Introductions

Recent years have documented a shift in the application of sexual orientation labels among youth, including an increase in reported changes in sexual orientation (Katz-Wise, 2015; Russell et al., 2009). These changes in sexual orientation are more prominent in adolescence and young adulthood, which are considered important periods for sexual orientation development (Perrin, 2002; Savin-Williams & Cohen, 2004). However, experiences and processes of sexual orientation development may differ; some sexual minority youth may identify with a fixed sexual orientation (consistent over time), whereas others may present instability in sexual identification (change over time; Diamond et al., 2017; Katz-Wise & Hyde, 2015; Ott et al., 2011). Additionally, youth may experience a change in one or more aspects of their sexual orientation over time (e.g., change in sexual attraction, behavior, identities or labels; Ott et al., 2011). These changes in sexual orientation also have been referred to as fluidity or mobility (Diamond, 2016; Everett et al., 2016; Ott et al., 2011; Scheitle & Wolf, 2018).

Changes in sexual orientation may also have implications for behavioral health outcomes (Everett, 2015; Fish & Pasley, 2015; Needham, 2012; Ott et al., 2013). For example, sexual orientation change may be stressful for some individuals, wherein a shift in their orientation may accompany a loss of support from previously held orientation groups or stress from navigating the social networks of a new sexual orientation group (Katz-Wise et al., 2019; Ott et al., 2011; Rosario et al., 2011). Some support for this exists; Everett (2015) found that reported changes in self-reported sexual orientation were associated with increased depressive symptoms (this association was stronger among those who identified as exclusively heterosexual at baseline as opposed to those moving from one sexual minority orientation toward another; e.g., bisexual to gay or lesbian).

That said, not all changes in sexual orientation have negative implications for health. For example, a shift toward an orientation that is more authentic and congruent may act as a protective factor and mitigate health risks. For example, Rosario and colleagues (2011), in their study with 156 sexual minority adolescents, reported that youth who presented greater sexual identity integration reported lesser depressive and anxious symptoms and higher self-esteem, both cross-sectionally and longitudinally. Yet the largely adult-focused literature has reported on fluidity in identification among cisgender male and female participants and inconsistencies among sexual attraction, behavior, and identity (Apostolou, 2018; Gruia et al., 2022; Hunt et al., 2018). Hunt et al. (2018), in their integrative review of sexual orientation change, reported on sexual fluidity among cisgender female adult participants and binary sexual orientation experiences (hetero- and homosexuality) among cisgender male participants. Similarly, Diamond (2016), in a review of evidence of sexual orientation change, reported longitudinal changes in sexual attraction and discordance among sexual attraction, behavior, and identity among adults. Although these reviews have provided summative evidence of sexual identity change in adults, how this phenomenon manifests in adolescence and young adulthood and the relationship between orientation change and health outcomes remain largely unexplored.

In the current systematic review, we conducted a synthesis of empirical research from the last two decades (2000–2020) among adolescent and young adult populations. We aimed to answer the following questions: (1) How do studies measure sexual orientation change?; (2) How common is sexual orientation change among sample participants?; (3) Are there patterns in the direction of change reported?; and (4) What is the relationship of self-reported sexual orientation change to health outcomes? Our review summarizes the approaches to measuring sexual orientation change; the prevalence, patterns, and directionality of changes in sexual orientation; and how changes in sexual orientation relate to health outcomes among adolescents and youth adults. The goal of this review is to describe, summarize, and evaluate literature in this area and in doing so, develop recommendations for future research involving sexual minority adolescent and young adult populations.

Method

Operationalization

In this review, we use sexual orientation as an umbrella term to encompass various aspects of sexual identification, including orientation, attraction, and labels. Similarly, a varied range of vocabulary has been used to refer to a change in sexual orientation or identity, including fluidity, mobility, instability, and transition. For this review, we use the term sexual orientation change, encompassing the reported changes in any self-reported sexual orientation dimension among youth.

Search Protocol

To identify studies for inclusion in this review, a search was undertaken in accordance with Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines (Moher et al., 2009). Title and abstract reviewing was applied to seven databases: PubMed, PsyArticles and PsycInfo, Cochrane, GenderWatch, EBSCO (LGBTQ + Source), ProQuest Central, and Google Scholar. Each database was searched by combining terms from two conceptual categories: sexual orientation change (sexual identity change, sexual identity mobility, sexual identity fluidity, sexual identity transition, sexual identity stability, sexual orientation identity change, sexual orientation trajectory, sexual orientation identity, mobility metrics, sexual orientation change, sexual orientation change longitudinal study), with age category (adolescents, youth, and young adult). Two researchers conducted the same search of electronic databases to ensure all studies were identified. Additionally, each time an article matched the inclusion criteria, the Google Scholar search tools “related articles” and “cited by” were used. All search results were managed in Zotero and then transferred to Covidence for screening. Citations in relevant studies were then reviewed to determine if any referenced articles fit the search parameters.

