Plain English summary

Reproductive and mental health are key domains of adolescent wellbeing but possible interrelationships are poorly understood. This study evaluated the association between psychopathology and reproductive health risk in a large multinational sample of European adolescents. A structured self-report questionnaire was delivered to 12,395 pupils of 179 randomly selected schools in 11 European countries within the EU funded “Saving and Empowering Young Lives in Europe” (SEYLE) project. Among 11,406 respondents, a wide spectrum of psychopathologic manifestations was positively associated with sexual initiation, independently of age and sex. The association appeared to be stronger for more overt manifestations, such as self-injurious behaviour and suicide attempts. However, having depression or anxiety was also associated with sexual initiation, consistently with a dimensional nature of adolescent psychopathology. The association between overt psychopathology and sexual initiation was stronger among pupils with low non-sexual risk behaviours compared to those with high non-sexual risk behaviours. Additionally, there was a stronger association between depression, serious suicidal ideation, self-injurious behaviour, or suicide attempts, and sexual initiation among pupils ≤15 years old compared to their older counterparts. Similarly, the association between anxiety or self-injurious behaviour and sexual initiation was stronger among girls compared to boys. Finally, pupils with psychopathology manifestations were more likely to have an increased reproductive health risk.

These findings suggest an alignment between mental and reproductive health risk and support the value of cross-domain collaborative efforts to prevent disease and improve the health of young people.

Background

Adolescence is a unique period of human life characterized by dramatic development, which includes reproduction-related changes and, often, sexual initiation. Developing sexuality is physiologic during adolescence, but early debut implies a risk for immediate or future adverse reproductive health outcomes, such as unwanted teenage pregnancies and their consequences [1, 2]. Unsafe sexual practices expose to the risk of sexually transmitted infections, which can also have repercussions on future health and fertility [3,4,5,6]. Women who had sexual debut before the age of 15, more often report gaps in contraceptive use, sex-partner concurrency, and serial monogamy later in life [4, 7, 8]. Apart from reproductive health risk outcomes, early sexual initiation is associated with non-sexual risk behaviours (e.g. smoking or substance abuse), context vulnerabilities (e.g. bullying, truancy and low parental involvement) and psychological difficulties [9], which are not uncommon among girls and boys [10].

While reproductive and mental health are acknowledged key aspects of European adolescents’ current and future wellbeing [11], both appear to be neglected in terms of systematic data collection and research in Europe. Many studies focus on health outcomes belonging to one specific domain (e.g. reproductive or mental), but likely interrelationships between different aspects of adolescent health are still poorly studied hence overlooked by ad-hoc policies [12]. However, the potential outcomes of adolescent sexual behaviours are obviously not only physical but also mental and social [13]. An association between early sexual initiation and psychopathology manifestations such as depression has been reported by studies that mostly proceed from the USA [5, 14,15,16,17], whereas fewer similar findings are available from European countries [18, 19]. Although there is evidence that early sexual debut may lead to adverse mental health outcomes [17], depression has also been observed to precede and predict sexual initiation, multiple partners and inconsistent condom use among adolescents [5, 16, 18]. Other longitudinal studies have found that mental health problems predict poor compliance to prescribed contraception [20] and depressive symptoms are associated with an increased risk of unintended pregnancies [21].

The complex and likely bidirectional relationship between mental health and sexual behaviours is consistent with the conceptual view that factors influencing sexual activity also determine the way adolescents perceive and evaluate their own behaviour [13]. While sexual behaviours may directly lead to physical health outcomes, it is their perception that arguably mediates outcomes in the mental and social health domains [13]. In view of this complexity, a broader perspective on adolescents’ sexual experiences taking into account other facets of their wellbeing, such as mental health, would be desirable in Europe. It would also be important to study whether other psychopathology manifestations beyond depression are associated with early sexual initiation and risk-taking, although that would clearly require more substantial research efforts. This study took advantage of comprehensive information about the health and wellbeing of more than 12,000 European adolescents, which was gathered by a recent EU-funded project called SEYLE (Saving and Empowering Young Lives in Europe) [22, 23]. We aimed at evaluating whether pupils with psychopathology manifestations, such as depression, anxiety, self-injurious behaviour and suicidality, are more likely to have a higher reproductive health risk, defined by earlier sexual initiation and associated risk factors. A secondary objective was to verify whether the strength of the association between psychopathology and sexual initiation is modified by age, sex or the co-occurrence of non-sexual risk behaviours.

