Symptoms of mental disorders and oral contraception use: A systematic review and meta-analysis

2 Worldwide, over 150 million adolescent and adult


Introduction
Oral contraceptives (OC) are the most widely used form of reversible contraception among adolescent and adult women globally, with over 150 million users (United Nations, 2019).The oral contraceptive pill, also commonly known as "the pill," provides a more than 99% effective contraception and safe family planning for sexually active women (Trussell and Kost, 1987).Alongside OC, other hormonal contraceptives (HC), such as intrauterine devices (IUDs) and the birth control patch, have also contributed to the expanding array of contraceptive options available to women.
Adolescence marks a pivotal stage for many women to initiate OC-use, coinciding with significant brain development and reorganization (Sharma et al., 2013).However, despite their widespread use, the uptake of OC has declined in recent years (Carrasco-Garrido et al., 2016;Marra et al., 2020).The underlying reasons for this decrease in utilization remain multifactorial and not entirely understood.Nevertheless, an increasing awareness of potential side effects, particularly concerning mental health issues like depression and anxiety symptoms, has garnered significant attention (Robakis et al., 2019).
The influence of OC on mental health is a pivotal area of research, as exogenous sex hormones present in OC may interact with various neurochemical systems, potentially impacting psychological conditions and overall well-being (Schandry, 2016).The effect of HC, including OC, is based on a negative feedback loop in the hypothalamus and the anterior pituitary gland, which suppresses the release of the gonadotropin-releasing hormone (GnRH).Subsequently, decreased levels of the luteinizing hormone (LH) and the follicle-stimulating hormone (FSH) lead to the absence of ovulation.In addition, the suppression of endogenous progesterone levels results in deteriorated conditions for implantation of the egg in the uterus.
Within the realm of OC, two major types are prescribed: progestin-only oral contraceptives (POC) and combined oral contraceptives (COC) (De Leo et al., 2016).COC, commonly known as the "micropill," contain both EE and progestins and are taken following a 21-day cycle with a 7-day pause to mimic the natural menstrual cycle.Besides EE, COC contain and differ in their forms of progestins.Norethisterone, levonorgestrel, dienogest, and desogestrel are among typical progestin types used also in COC (Skovlund et al., 2016), varying in their androgenic activity.On the other hand, POC exclusively contain progestins and find use in specific populations, such as breastfeeding women or those susceptible to estrogen's adverse effects (Worly et al., 2018).They are also known as the "minipill" and are taken continuously without interruption.After the first low dose "minipill" with 0.35 mg norethindrone has been introduced, different forms of progestins associated with different ways of action followed (Shoupe, 2021).Norethisterone (norethindrone) and levonorgestrel, both with a moderate androgenic activity, are among the commonly used progestins in POC (Shoupe, 2021;Yusuf et al., 2019).
Existing literature and systematic reviews have examined the relationship between OC-use and mood or affective disorders, particularly depression (Robakis et al., 2019;Schaffir et al., 2016;Worly et al., 2018).However, none of these studies have systematically reviewed the results of studies focusing on OC-use and various mental symptoms beyond mood disorders.Considering the significant role of sex hormones for mental health (Gogos et al., 2019(Gogos et al., , 2015;;Pinares-Garcia et al., 2018) and the millions of women using OC containing synthetic sex hormones (United Nations, 2019), the primary objective of this review was to investigate whether OC-use is associated with changes in symptoms of mental disorders, focusing on clinically relevant outcomes beyond moodrelated variables including anxiety and depressive symptoms, as well as other psychiatric symptoms.Furthermore, this review aimed to explore the potential impact of the type of OC (COC vs. POC) on the magnitude of the association between OC-use and mental symptoms.COC, containing both estrogen and various progestins, may elicit different effects compared to POC, which consist only of progestins.Additionally, this review considered the duration of OC-use and the age of first-use as potential mediating or moderating variables examining whether earlier initiation of OC-use might have a more significant impact on mental health.Lastly, the review addressed the research question of whether individuals with a prior or current diagnosis of mental disorders may be more vulnerable to the effects of OC on mental symptoms, suggesting a possible interaction between OC-use and pre-existing mental health conditions.

Methods
This systematic review was conducted by using Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines (Moher et al., 2009;Page et al., 2021) and registered on PROSPERO (registration number 338618).A systematic literature search of the electronic databases PubMed, Web of Science, and PsycInfo was performed.The search of titles, abstracts, and keywords with established search strategies was conducted on July 12 th , 2022, by using variations of the keywords of the two categories: "oral contraception" and "mental disorders".Complete search strategies for all three databases and the number of hits can be found in Supplementary Material 1. Further, citations and references of recent reviews, meta-analyses and already included primary studies were analyzed to identify additional studies that could not be found in the electronic search.

Eligibility criteria
After the identification, relevant studies were selected by defined inclusion and exclusion criteria.Studies were included if they met all the following criteria: (1) published between 2000 and June 2022.In recent years, since ~2000, newer OC with reduced doses of synthetic estrogen, known for their decreased side effects, have been introduced and are now widely prescribed and utilized (De Leo et al., 2016;Sondheimer, 2008) (2) in English or German language (3) human study, (4) experimental and epidemiological study designs including randomized controlled trials, non-randomized controlled trials, quasi-experimental, longitudinal studies, prospective and retrospective cohort studies, case-control studies and cross-sectional studies comparing groups of women using no OC such as naturally cycling women or females on other HC with females using OC including COC and POC, (5) participants older than twelve years, (6) type of progestin and (7) (self-reported) clinically relevant symptoms of mental disorders measured with a formal psychological instrument, for example Beck Depression Inventory (BDI) (Beck et al., 2011) or the Symptom-Checklist-90 (SCL-90) (Derogatis and Unger, 2010), a diagnosis of a mental disorder, or psychotropic drug use.All in-and exclusion criteria for studies are completely presented in Supplementary Material 2.