Inclusion and Exclusion Criteria

The authors reviewed the abstracts to determine which studies met the review’s inclusion criteria. To be included, studies must have (a) been published between January 2000 and September 2020; (b) been published in a peer-reviewed journal; (c) been empirically based; (d) reported original research findings; (e) been published in English; (f) included sexual minority or nonheterosexual identified adolescents or young adults aged 12 to 26 years at baseline; and (g) examined or reported changes in self-reported sexual identity or orientation.

Though we recognize that sexual orientation development and change may continue over one’s life course, the psychological processes associated with sexual orientation identity development, as well as practice and interventions, differ between adolescence and adulthood (Craig et al., 2013; Everett et al., 2016; Fadjukoff & Kroger, 2016; Hash & Rogers, 2013; Phillips et al., 2020; Shilo & Savaya, 2011). In addition, adolescence and youth are a critical time for behavioral health and for sexual identity development, and until now, nothing exists to guide youth researchers and practitioners on how to incorporate identity change into their research. Hence, the current review only contains articles that included sexual minority or nonheterosexual identified adolescents or young adults aged 12 to 26 years at baseline. Additionally, given our team’s language capabilities, only studies that were published in English, irrespective of geographical location, were included.

All studies were independently reviewed by two authors and abstracted using a standardized form. The first author reconciled the work of the two reviewers and organized the summaries into one database that allowed aspects of all included studies to be compared and summarized.

Results

Our search resulted in 2,747 records (including five records from portal suggestions) identified through database searching. Of these, 2080 were excluded for not meeting the primary inclusion criteria (i.e., did not examine or report changes in sexual identity or orientation). Of the remaining 667 records, 249 were excluded for duplicates. Of the remaining 418 non-duplicate records, 368 records were excluded for not meeting the requirements of empirical and peer-reviewed publications; these included books, book chapters and essays (n = 116), dissertations (n = 195), essays and gray literature (n = 41), and non-English language publications (n = 16). This left 50 articles for full-text review for eligibility, including quality assurance and data extraction. Twenty articles were excluded for not meeting the age (n = 18) and sexual orientation change (n = 2; reported gender identity change) requirements. Outcomes of applying the inclusion and exclusion criteria to the retrieved articles are presented in Fig. 1. This process left 30 studies for the review. The reviewed studies differed in their research foci and approaches, and key aspects of each study are summarized in Table 1. Most studies were conducted in the USA (n = 27), along with one each in Australia, Croatia, and New Zealand. As shown in Table 2, several articles were based on common data sources.

Fig. 1
figure 1

PRISMA flow diagram

Full size image
Table 1 Descriptive summary of 30 reviewed articles
Full size table
Table 2 Linked publications included in the review
Full size table

Study Designs and Participants

Most reviewed studies were quantitative (n = 25); two were qualitative; and three reported a mix of quantitative and qualitative results. With a few exceptions, most studies reported results from longitudinal investigations (n = 26). Multiple studies included in the review used same data source (i.e., National Longitudinal Study of Adolescent to Adult Health; Table 2). The studies using a cross-sectional approach had direct measures of sexual fluidity and change, attitudes, and prevalence, reported both retrospectively and prospectively (Katz-Wise, 2015; Katz-Wise & Hyde, 2015; Katz-Wise et al., 2019). Studies reporting on nonprobability samples recruited participants from lesbian, gay, bisexual, transgender, and queer venues; university campuses; and other locations. Probability samples included nationally representative school-based samples (e.g., Cranney, 2016; Everett, 2015; Fricke & Sironi, 2020; Savin-Williams et al., 2012; Silva, 2018) or followed youth based on birth cohorts (e.g., Campbell et al., 2021; Dickson et al., 2013).

Most studies included both cisgender male and cisgender female participants, except six studies reporting on only cisgender female samples and one with only cisgender male sample (e.g., Cohen et al., 2020; Cranney, 2016; Dickson et al., 2013; Everett, 2015; Liu et al., 2019). However, most studies were conducted with cisgender participants; only two studies reported data from a more inclusive sample of genders (including transgender and another genders; Katz-Wise & Hyde, 2017; Katz-Wise et al., 2019). The racial and ethnic composition of the sample was reported in all studies. Morgan and colleagues (2018) conducted a qualitative investigation with four time points (3 months apart) with only Black men. Others reported a more diverse composition; for example, Stewart et al. (2019) reported that the racial composition of their study sample as 48% White, 24% Hispanic or Latinx, and 21% Black or African American. Berona et al. (2018) found that racial and ethnic minority status was associated with a three-class solution of sexual orientation trajectories, with racial and ethnic minority status significantly associated with membership in the primarily other-sex attraction class and the primarily same-sex attraction class, but not with bisexual attraction class. Ages ranged from 12 to 26 years at the time of first enrollment or baseline. However, participants’ age spanned decades once recruited and followed in longitudinal studies. For example, a birth cohort study from New Zealand measured changes in sexual identity and orientation, with data collected at ages 21, 26, 32, and 38 years (Dickson et al., 2013). Similarly, Diamond (2000, 2008) followed a sample of sexual minority cisgender female participants aged 16 to 23 years at enrollment for more than 10 years.