Methods

Study population

Between 2009 and 2011, 12,395 adolescents (median age 15; interquartile range [IQR] 14–15; mean age 14.91 ± 0.90) were enrolled into the EU-funded SEYLE project [22]. This population consisted of young females (55.2%) and males (44.8%) attending 179 schools in 11 countries: Austria, Estonia, France, Germany, Hungary, Ireland, Israel, Italy, Romania, Slovenia and Spain. Karolinska Institutet, Sweden, acted as scientific coordinating centre, and the Child Psychiatric Epidemiology Group at Columbia University and New York State Psychiatric Institute participated as methodology experts. The two core objectives of SEYLE were to gather epidemiological data about European adolescent health and wellbeing, and to actively test suicide-preventive interventions in a randomized controlled trial [24]. The included schools were randomly selected out of a list of eligible schools located within the study sites of each country, according to the selection criteria described by Wasserman et al. [22]. A 67.8% response rate was obtained from the selected schools. Elsewhere published analyses show that SEYLE’s study sites were representative of each respective national population, based on none or small significant differences with the respective country in several key parameters, including age, sex distribution, proportion of 15-year old males and females, net income, immigrants and unemployment rates [23]. At recruitment, pupils were asked to complete a structured, self-report questionnaire addressing socio-demographics, risk factors, lifestyle, and mental health. This paper-based survey was administered in the official national languages during a single classroom session. Research staff was available in each location to supervise and assist the pupils with the process. For the current study, we selected all SEYLE cases with complete baseline data about age, sex, and reproductive health risk (N 11,406; 92.5%).

The consent of pupils and their caregivers was obtained before inclusion. Further information about the sample and SEYLE’s core methodology is elsewhere published [22, 23].

Outcome variable

The main outcome in the study was the prevalence of self-reported sexual initiation, which was evaluated by one closed-ended question (“have you ever had sexual intercourse?”). SEYLE’s questionnaire investigated reproductive risk factors (RRF), that were included in analysis as dichotomous variables: number of sexual partners (dichotomized as 1 versus ≥2); use of condom (dichotomized as “rarely or never” versus “always or almost all the time”); pregnancy involvement (yes or no). For each study subject, an additional outcome variable called “reproductive health risk” was computed by subcategorizing the outcome sexual initiation depending on associated risk factors (no initiation; initiation without RRF; initiation and 1 RRF; initiation and 2 or 3 RRF).

Psychopathology measurements

Psychopathology was evaluated with known instruments such as the Zung Self-Rating Anxiety Scale (Z-SAS) [25], the modified 20-item Beck Depression Inventory-II (BDI-II) [26,27,28], a modified six-item version of the Deliberate Self-Harm Inventory (DSHI) [29, 30], and the Paykel Suicide Scale (PSS) [31]. When official validated versions were unavailable in the required language, the surveys were translated, back-translated, and linguistically adapted. Internal reliability of psychometric scales was confirmed by a consistently high Cronbach’s alpha [23]. Dichotomous psychopathology variables were computed for each study subject. Anxiety and depression were respectively defined by a Z-SAS score ≥ 45 and a modified BDI-II score ≥ 17. Self-injurious behaviour was defined by a sum of ≥3 obtained adding all points of the modified DSHI [10]. Severe suicide ideation was defined as having seriously considered taking one’s own life, or having made plans about that, at least sometimes during the previous two weeks (4th item of the PSS) [24]. Self-reported suicide attempts were investigated by means of a multiple choice closed-ended question (“Have you ever tried to take your own life?”; possible answers: yes, during the past 2 weeks; yes, during the past 6 months or longer; no, never) and recoded as a dichotomous variable (yes or no).

