We conducted a systematic review of quantitative studies that examined the associations between exposure to pornography and sexual behaviors in adolescents aged between 10 and 19 years. We followed the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines (Figure 1 and Multimedia Appendices 1 and 2) [27,28]. This systematic review was registered with the International Prospective Register of Systematic Reviews (PROSPERO registration number: CRD42021227390).

We searched 6 databases: CINAHL (via EBSCOhost), Cochrane Reviews (via Ovid, see Multimedia Appendix 3 for specific Cochrane databases searched), Embase (via Ovid), MEDLINE (via Ovid), PsycINFO (via Ovid), and Web of Science (all databases: Web of Science, Current Contents Connect, BIOSIS Previews, CAB Abstracts, and MEDLINE). The search strategy, developed in MEDLINE and later adapted to other databases, used controlled vocabulary and free-text terms that related to the 2 key concepts of adolescents and pornography (Multimedia Appendix 3) as appropriate for each database. The search strategy, including adaptations, was developed by an experienced specialist research librarian. By design, search terms did not include “preteen*” and “prepubert*” because these terms were considered less likely to yield studies about young users of pornography than about child pornography (see Inclusion and Exclusion Criteria). Similarly, the names of apps were not included as search terms, because the apps are used not only for pornography by our definition (in subsequent section) but also for interactive sexual experiences. The searches were limited to peer-reviewed papers published from January 1, 2010, to November 30, 2022, to update a previous review article [19]. Other studies that were found to be independent of our search strategy were identified through citation searches.

To be considered eligible, studies must have reported quantitative empirical findings on the association between exposure to pornography and sexual behavior outcomes (Table 1). Qualitative studies and systematic reviews were excluded from analysis. For this review, exposure to pornography was defined as a participant’s report of any contact with sexually explicit or erotic media (pornography), with the intent of sexual arousal [29]. We included studies that did not define the exposure but used synonymous terms such as “sexually explicit material,” “adult or X-rated material,” or “pornography” without defining it in the methods. We excluded studies on child pornography, that is, sexually explicit or erotic material portraying individuals aged <18 years.

Sexual behavioral outcomes included those with established associations with adverse health and social outcomes (Table 1). Studies describing sexual aggression were limited to those that measured forced or coercive penetrative sex, excluding those that reported only on forced or coercive kissing and touching or used only the term “sexual acts” without further specification. Composite indices of sexual behaviors were excluded, as the association could not be clearly mapped to a specific behavior. The included studies were limited to those with adolescent participants (aged 10-19 years, inclusive), in which the age range was not provided, and the reported mean or median age was within this range. Included studies were those from a general population sample, and studies with clinical or specific population groups (eg, juvenile detention) were excluded. There was no restriction on the ethnicity of the participants or the study country. Multiple studies that analyzed the same cohort were considered together if they reported different associations.

The first author (PP) conducted the search and removed duplicates using EndNote X9 [30]. PP reviewed the titles and abstracts (N=5852) and reviewed the full text (n=337) for eligibility (Figure 1). In cases of doubt (n=7), another author (SRS) was consulted. A total of 19 studies were eligible for inclusion. Information about the study location, design, sample size calculation, participants, setting, exposures, outcomes, control variables, methods of statistical analysis, effect sizes, and their variations were extracted by PP and MR. Where possible, effect sizes were calculated from the data if not reported through the study. Included studies either reported on sex or gender. For this systematic review, we used male/female/nonbinary gender or male/female sex to describe the gender or sex of the cohorts as reported in each study.

Study quality was assessed independently by the authors PP and MR using the Newcastle-Ottawa Scale [31]. For the Newcastle-Ottawa scale used to assess longitudinal studies, we replaced the item “was follow-up long enough for outcome to occur” with “whether the outcome was assessed to be at low levels at baseline.” Cross-sectional studies were assessed based on the representativeness of the sample, sample size, addressing the influence of nonrespondents on the reported effect estimates, exposure, and outcome, reporting of statistical analyses, and comparison of outcome and control groups (Multimedia Appendix 4 [32-50]). Longitudinal studies were assessed based on the representativeness of the sample, selection of a nonexposed cohort, assessment of exposure, outcome (at baseline and subsequent waves), follow-up rates, and comparability of outcome and control groups (Multimedia Appendix 4). In cases of conflicts, 2 other authors (ML and JM) were consulted. Low risk of bias was classified as a score of ≥7, moderate risk of bias as a score of 4 to 6, and high risk of bias as a score of 0 to 3.