Except for nine studies, all studies reported results from samples that included both heterosexual and nonheterosexual identified youth at baseline. The proportion of participants who identified with a nonheterosexual identity at baseline ranged from 6.4% to 86.9%. Additionally, sample sizes also varied depending on the methods and approach, with the smallest sample being 15 in a qualitative study (Morgan et al., 2018) and the largest being 15,678 from a national probability study (Needham, 2012).

Measuring Sexual Orientation Change

The studies in our review differed in terms of assessment of sexual identification (Table 3). Many studies collected data on more than one aspect of sexual orientation, including sexual identity or label or orientation, sexual or romantic attraction, and sexual behavior. Most studies used a Likert scale based on Kinsey’s sexual orientation spectrum; exclusively heterosexual (or 100% heterosexual) to exclusively homosexual (or 100% homosexual). However, in its last wave (2008–2009), the National Longitudinal Study of Adolescent to Adult Health added an additional option to the scale—not sexually attracted to either men or women—to capture no sexual attraction or asexuality (Cranney, 2016).

Table 3 Sexual orientation measurement and reported change in the review
Full size table

Sexual attraction was also measured using a similar scale as Kinsey’s. For example, Cohen et al. (2020) captured romantic or sexual attraction by asking respondents about their attraction toward exclusively the other sex, mostly the other sex, equally the other and same sex, mostly the same sex, and exclusively the same sex. Others captured sexual attraction in terms of fantasies and sexual thoughts by asking about “the extent to which their recent sexual attractions, thoughts, and fantasies were focused on the same or the other sex (a) when in the presence of other individuals, (b) while masturbating, dreaming, or daydreaming, and (c) when viewing erotic material in films, magazines, or books” (Rosario et al., 2006, p. 49).

Three studies measured responses using discrete values on sexual labels and identities (e.g., lesbian/gay, bisexual or straight; Dickson et al., 2013; Feinstein et al., 2019; Rosario et al., 2006), while all others used a Likert scale based on Kinsey’s sexual orientation spectrum; exclusively heterosexual (or 100% heterosexual) to exclusively homosexual (or 100% homosexual). At each wave of data collection, Feinstein et al. (2019) asked participants to report their sexual orientation, whereas Rosario et al. (2006) offered response items: “When you think about sex, do you think of yourself as lesbian/gay, bisexual or straight?” Dickson et al. (2013) provided response items of heterosexual, homosexual, lesbian or gay, bisexual, and other. A few studies also used “romantic partnering” and “sexual behavior” to capture sexual orientation; however, they were used in conjunction with other variables like sexual identity, attraction, or orientation. One study used “same-sex experience” to capture sexual identification. Oi and Wilkinson (2018) operationalized same-sex experiences as reporting a same-sex romantic attraction, listing a same-sex romantic relationship, or reporting one or more same-sex sexual encounters.

Qualitative assessment (five studies) differed from quantitative assessments of sexual attraction and used more innovative techniques to capture sexual attraction. For example, to assess same-sex attractions, Diamond (2000, 2003, 2008) used a blank pie chart divided into 16 equal regions totaling 100%. Participants were asked to fill the chart to represent the percentage of their current sexual attractions directed toward the same sex on a daily basis, and the item was reassessed at follow up interviews. Katz-Wise and Hyde (2017) did a narrative and descriptive assessment of sexual orientation development and fluidity (e.g., Have you ever experienced a change in attractions to others over time?).

Except cross-sectional studies, all studies used repeated measures of sexual orientation (including labels, identities, attraction, and behavior) to examine changes. This approach was used in varied forms by most studies and accounted for any change in sexual orientation and number of changes over time (Campbell et al., 2021; Cohen et al., 2020; Everett, 2015; Oi & Wilkinson, 2018). Most studies used repeated measures on single or multiple components of sexual orientation (identity, attraction, behavior) to measure change in follow-up waves. Additionally, because most studies used Kinsey-informed Likert scales, they measured changes in direction: reported orientation which remained consistent between waves, or those that transitioned toward or away from same-sex attraction (100% homosexual).

Unlike longitudinal studies, cross-sectional studies relied on retrospective reporting of identity change (Katz-Wise, 2015; Katz-Wise & Hyde, 2015; Katz-Wise et al., 2019). For example, Katz-Wise and Hyde (2015) asked participants if they had ever experienced a change in attractions to others over time and if they did, their age at the initial change and whether it resulted in a change in their sexual label or identity. They were also asked about the number of sexual identity changes and the likelihood of future changes. In addition to reported change in sexual attraction, participants were asked to report on the Attitudes Regarding Bisexual Scale and a 5-item Sexual Fluidity Belief Scale (developed by the authors). In another study, in addition to reporting on changes and attitudes, Katz-Wise et al. (2019) measured entity and incremental views of sexual orientation with two subscales based on sexual orientation and person (for example, a person’s sexual orientation is something very basic about them that they can’t change). All items were measured on a 7-point continuous scale (from strongly disagree to strongly agree).