Potential confounders and effect modifiers

Self-reported age and sex of each participant were obtained through SEYLE’s questionnaire. Age was dichotomized with a 15-year-old cut-off (≤15 versus > 15), based on evidence that sexual debut in Europe more often occurs after 15 years old [32]. Besides, other large surveys collect data on sexual initiation by the age of 15 and European teenagers believe that one may be “too young to have sexual intercourse” at an average age of 15.5 [33].

SEYLE’s questionnaire also provided data regarding non-sexual risk behaviours such as substance abuse (smoking, alcohol and illegal drugs), truancy, sedentariness, poor sleep, and high media use. Previous latent class analysis research identified different patterns of risk clustering in the SEYLE cohort [10]. Building up on those findings, participants in the present study were sub-grouped according to their non-sexual risk behaviour profile into a low risk profile group, including pupils with low scores on all examined risk behaviours; and a high risk profile group, including pupils clustering on multiple overt (i.e. smoking, alcohol and illegal drugs use, truancy) and/or “invisible” (i.e. poor sleep, sedentariness and high media use) risk behaviours [10].

Statistical analyses

Descriptive statistics were used to describe the study population, with a focus on the prevalence of reported sexual initiation, reproductive risk factors, and psychopathology. Chi-square test was used to compare the prevalence of reported sexual initiation depending on age, sex, non-sexual risk behaviour profile, and psychopathological manifestations, as previously defined. The association between each psychopathology variable and sexual initiation was tested with multivariable logistic regression analyses. These analyses were repeated after stratifying the sample for age and sex, as well as for the two subgroups of opposite non-sexual risk behaviour profile. Additional analyses including interaction terms were performed when the stratified findings suggested differences among the strata, in order to test interactions between each psychopathology manifestation and the variables age, sex or non-sexual risk behaviour profile. Ordinal and multinomial logistic regression analyses were used to measure the strength of association between psychopathology variables and the outcome reproductive health risk. The logistic regression analyses were by default adjusted for the age and sex variables because of the expected association with the study’s main outcome measure, with the exception of the stratified analyses where adjustment was not applicable. Statistical significance of differences was set at a two-tailed p value of less than 0.05. Odds ratios (OR) were calculated together with 95% confidence intervals (CI). The software IBM® SPSS® Statistics ver. 23 for macOS was used for statistical analyses.

Ethical approval

SEYLE was approved by the European Commission, as a precondition for funding, as well as by the local research ethics committees of each national recruiting centre [22]. An external advisor from the University of Basel, Switzerland, provided independent ethical assessment and supervision of the project [23].

Clinical trial registration

SEYLE was registered in the German Clinical Trials Register (DRKS00000214).

Results

Complete data from SEYLE’s reproductive health items were available from 11,406 pupils, accounting for the 92.5% of the whole SEYLE population (median age 15; IQR 14–15; mean age 14.88 ± 0.88). Forty-three percent (N 4889) of the respondents were males, while 57% (N 6517) were females (Table 1). The majority of respondents were 15 years old or younger (N 8844; 77.2%).

Table 1 Age, sex and psychopathology in 11,406 adolescents with or without sexual initiation
Full size table

Sexual initiation

Sexual initiation was reported by 18.8% of the respondents (N 2143; mean age 15.39 ± 0.87). It was significantly more common among boys (21.3% versus 16.9% in girls; p < 0.001) as well as among those older than 15 years old (38.0% versus 13.2% in pupils ≤15 years old; p < 0.001). The rate of sexual initiation was also significantly higher in the high non-sexual risk behaviours profile group (20.7% versus 18.2% in the low profile group; p = 0.009).

