequent increased infection risk, particularly for homosexual men. Sexual minority females (e.g. bisexual, lesbian) may be more likely to experience a teen pregnancy and less likely to receive preventative advice (Charlton, Missmer, DiVasta, Rosario, & Austin, 2016). In the past, this lack of emphasis on anything other than vaginal intercourse may have been due to a primary focus on pregnancy prevention (Albert, Brown, & Flanigan, 2003;Teenage Pregnancy Independent Advisory Group, 2010). It would therefore seem important that current research and interventions recognise a broad range of potential sexual behaviour, and acknowledge the physiological influences that impact on adolescent sexual activity during this time.
A recently published review of romance and sex during adolescence by Suleimann, Galvan, Harden, and Dahl (2016) highlights the fact that support "scaffolding" can help adolescents to negotiate the relationship terrain during their physiological and maturational development. Support services are not only welcome by adolescents, but can help to reduce health disparities (Mason-Jones et al., 2012). We hope that our findings will help to inform the knowledge base for such interventions.

Conclusions
This review has highlighted the role that hormones play in influencing sexual behaviour during adolescence. It has also discussed the influence of psychosocial factors, including peers, and how these may differ between males and females.
With regard to maturity outcome measures (e.g. Tanner stage), the influence of maturity perceptions needs to be taken into account. For example, perceived maturity may be a confounding factor, irrespective of objective maturity, in terms of engagement in sexual activity. Social context and cultural or religious persuasion may also be mediating factors in the decisions adolescents make about sexual activity, in addition to physiological drivers. To more fully establish these influences, social context, cultural and religious affiliation need to be made more explicit for the populations under study.
Furthermore, sexual orientation needs to be considered in terms of sexual behaviour to avoid potential discriminatory reporting, or the under reporting of potentially risky sexual behaviours.
Further research relating to the impact of physiology on adolescent sexual behaviours would be beneficial, especially in a wider variety of populations. In particular, longitudinal research is needed to more fully clarify the causal links between hormones and their influence on sexual behaviour. In addition, there is a need for greater consistency and clarity regarding specific sexual behaviour outcomes to allow comparison between study findings. For example, coitus needs to be defined in terms of vaginal intercourse, or otherwise, for results (and risks) to be more fully evaluated.

Intervention development
The review has highlighted specific physiological gender differences that need to be taken into account when designing interventions to promote sexual health. Since many physiological factors, such as hormonal levels and brain development are non-modifiable, understanding how these interact with more modifiable factors, such as peer and social influences, can contribute to more nuanced intervention development (Wight, Wimbush, Jepson, & Doi, 2016).
Stage of readiness to receive advice and information relating to sexual behaviour may differ according to physiological maturity. However, early behavioural problems may indicate the need for earlier intervention, especially if linked to early physiological maturity. Interventions for those at risk may need to be adapted to be appropriate for both the age and maturity of the intended audience.
Intervention development needs to take account of the influence of social context, culture and religious affiliation. Some of these factors may offer protection against risk, by opposing physiological drivers. However, further research is needed to more fully establish how these protective factors may be harnessed in intervention development, or the identification of populations at greater risk.