Sexual Orientation Change Prevalence

The prevalence of self-reported sexual orientation change differed across the studies, based on sample composition, demographics, sampling techniques, and assessments. Prevalence of sexual orientation change ranged from 6 to 30% in probability samples of heterosexual and sexual minority youth and 28% to 67% in nonprobability samples of sexual minority youth (Cohen et al., 2020; Diamond, 2008; Morgan et al., 2018; Silva, 2018). Multiple studies examined changes in sexual orientation between two consecutive time points. For example, Campbell and colleagues (2021) reported incidence of change in sexual orientation label as 19.3% (wave to wave), whereas Cohen and colleagues (2020) found 7.0% to 9.2% of the sample reported a change between consecutive time points. Similarly, Everett (2015) reported 12% of participants changed their sexual orientation identity between two waves, of whom 70% changed their sexual orientation to a more same-sex-oriented identity (identity scale: exclusively heterosexual, mostly heterosexual, bisexual, mostly gay or lesbian, and exclusively gay or lesbian).

In addition to reporting any sexual orientation change, some studies reported the number of sexual orientation changes over time. For example, Feinstein et al. (2019) found that 34.4% of participants reported at least one change in sexual orientation during the study period, with 24.6% reporting one change and 9.8% reporting two or more changes. Similarly, Diamond (2008) found that in a sample of sexual minority women, one third changed their sexual orientation two or more times, whereas Katz-Wise and Hyde (2015) found multiple changes in attraction reported by 21% of cisgender female (M = 2.58 changes) and 19% of cisgender male (M = 2.75 changes) participants.

In addition, two studies reported on the qualitative aspects of sexual orientation development and change. Morgan et al. (2018) interviewed 15 Black bisexual-identifying men at four points with 3-month intervals. Of 15 bisexual-identifying participants at baseline, only seven consistently identified as bisexual; eight reported a change to gay or questioning. Those who consistently identified as bisexual described similar experiences related to same- and other-sex sexual and romantic attractions and more consistently described same- and other-sex sexual behaviors. Those who reported a change in sexual orientation also described changes in same- and other-sex sexual and romantic attractions over time. Katz-Wise and Hyde (2017) interviewed 18 participants, who were asked to reflect on how they explain the changes that have occurred in attraction or sexual orientation and what potential factors prompted the change. Participants discussed the role of facilitating environment (such as, exposure to labels, involvement with LGBTQ community, LGBTQ friends) and social norms (such as, gendered norms on appearance and heteronormativity) as contributing factors to their current perception of changes in sexual orientation. Participants also discussed future changes in sexual orientation with respect to their speculations about multiple factors, such having a partner of a particular gender, or living in a more or less restrictive social environment.

Change in Sexual Orientation by Sex

Reported changes in sexual orientation also differed by sex assigned at birth. Compared to cisgender male participants, cisgender female participants were significantly more likely to report changes in orientation between waves (Dickson et al., 2013; Everett, 2015; Savin-Williams et al., 2012). Oi and Wilkinson (2018) found 13.3% of male and 19.3% of cisgender female participants reported a transition between exclusively heterosexual and same-sex experiences. This reported change in sexual orientation between cisgender female and cisgender male participants was higher in other studies: 18.0% vs. 6.2% in Fricke and Sironi (2020), 17.8% vs. 6.2% in Savin-Williams et al. (2012), and 26% vs. 11% in Stewart et al. (2019), respectively. This difference was further elaborated in results from Katz-Wise and Hyde (2015), who reported that cisgender female participants were more likely than their male counterparts to endorse beliefs in sexual fluidity, that sexuality is changeable, and that sexuality is influenced by the environment. Diamond (2008) found that over a 10-year span, two thirds of cisgender female participants changed their orientation labels, of whom one third changed sexual orientation labels two or more times. Multiple studies reported a smaller proportion of cisgender female than male participants reported exclusive heterosexuality at baseline (Cohen et al., 2020). For example, Kaestle (2019) reported 74% of cisgender women compared to 87% of cisgender men fell into the heterosexual class. Similarly, Fish and Pasley (2015) reported that more cisgender women were likely to report heteroflexibility (6.45%) and later bisexual identification (3.32%). Additionally, more cisgender female participants reported a change in orientation away from heterosexuality (Campbell et al., 2021). For example, Needham (2012) found that those who reported a transition to lesbian, gay, or bisexual attraction were disproportionately female (75%), whereas those who reported a transition to heterosexual attraction were disproportionately male (59%). However, it is important to note that not all studies found significant differences in sexual orientation change by sex assigned at birth. For example, Katz-Wise (2015) in their study reported that 64% of cisgender females and 52% of cisgender males reported fluidity in sexual orientation, though chi-square analyses did not reveal any significance in the difference. While Ott and colleagues (2011) found limited support on differences in sexual orientation change by sex assigned at birth. In their total sample, there was a significant difference in mobility scores by sex assigned at birth, with cisgender females having a higher mobility than cisgender males in the younger group and older group. However, among the group with self-reported minority sexual orientation, there were no significant difference in mobility scores by sex assigned at birth.