More than 97% of participants responded to each one of the considered mental health items (Table 1). The prevalence of reported sexual initiation was significantly higher among pupils with anxiety, depression, self-injurious behaviour, severe suicidal ideation and suicide attempts, as defined in the previous section (Table 1). All psychopathology manifestations were significantly associated with reported sexual initiation even when adjusting for age and sex (Table 2; Fig. 1). The results of the logistic regression analyses remained unaltered even when adjusting for pupils’ country of origin. Stratified analyses suggested a stronger association among pupils with a low non-sexual risk behaviour profile (Table 2). Further analyses with interaction terms confirmed a significantly stronger association between self-injurious behaviour and sexual initiation among pupils with low non-sexual risk behaviours compared to those with high non-sexual risk behaviours (p = 0.024). Similarly, the association between suicide attempts and sexual initiation in the low non-sexual risk behaviours group was significantly stronger than in the high non-sexual risk behaviours group (p = 0.038).

Table 2 Prevalence and odds of sexual initiation among European adolescents with and without psychopathology
Full size table

The age-sex stratified analysis showed particularly strong associations between psychopathology and sexual initiation among younger (versus older) and female (versus male) pupils (Table 3). Therefore, we performed separate analyses including interaction terms which confirmed a significantly stronger (p < 0.05) association between depression, serious suicidal ideation, self-injurious behaviour, or suicide attempts, and sexual initiation among pupils ≤15 years old compared to their older counterparts. Similarly, the association between anxiety or self-injurious behaviour and sexual initiation was significantly stronger (p < 0.05) among girls compared to boys.

Table 3 Association between psychopathology and sexual initiation, stratified for age and sex
Full size table

Reproductive health risk

Sexual experience was associated with at least one of the considered reproductive risk factors in 60% of the cases: 52.4% reported more than one sexual partner (40.1% of females and 65.4% of males); 14.7% reported rare/no use condoms (18.3% of females and 10.8% of males); and 3% reported pregnancy involvement (2.5% of females and 3.6% of males) (Table 4). Half of the adolescents reported one reproductive risk factor, while 9% reported two, and 0.6% reported all the three considered risk factors (Table 4).

Table 4 Characteristics of 2143 adolescents reporting sexual initiation and associated risk factors
Full size table

Having each psychopathology manifestation significantly increased the odds of being in higher levels of the outcome reproductive health risk (Table 5). This was particularly evident when the outcome was sexual initiation with multiple RRF (versus no sexual initiation) in the multinomial analyses (Fig. 1).

Table 5 Association between psychopathology and reproductive health risk
Full size table
Fig. 1
figure 1

Psychopathology and sexual initiation among European adolescents. Logistic regression derived odds ratios (aOR) and 95% confidence of interval (95% CI) of reported sexual initiation, with or without associated reproductive risk factors (RRF) in European adolescents with different manifestations of psychopathology. The reference category of the outcome is “no sexual initiation”. The odds ratios are adjusted for age and sex

Full size image

Discussion

This study evaluated the association between psychopathology and reproductive health risk in a large, multinational and representative sample of more than 11,000 European adolescents. A wide spectrum of psychopathologic manifestations was positively associated with sexual initiation, independently of age and sex. The association appeared to be stronger for more overt manifestations, such as self-injurious behaviour and suicide attempts. However, having depression or anxiety was also associated with sexual initiation, consistently with a dimensional nature of adolescent psychopathology [28]. The association between overt psychopathology and sexual initiation was stronger among pupils with low non-sexual risk behaviours compared to those with high non-sexual risk behaviours. Additionally, there was a stronger association between depression, serious suicidal ideation, self-injurious behaviour, or suicide attempts, and sexual initiation among pupils ≤15 years old compared to their older counterparts. Similarly, the association between anxiety or self-injurious behaviour and sexual initiation was stronger among girls compared to boys. Finally, pupils with psychopathology manifestations were more likely to have an increased reproductive health risk.