Initial searches identified 5852 abstracts, and 19 studies met the inclusion criteria (Figure 1). This included 11 cross-sectional studies [32-42], 2 case-control studies [43,44], and 6 prospective cohort studies [45-50]. Furthermore, 8 studies addressed early age (<16 years) of first sex [32,34,35,40,41,43,48,50]; 5 studies addressed sexual behaviors including condomless sex [37,42,46,47,49]; 2 studies examined multiple sexual partners during lifetime; a single study examined group sex [35]; 3 studies examined coercive or forced sex [33,39,41]; and 5 studies examined other behaviors including paid sex [38,41,45], teenage pregnancy [44], and history of sexually transmitted infections [36].

The included studies were conducted across 13 countries (Multimedia Appendix 4), and 2 studies were conducted internationally across various countries in Europe [33,40] (Multimedia Appendix 4). Most studies were conducted in European countries (n=9) [33,35,40,41,45,46,48-50], followed by countries in Asia (n=6) [32,36,38,43,44,47], the United States (n=3) [37,39,42], and Eastern Uganda (n=1) [34]. Ten studies reported on the frequency of pornography exposure [33,35,37,38,41,45,46,48-50], and of these studies, one [33] reported “regular” exposure without further definition. Four studies reported exposure to different pornography contents [35,37,39,42], including violent pornography (aggressive sexual acts) and pornography with condomless sex (Multimedia Appendix 4). Exposure to pornography was not uniformly measured; 7 studies did not define pornography in their methods [32,33,35,41,43-45] and a further 6 studies defined it as “sexually explicit content,” “adult content,” “banned media,” or “R-rated media” [34,36-38,40,47] (Multimedia Appendix 4). Similarly, the definitions of outcomes varied across studies. Age at first sex was measured prospectively in one study by asking at each follow-up whether an adolescent had ever had sexual intercourse [50] and retrospectively in 2 studies by asking the age at first sexual intercourse [35,41]. Another longitudinal study assessed the outcome of the first sexual experience in a sexually naive population at baseline, with a mean age at baseline of <16 years [48]. Four studies had a mean sample age of <16 years, so any participant in these samples who had sexual intercourse was considered to have had a young age of first sex [32,34,40,43]. “Multiple sexual partners” included an adolescent self-reporting sex with more than one partner over their lifetime (n=2) [46,47] and having ever participated in group sex (n=1) [35]. Studies of forced or coercive sex included participants as perpetrators (n=3) [33,39,41] and survivors (n=2) [33,39]. Studies of paid sex included participants as buyers (n=3) [38,41,45] or sellers (n=2) [41,45]. Because of the heterogeneity in how exposure and outcomes were measured across studies, a meta-analysis was not possible.

A total of 8 studies (2 prospective cohorts, 1 case-control, and 5 cross-sectional) assessed the relationship between pornography exposure in adolescence and earlier age of first sex (defined as below the age of 16 years; Table 2 [32,34,35,40,41,43,48,50]). Of the 6 cross-sectional studies that examined this outcome, 2 studies [35,41] assessed only male adolescents, whereas the other 4 studies [32,34,40,43] studied cohorts of both males and females and the analysis was not stratified by sex. Two prospective cohort studies reported conflicting findings regarding this association. The first study, of 639 male adolescents, conducted in Belgium [50] found a positive longitudinal relationship between exposure to pornography and an earlier age of first sex. The second study [48] reported no association between exposure to pornography and an earlier age of first sex in the overall sample. However, adolescents with a later age of first exposure to pornography had a significantly lower probability of reporting first sexual experience in subsequent waves than their counterparts with earlier exposure to porn. Subsequent sensitivity analyses to account for missing values showed no major association between the timing of porn exposure and first sex in males but showed an association between regular exposure to porn versus no exposure and first sex in females [48]. Both studies were assessed to have a low risk of bias. Four cross-sectional studies of low to moderate risk of bias conducted in Sweden [35], Thailand [32], Eastern Uganda [34], and 23 EU countries [40] supported this association. Two studies did not identify any substantial associations. In 2013, Sahay et al [43] presented a P value close to the threshold and conjectured that the rarity of the outcome in their sample limited the power to detect a true association.