Change in Sexual Orientation by Other Covariates

Only two studies examined changes in orientation with covariates other than gender. Silva (2018) found non-Latino mixed-race men, as compared to non-Latino White men, had higher odds of changing to a straight orientation (OR = 8.39, p < 0.05). Additionally, cisgender female participants for whom religion became more important, as compared to less important, had higher odds of changing to a straight orientation (OR = 1.97, p < 0.01). Cohen et al. (2020) found that religiosity was associated with initial sexual attraction (more religious individuals were more likely to report exclusively heterosexual attraction), but not with changes in romantic and sexual attraction over time.

Patterns in Sexual Orientation Change

Some studies, in addition to reporting on prevalence of change, provided information on the patterns and directionality of sexual orientation change. For example, Berona et al. (2018) reported that the proportion of individuals endorsing same-sex and bisexual orientations in each class increased over time. They reported that 63.2% participants reported at least one change in sexual orientation, whereby the primarily same-sex class reported more frequent sexual orientation changes than the bisexual class and primarily other-sex class.

Two studies used life stages to categorize orientation development over time (Hu et al., 2016; Liu et al., 2019). For example, Liu et al. (2019) reported four categories by life stage: (a) early sexual minority status continuing into adulthood (same-sex sexual orientation in both adolescence and adulthood; 7.4% of cisgender women and 4.5% of cisgender men); (b) late sexual minority status (same-sex orientation in adulthood only; 10.1% and 3.1%, respectively); (c) early sexual minority status that did not continue into adulthood (same-sex orientation in adolescence only; 7.8% and 10.2%, respectively); and (4) the reference category of heterosexual throughout adolescence and young adulthood (no same-sex attraction in adolescence or adulthood; 74.7% and 82.2%, respectively).

The changes in sexual orientation were also found to be associated with endorsement of sexual orientation labels and attraction. For example, most studies reported that exclusive heterosexual orientation labels remained stable over time (Campbell et al., 2021; Fish & Pasley, 2015); however, among those identifying as nonheterosexual, same-sex attraction was found to be more stable over time compared to bisexual attraction (Hu et al., 2016). For example, Cohen et al. (2020) found that stability of attraction was observed in 91.5% to 96.1% of exclusively heterosexually attracted adolescents, compared with only 28.5% to 33.7% of adolescents with other attractions.

Bisexual, mostly heterosexual, or mostly homosexual identities were found to have more reported changes in sexual orientation than their counterparts. For example, Savin-Williams et al. (2012) reported that bisexual-identifying cisgender male and female participants were more likely to report a high rate of temporal shifts compared to those with an exclusive same- or opposite-sex attraction; in addition, more bisexual and mostly heterosexual young adults of both sexes moved toward heterosexuality than toward homosexuality over time. Feinstein et al. (2019) reported similar findings: Cisgender female participants who initially identified as bisexual were more likely to report two or more changes than those who initially identified as lesbian (b = 0.61, p = 0.026).

Directionality of Change

A few studies reported on the directionality of sexual orientation change (Table 3). For example, Sabia (2015) found that among participants who reported a change, 11.1% moved toward same-sex and 4.9% away from same-sex attraction among female respondents, compared with 3.0% and 1.9%, respectively, among male respondents. Similarly, Silva (2018) found only 4.3% of the sample reported a change toward heterosexual orientation and 8.4% toward nonheterosexual orientation. Similarly, Everett (2015) reported that among those who reported a change in sexual orientation, 70% changed to a more same-sex orientation.

In another study, Ott et al. (2011) found differences in the directionality of change by both age and gender. Among younger male participants (12–15 years old), 2.3% reported a change toward completely heterosexual; 5.1% reporting moving toward gay or bisexual; and 3.5% reported a multidirectional change. Among older male participants (15 or older), 1.9% moved toward completely heterosexual; 5.2% moved toward gay or bisexual; and 3.5% reported a multidirectional change. Similarly, among younger cisgender female participants, these figures were 3.0%, 8.9%, and 6.7%, respectively; and among older cisgender female participants, they were 4.4%, 8.9%, and 6.5%, respectively. This directionality in change was also observed in sexual minority samples. Rosario et al. (2006) found that 18% of their sample transitioned toward a gay or lesbian identity, whereas 5% transitioned toward bisexual and 5% toward straight.