Physiological reproductive and sexual development characterizes the adolescent transition from childhood to adulthood. Females and males become fertile during this period when many have also their sexual initiation. In European countries, sexual debut often occurs between 16 and 18 years of age, while it is less common before the age of 15 [32]. This was confirmed by our findings since only 13% of adolescents ≤15 years old reported sexual debut compared to 38% of those over 15 years old of age. The evident link between sexual behaviours and the risk of adverse reproductive health outcomes is often targeted by research and policy efforts in the field of adolescent health. The idea of early sexual debut as a gateway to sexual and reproductive risk has commonly informed responses directed to avoiding or postponing sexual activity, or to minimizing risks through specific education and care. However, the timing of sexual initiation alone may not be considered as an independent marker of risk, as risky sexual activity is often associated with non-sexual risk behaviours (e.g. substance abuse) [3, 9, 34, 35], exposure to childhood trauma [9, 34, 36] or problematic social interactions (e.g. with peers, school or family) [9, 34]. Therefore, a reductionist perspective of reproductive health as an isolated domain may only offer a limited understanding of adolescent risk behaviour and is hardly functional to the development of much needed and advocated intersectoral and multicomponent actions for adolescent wellbeing [37].

Aspects of the relationship between sexual and non-sexual risk behaviours may be interpreted through a gateway effect (e.g. substance abuse or victimization facilitating exposure to unsafe sex) [12]. However, such interpretation overlooks the fact that adolescents often cluster on multiple risk behaviours, which leads to ineffective surveillance and responses [12]. An alternative or complementary interpretation may be informed by a behavioural risk syndrome model, which frames adolescent risk-taking in the context of broader correlates, such as individual and social vulnerabilities [12, 38,39,40]. Therefore, our initial hypothesis that adolescent psychopathology could be associated with early/risky sexual activity appeared to be well grounded. Nevertheless, most related observations proceed from the USA and focus on depression [5, 15,16,17, 41, 42], whereas less studies have considered young Europeans [18, 19] or other psychopathology manifestations such as suicidality [14, 19].

The present study adds insights to existing literature: first of all, the association between psychopathology and sexual initiation exists also among European adolescents, and involves a wider range of manifestations beyond depression; secondly, our findings suggest an alignment between the two domains of mental and reproductive health, as the strength of association increases with more overt psychopathologic manifestations or higher reproductive health risk. Another peculiarity of this study is that it considered factors that, by altering the strength of the association between psychopathology and sexual initiation, may inform our interpretation. The stronger association among those who report sexual debut by the age of 15 is arguably explained by the fact that sexual activity is less normative at younger age. It therefore occurs more often among vulnerable adolescents, who also engage in other risk behaviours and are exposed to worse health outcomes during adolescence and adult life. Similarly, the strong association in case of overt psychopathology observed among young females, who generally experience a later sexual debut [32, 43], suggests that norm-breaking behaviours may relate to dysfunctional situations where risk-taking and poorer mental health coexist.

Finally, the stronger association between overt psychopathology and sexual initiation among pupils with low risk behaviour profile is an interesting and partially unexpected finding. In the absence of psychopathology, the higher prevalence of sexual initiation among pupils engaging in multiple non-sexual risk behaviours may relate to gateway mechanisms (e.g. non-sexual behaviours as facilitators to risky sex) and is consistent with the known clustering of risk behaviours among undercontrolling adolescents with conduct and hyperactivity (i.e. externalizing) problems [10, 44]. On the contrary, sexual initiation was most commonly reported by pupils with psychopathology, independently of their non-sexual risk behaviour profile. Therefore, internalizing symptoms, such as the ones captured by this study, may play a specific role when it comes to sexual behaviours, and the possibility of a bidirectional relationship should be acknowledged. Altogether, these findings explain why targeting individual risk behaviours without addressing underlying vulnerabilities, such as poor mental health, may be an ineffective way to promote adolescent well-being.