A total of 5 studies (3 prospective cohorts, 1 case-control, and 1 cross-sectional) assessed the relationship between pornography exposure in adolescence and participation in condomless sex (Table 3 [37,42,46,47,49]). Two studies [37,42] were conducted in the United States, and one each in Croatia [46], Taiwan [47] and the Netherlands [49]. The largest study assessing this outcome was a prospective cohort study of 2054 participants conducted in Taiwan [47], and the smallest was a cross-sectional study of 206 participants conducted in the United States [37]. Of the 2 cross-sectional studies that examined this outcome, one studied a cohort of male adolescents [37], whereas the other examined a cohort of males and females, where analysis was not stratified by gender [42]. The 3 longitudinal studies reported cohorts of male and female adolescents, where the analysis was not stratified by sex. However, one study [46] included gender as an interaction term to analyze the effects of gender on the association, while another [47] adjusted for gender in the regression models.

The results of these studies were mixed. Among 3 longitudinal studies [46,47,49], one conducted in Taiwan [47] reported a positive association and was assessed to have a moderate risk of bias, while the studies from Croatia and the Netherlands [46,49] reported no major associations and were assessed to have a low risk of bias. Both cross-sectional studies supported a positive association between exposure to pornography and condomless sex in the samples [37,42]. Nelson et al (2019) examined pornography patterns in non-heterosexual male adolescents [40], a study that was assessed to have a low risk of bias. Specifically, this study found that those who viewed more condomless sex (>50% of total pornography viewed) were more likely to have reported participating in condomless sex [37]. Another cross-sectional study that found an association, by Wright et al [42], was assessed to have a high risk of bias owing to poor adjustment of confounders and ascertainment of exposures and outcomes.

Two prospective studies examined the association between pornography exposure and multiple lifetime sexual partners [46,47]. These studies were conducted in Croatia [46] and Taiwan [47]. Both studies examined male and female cohorts. Koletić et al [46] included gender as an interaction term to analyze the effects of gender on the association, whereas Lin et al [47] adjusted for gender in the regression models. The study by Lin et al, 2020 [47] found an association between pornography exposure and multiple sexual partners and was assessed to have a moderate risk of bias owing to not measuring the outcome at baseline, having a low proportion of the sample reporting the outcome, and limited assessment of follow-up and retention rates (Table 3). This study reported exposure to pornography across different media (such as comic books and web sites) in a single exposure variable. Koletić et al [46] found a major association but only in a nonadjusted model. A single cross-sectional study conducted in Sweden on a cohort of male adolescents reported a nonsignificant association between exposure to pornography and participation in group sex [35], that is, sex with multiple partners simultaneously (Tables 1 and 3).

A total of three cross-sectional studies [33,39,41] assessed the relationship between pornography exposure in adolescence and experiences of forced or coercive penetrative sex (Table 4). These studies were conducted in the United States [39], Sweden [41] and 5 European countries [33]. The largest study was by Barter et al [33] in 2021 of 3277 adolescents, and the smallest was conducted by Rostad et al [39] in 2019 with 1766 adolescents. Furthermore, 2 out of 3 studies reported a major positive association in reported experiences of forced or coercive sex in males only [33,39]. Two of the 3 studies [33,39] stratified their analyses as male or female, and one of the studies examined only male adolescents [33].

The relationship between pornography exposure and “paying for sex” was inconsistent (Table 4). Studies have been conducted in Singapore [38], Sweden [41], and Switzerland [45]. A prospective cohort study [45] reporting this exposure and outcome found a positive association between exposure to pornography and sex. This was supported by one [38] of the 2 cross-sectional studies [38,41] that examined this association. There are an insufficient number of studies examining both male and female adolescents to conclude whether this association was gender dependent, as both Ng et al [38] and Svedin et al [41] reported on male adolescents yet found opposing substantial associations. Regarding the outcome of selling sex, a prospective cohort study found no association with pornography viewing [45]. However, they reported an interaction effect with gender, whereby females with high pornography consumption were more likely to sell sex than those with low consumption. One cross-sectional study [41] that sampled only male adolescents found a positive association between selling sex and viewing pornography. There are an insufficient number of studies examining this association to draw conclusions.

One study reported a major association between exposure to pornography and teenage pregnancy [44], whereas another found an association with a history of sexually transmitted diseases [36] (Table 4). The study of teenage pregnancy was assessed to have a moderate risk of bias, largely due to poor assessment of nonrespondents and limited justification of sample size. Both studies were limited by a lack of representative samples.