Relationship with Mental Health and Substance Use

Six of 30 studies measured a mental health or substance use outcome associated with change in sexual orientation. These studies used directionality of change (a shift in orientation toward or away from a same-sex orientation) to examine the association of the orientation change with behavioral health outcomes. For example, Everett (2015) reported that only changes in sexual orientation toward a same-sex orientation were associated with increases in depressive symptoms. This directionality of change was also reported by grouping orientation changes. For example, one study reported that sexual minority groups (heteroflexible, later bisexual, and lesbian, gay, or bisexual) experienced greater depressive symptomology and suicidality compared to groups characterized by opposite-sex attraction, behavior, and identities (heterosexual early daters and heterosexual late daters; Fish & Pasley, 2015).

A few studies reported the impact on health outcomes by gender. For example, Needham (2012) found cisgender female participants who reported transitioning (lesbian, bisexual, or heterosexual) had higher levels of tobacco smoking and marijuana use compared to those who reported consistent heterosexual attraction. In addition, cisgender female participants who reported consistent lesbian or bisexual attraction at the beginning of the study period reported significantly higher levels of depressive symptoms and suicidal thoughts than those who reported consistent heterosexual attraction. However, young men who transitioned to heterosexual attraction reported higher initial levels of smoking, heavy drinking, and marijuana use compared to those who reported consistent heterosexual attraction. Similarly, Oi and Wilkinson (2018) reported that cisgender men and women who reported same-sex experiences in both adolescence and adulthood had the greatest risk of suicidal ideation. However, cisgender female participants who had their first same-sex experience in adulthood had a lesser decline in suicidal ideation over time relative to those with no same-sex experiences and those with same-sex experiences in adolescence only.

One study found the number of changes in sexual orientation was positively associated with typical weekly alcohol consumption and depression (Feinstein et al., 2019). Additionally, Ott et al. (2013) found higher levels of reported sexual orientation mobilityFootnote 1 were associated with higher levels of substance use (marijuana and substance use). Specifically, the researchers found differences by age and gender. For example, female and male adolescents with a mobility score of 1 had 3.2 and 1.9 higher odds of marijuana use, respectively, than those with a mobility score of 0. Similarly, for smoking among male participants regardless of age, those in a “toward completely homosexual” group had a greater risk of smoking in the past month than the “immobile” group (consistent reporting on sexual orientation).

Other Health Outcomes

Four studies reported on outcomes other than mental health and substance use, including physical activity, body mass index (BMI), diabetes, and earned wages. Fricke and Sironi (2020) found that men who reported a change toward a more homosexual identity indicated significantly more physical activity and lower BMI and those who reported a change toward a more heterosexual identity indicated less physical activity and higher BMI, compared to those who did not report a change. However, for women, a change toward a more homosexual identity was significantly associated with more physical activity and lower odds of being obese compared to no change. Katz-Wise et al. (2014) found that sexual orientation mobility (calculated at each wave based on changes in reported sexual orientation) differed by sexual orientation, with sexual minority female and male participants reporting greater mobility than completely heterosexual participants. Additionally, the study found that sexual orientation mobility was positively associated with BMI in females, but not in males.

Using all four waves of National Longitudinal Study of Adolescent to Adult Health data, Liu et al. (2019) reported that respondents who reported sexual minority status in adulthood only or sexual minority status in both adolescence and adulthood had significantly higher odds of having diabetes compared to heterosexual respondents. Additionally, cisgender female participants with sexual minority status in both adolescence and adulthood had significantly higher odds of having diabetes than heterosexual cisgender female participants. Sabia (2015) reported on earned wages and sexual orientation change. The study found that among cisgender male participants, those who transitioned either toward or away from a same-sex orientation earned lower wages than those who consistently identified as heterosexual. However, among cisgender female participants, only those who transitioned toward a same-sex orientation also reported lowered wages (by 4.0% to 5.5%) compared to those who consistently identified as heterosexual.