The present study has some limitations which should be taken into account. Not all SEYLE participants could be included because of missing data regarding the reproductive health outcome measures. The amount of excluded participants was however very low, as 92% were eventually included. Additionally, the response rates to the psychopathology measurements were consistently very high (97–99%). In terms of generalizability, the samples recruited at the eleven study sites were population-based and fairly representative at national level [23], though it should be acknowledged that Europe is a wide and heterogeneous geopolitical area. It is however reassuring that the association between depression and sexual initiation has also been observed in countries not participating to SEYLE [18], and that the clustering and psychosocial correlates of adolescent risk behaviours appear to be consistent across European countries [40]. Regarding the study methodology, a cross-sectional design does not allow for inferences about directionality or causality. A further consideration needs to be made regarding the study variables, which were obtained through self-report. Diagnostic interviews would allow for a more correct classification of psychiatric disorders, but they may be unfeasible in the context of a large study as well as they may underestimate symptoms. Self-report of risk or sexual behaviours may be biased by social pressure, although SEYLE participants were made aware about the strictly confidential procedures and anonymous data management and analysis. Respondents may have variably interpreted what “sexual intercourse” is, as no distinction between different sexual activities was made in the survey. Besides, our choice of reproductive health risk variables was restricted to those available in SEYLE’s questionnaire, while it would have been interesting to collect additional data, such as about contraceptive use, sexual orientation, gender identity, or exposure to sex-related trauma. Finally, SEYLE did not collect individual data about the age of sexual debut nor about pubertal timing, which may be a correlate of adolescent sexual experiences.

Despite those considerations, a number of implications for further research and practice may be suggested by this large and multinational study. It would be interesting to test our findings in longitudinal studies, as insights about the directionality of the relationship between psychopathology and reproductive health risk are needed to verify how much policy responses may rely on gateway models [12]. However, establishing directionality does not prove causality nor it ensures that interventions on the presumed gateway would have downstream effects on the outcome. Besides, the possibility of reverse or bidirectional relationships should be considered.

As a matter of fact, psychological vulnerabilities are possibly one of the numerous determinants of adolescents’ sexual behaviours and their self-evaluation, which is then mediating mental and social health outcomes [13]. A comprehensive view on those factors would help us understanding why physiologic events such as sexual debut and developing sexuality sometimes deviate from being positive experiences, as they instead could and should be [13]. Instead of addressing adolescent sexual behaviours only in view of their potential negative outcomes on physical health, it should therefore be acknowledged that coexisting problems in different domains of adolescent wellbeing may constitute different manifestations of a more complex behavioural risk syndrome [12], which entails also social and contextual factors. For instance, family affluence appears to be related to adolescent sexual behaviours, although the extent and direction of that association is unclear [43]. On a related note, a very recent study shows that depression generally increases the risk of unintended pregnancy among young American women although this does not apply to those who are white or have a higher socioeconomic status [21]. From a behavioural risk syndrome perspective, it is reductive to abstract each relationship from the whole picture whereas it would be interesting to look into broader correlates of adolescent health and risk behaviours [12]. The disparities between adolescent males and females, and their cross-national variation, would be another interesting topic for further research. In fact, while boys are in general more likely than girls to report having had intercourse by the age of 15, that disparity is not observable in all countries or regions [43]. Besides, it would be interesting to study the relationship between psychopathology and sexual behaviours separately for girls and boys, as they cluster on different risk taking profiles [10].

Adolescent health practitioners need to be aware of the possibility that reproductive health risk is overrepresented among adolescents with psychopathology. This knowledge may be a useful tool for the early identification of young people who are exposed to specific risk or in need of tailored interventions [39]. Broad surveillance and comprehensive policies should also take into account the limitations of gateway effect interpretations and the diversity between boys and girls, who are particularly vulnerable because of the obvious peculiarities of female reproductive health outcomes. Possible interactions between key domains of European adolescents’ wellbeing have often been neglected [11, 12, 45]. However, clustering and correlates of their risk behaviours are cross-nationally consistent [40] and it is recognized that effective adolescent health strategies require intersectoral and multicomponent actions [37]. In this context, our findings highlight the need and potential for broad collaborative efforts to prevent disease and improve the current and future health of young people.

Conclusions

Among European adolescent girls and boys, psychopathology is associated with early sexual initiation and reproductive health risk. This is particularly evident in case of more overt manifestations, such as self-injury and suicidality. Besides, the association between psychopathology and sexual initiation is particularly strong among pupils ≤15 years old and, for overt psychopathology, among pupils with low non-sexual risk behaviour profile or females. These findings should be taken into account when tailoring health-promoting or disease-preventing interventions for adolescents.