Discussion

The reviewed studies lacked agreement in operationalization and assessment of self-reported sexual orientation changes, using repeated measures of one or more aspects of sexual orientation (e.g., sexual identity, attraction, partnering, and behaviors) to report change. The differences in assessment techniques have important implications for how the study results are understood and applied in future work. Some studies have examined changes in multiple dimensions of sexual orientation (such as, Katz-Wise, 2015; Stewart et al., 2019); however, future studies may further explore within-person and between-person differences in changes in multiple dimensions of sexual orientation over time. For example, if there is a cross-lagged effect between multiple dimensions of sexual orientation and if there are differences in trajectories of dimensions of sexual orientation over time. Most studies included in the review used Kinsey-informed Likert scale that measured sexual orientation or attraction on a continuum ranging from 100% opposite sex to 100% same-sex attraction, which assumes that sexual orientation is a two-dimensional experience. Further work must be responsive of current evidence and include language that is inclusive of diversity and breaks away from dichotomous understanding of orientation experiences. Additionally, studies reporting on sexual identity labels and changes can capture contemporary vocabulary and associated meanings. For example, sexual identity labels like pansexual, demisexual, and queer are based on an expanded understanding of self, both in terms of sexual orientation and gender identity, and at times, fall outside of the binary (Belous & Bauman, 2017; Horner, 2007; Morandini et al., 2017; Russell et al., 2009). In contrast, the more traditional assessment of sexual attraction has a more binary operationalization of sexes (male and female) to capture attraction and views heterosexual and homosexuality as two opposite and definitive ends. An under researched area in sexual orientation assessment is endorsement on multiple sexual orientation identities by participants (i.e., studies that allow to select multiple response options), and its implication on change in sexual orientation over time. Limited evidence on who endorse multiple identities exists; where those with nonnormative identities (i.e., plurisexual and gender deviser) are more likely to endorse multiple sexual orientation identities compared to those with normative identities (monosexual and cisgender; Galupo et al., 2015). Though, the prevalence of sexual orientation change among those who endorse multiple sexual orientation identities remain unknown. However, assessing self-perception of sexual orientation change (irrespective of single or multiple endorsements) may help in addressing this. For example, a few studies asked participants to report their attitudes on sexual fluidity and if they experienced changes (Diamond, 2008; Katz-Wise, 2015; Katz-Wise & Hyde, 2015). Such assessments may help capture participants’ self-perceptions regarding identity change and determine if these changes were part of a conscious individual decision (Carter, 2017).

There were mixed findings on difference in reported change in self-reported sexual orientation by gender, whereby many studies indicated that cisgender female participants were more likely to report a change than cisgender male participants (by more than 10 percentage points; Fricke & Sironi, 2020; Oi & Wilkinson, 2018; Savin-Williams et al., 2012; Stewart et al., 2019). However, it is important to note that some studies included in the review (such as, Katz-Wise, 2015; and Ott et al., 2011) found no significant differences in sexual orientation change by gender, especially, among sexual minority participants. To contextualize, a review by Diamond (2016) indicated that sexual orientation change was somewhat more common among cisgender female than cisgender male participants, though no conclusions about the extent of gender differences in sexual fluidity and the causes of such differences could be determined. In contrast, Apostolou (2018) found that many cisgender female participants did not have fluid sexuality but instead had stable attractions over time. This lower prevalence among male respondents could be associated with stricter notions of masculinity, social, and cultural expectations regarding gender, and more negative attitudes toward gay men by heterosexual men (Breen & Karpinski, 2013; Kilianski, 2003); however, these differences and underlying factors have yet to be examined. Although the extant research suggests that these phenomena are somewhat more common in cisgender female than cisgender male participants, it is difficult to draw reliable conclusions about the extent and cause of gender differences in sexual fluidity.

Additionally, only two studies examined other covariates associated with change in sexual orientation. Non-Latino mixed-race cisgender male participants, as compared to non-Latino White cisgender male respondents, had higher odds of changing to a straight identity (Silva, 2018). There were mixed findings on the relationship between religiosity and sexual orientation change, with one study finding support that religiosity among cisgender female was associated with higher odds of changing to a straight orientation (Silva, 2018), while another study did not find any association between religiosity and changes in romantic and sexual attraction over time (Cohen et al., 2020). However, these associations of race and religion with sexual orientation change were not further explored or discussed. It is imperative to further examine how multiple identities (like race, religion, and region) may be associated with sexual orientation development and how this relationship changes over time.

In a review, participants who identified with a nonheterosexual identity or as a sexual minority at baseline were more likely to report a change in sexual orientation (Cohen et al., 2020). The prevalence in sexual orientation change differed by sampling (probability vs. nonprobability) and sample demographics (Cohen et al., 2020; Diamond, 2008; Morgan et al., 2018; Silva, 2018). Studies have found multiple pathways in terms of directionality of change, with more studies reporting a change in the direction of a same-sex identity rather than toward a heterosexual identity (Sabia, 2015; Silva, 2018). However, the directionality of change was observed as toward heterosexual, toward a same-sex identity, or in both directions. Participants who reported a nonheterosexual or bisexual identity orientation at baseline were more likely to report a change toward a more same-sex orientation. Most studies used Kinsey’s scale to operationalize sexual orientation, ranging from completely heterosexual to completely homosexual. These finding are consistent with Hunt et al. (2018) and Morgan (2013), who used traditional and contemporary models of sexual orientation development, identity development trajectories, and change in sexual orientation over time.

A few studies examined the impact of a change in sexual orientation on behavioral health outcomes. This association was examined using the directionality of such change (e.g., a shift away or toward same-sex orientation). Studies found that those who reported changes in sexual orientation toward more same-sex orientation were associated with an increase in depressive symptoms (Everett, 2015) and those who reported heteroflexibility or same-sex orientation had a greater likelihood of reporting depressive symptomology, suicidality, and substance use (Fish & Pasley, 2015; Needham, 2012; Oi & Wilkinson, 2018), compared to those who did not report a change or reported consistent heterosexuality. In addition, more changes in sexual orientation were found to be associated with depression and substance use (Feinstein et al., 2019; Ott et al., 2013). One plausible explanation for these association is that sexual orientation changes may be associated with unique stressors in the form of cognitive and emotional disruptions, as people reconfigure self-relevant schemas and navigate social support networks. This could be characterized as either the loss of existing social support associated with a past orientation or stress with navigating new orientation experiences and new social networks.

Future studies should address the gaps in the literature on the impact of sexual orientation changes on mental health outcomes while reflecting on theoretical frameworks to contextualize the findings. Drawing from sexual minority stress framework (Meyer, 2003), change in sexual orientation processes may also indicate presence of underlying processes (such as, stressors associated with identity management and change) that may help elucidate its relationship to mental health outcomes. For example, Caplan (2017) found that discordance between sexual orientation labels and sexual attraction and behavior was positively associated with higher depression, which may be indicative of stress associated with ability to manage one’s sexual orientation and associated dimensions. In addition, applying social identity theory may be helpful in examining the association between sexual orientation change and mental health outcomes. Social identity theory would suggest that because social identity (such as, sexual orientation) is in part derived from membership in a certain group (i.e., one’s current sexual orientation), experiencing the loss of support from one identity group during a shift in identity would be stressful (Everett, 2015; Stets & Burke, 2000). Though sexual orientation change may be associated with cognitive and emotional disruptions, it is important to reflect on meaning and value one ascribes to different dimensions of sexual orientation which may help explain the relationship between change and mental health (Haslam et al., 2009; Rosario et al., 2011). For example, the literature on social orientation change suggests that the more severe someone perceives their experience of identity change, the greater their level of negative mental health outcomes; however, when they perceive that the direction of their identity change is progressive (rather than regressive), they are less likely to be depressed (Carter, 2017). However, this complicated relationship between orientation change and health outcomes needs further exploration.

Additionally, further research is needed to understand how demographic factors, particularly the experience of multiple minority identities, are associated with changes in sexual orientation. This is imperative in understanding sexual orientation development in a framework of multiple minority processes (e.g., internalized homophobia, racial and homophobic discrimination). Additionally, with limited understanding of the relationship between sexual orientation change and health outcomes, future research must examine how underlying factors and processes like minority stressors, social networks, and community access and belongingness are associated with this relationship over time.

Finally, most studies we identified for this review were conducted with cisgender participants. Given the contemporary evidence that transgender and nonbinary youth also commonly identify with sexual minority labels and orientations, researchers must be cognizant of gender diversity in their samples (Horner, 2007; Morandini et al., 2017; Russell et al., 2009). Future research may also examine gender identity and sexual orientation development among transgender and nonbinary youth and assess if changes in gender identity have an impact on sexual orientation over time.

Limitations of the Review

The conclusions are based on a selection of studies that we identified for inclusion in our review after applying inclusion and exclusion criteria. We used seven electronic databases for our literature search, and it is possible some relevant studies may have been overlooked. Studies that were published in a language other than English or not published in a peer-reviewed journal were not represented. Additionally, as with any literature review, we would like to note publication bias, wherein significant findings are given precedence over null finings in publications. Moreover, given the diversity in capturing various aspects of sexual orientation (identity, labels, and attraction), the extent to which some older studies may be applicable to contemporary evidence is unclear. For example, some studies in this review captured more traditional sexual identities (gay, lesbian, bisexual), whereas more recent studies have documented a rise in nontraditional youth identities, like queer and questioning (Russell et al., 2009). It is also not possible to construct a clear historical narrative on how sexual orientation change has been measured and reported because of the varied study settings and assessments used in the studies. Finally, a very small proportion of studies examined the association between change in sexual orientation and health outcomes, limiting our ability to conclude how changes and directionality of changes may apply to health outcomes in various settings or understand the underlying mechanisms driving these associations.

Conclusion

This review summarized empirical research from the last two decades among adolescent and young adult populations as pertains to self-reported sexual orientation change. The review syntheses the approaches to measuring sexual orientation change; the prevalence, patterns, and directionality of changes in sexual orientation; and how changes in sexual orientation relate to health outcomes among adolescents and youth adults. Given evolving understanding of gender and sexual identities, and access to resources and support, sexual minority adolescents and young adults may present fluid and evolving understanding of self and sexuality. The expanding evidence that sexual orientation changes are associated with negative health outcomes is essential to how we understand sexual orientation identity development among adolescents and young adults. Hence, practice and interventions in the fields of social work, psychology, pediatric medicine, and school settings must be cognizant of the fact that many adolescents and young adults experiencing sexual orientation change may need more support for healthy psychosocial adjustment. Additionally, adolescent sexual orientation identities are a core component of adolescent development and merit continued focus, considering the significant public discourse on non-heterosexual identities and the changing realities of the lives of young people.