Literature screening and selection process
Studies from all three databases (PubMed n = 821, Web of Science n = 1,845, and PsycInfo n = 1,126) were downloaded and manually included into a Mendeley database.Six additional records identified through other resources were additionally included.Duplicates were excluded (n = 1088).2,710 articles were examined according to the eligibility criteria.First, publication date of records was checked.Articles' titles and abstracts were assessed.Sixty-two studies that appeared to meet the eligibility criteria were included in the full-text analysis and checked based on the eligibility criteria.After excluding 40 studies, the remaining 22 articles were included in this review and further analyzed (see Figure 1).

Data collection
Data of each included study were extracted.This included basic information of the study such as authors and publication year, study design and methodology, as well as information about participants, e.g., participant demographics and baseline characteristics.Also, intervention and results including OC method(s), duration of OCuse, study groups, relevant outcomes, measurements for changes in symptoms of mental disorders and significance of effect were collected.Data was listed according to patterns in response to similar OC methods, populations, outcomes, and results.
Studies' results were analyzed regarding changes in symptoms of mental disorders in females while or after using OC.Measures of effect for the main outcomes included relative risks, odds ratios, risk difference, hazard ratio, and/or mental diagnoses.Also, significance of effects was noted.For binary studies, results were statistically pooled and summarized.Furthermore, descriptive and narrative forms for presenting synthesized results were chosen.

Meta-analysis
The meta-analysis was performed using the R statistical software (version 2022.02.2) and the 'meta' and 'metafor' packages.Both fixed-effects and random-effects models were utilized to estimate the pooled risk ratio (RR).Heterogeneity among studies was assessed using the I-squared statistic.Publication bias was assessed through a combination of funnel plot visualization and the Egger's test.The funnel plot displayed the relationship between effect sizes and their precision, aiding in the visual detection of asymmetry.The Egger's test provided a formal statistical assessment of publication bias.2.6.Methods for risk of bias and level of evidence assessment Studies were evaluated regarding their risk of bias and certainty of evidence.The method sections of each included article were screened and risk of bias in individual studies was evaluated in accordance with the standardized critical appraisal tools by Joanna Briggs Institute (Joanna Briggs Institute, 2020).By reading the method section, criteria regarding for example participants, allocation to treatment groups, outcome measurement, or use of appropriate statistical analysis specified in the critical appraisal tools were checked.Studies of at least moderate and higher methodological quality were included in the final analysis.Methodological quality of all included studies was found to be at least moderate.Hence, no study was further excluded based on its risk of bias.Detailed information on individual study procedure and analyses are given in the individual studies itself.
In addition to quality assessment of each individual study, they were assigned to various levels of evidence.Figure 2 shows the hierarchy of evidence level produced for and used in this systematic review.It was deduced and adapted to the evaluation in this systematic review (Burns et al., 2011).Due to excluding studies without original empirical data in this systematic review, randomized controlled trials were assigned to level I.Each included study was allocated to the appropriate level of evidence (see Table 2).
To reduce the risk of bias, ClinicalTrial.gov was searched.The aim was to identify studies in progress that may be relevant to the topic or for studies having been registered, but not published yet or not published at all for any reasons.Search in this register was conducted on September 7 th , 2022.A similar strategy to the search strategies used in this systematic review was applied to the search in ClinicalTrial.gov.Terms of the two categories "oral contraception" and "mental disorders" were multistage combined.Subsequently, the titles and, where appropriate, the abstracts of the given hits were reviewed for possible inclusion in this work.Potential eligibility was assessed by considering the questions, hypotheses, and eligibility criteria of this systematic review.Moreover, status of identified examinations was noted.The search in ClinicalTrials.govon 9 th September 2022 using the terms "Oral contraception" and "Psychiatric Disorder", which also included the terms "Oral Contraceptives", "Mental Disorder", "Birth control pill" and more resulted in 26 hits.By evaluating the titles, no identified study was considered as potentially relevant for inclusion in this systematic review.

Literature screening and study selection
A total of 22 studies were eligible to be included in this systematic review (see Figure 1, Table 1).

Characteristics of included studies
Relevant information including references, the examined diagnosis of mental disorders, information about study design, participants, and OC methods, used measurements and results extracted from the individual studies are presented in Table 2.In Supplementary Material 3, detailed information about the study design, participants, OC preparations, and study groups are listed.Supplementary Material 4 provides detailed information about measurements, including relevant outcomes, times of measurement and used instruments, descriptive summaries of the results of the included studies, and author's conclusions.Supplementary Material 5 provides a comprehensive and elaborated description of all included studies.

Mental health conditions in OC-use
The majority of studies (n=15) investigating the relationship between OC-use and symptoms of mental disorders reported no significant association (Akin et al., 2010;Cinar et al., 2012;Duke et al., 2007;Larsen et al., 2022;Lundin et al., 2022;McKetta and Keyes, 2019;Morssinkhof et al., 2021;Neri et al., 2017;O'Connell et al., 2007;Raeder et al., 2019;Rapkin et al., 2006;Toffol et al., 2012Toffol et al., , 2011;;Wit et al., 2020;Zethraeus et al., 2017).However, one study examining the impact of OC-use on exposure therapy benefits in specific phobia found no significant difference between the naturally cycling group and the OC group from pre-to post-treatment but observed a significant difference between groups at 6 weeks follow-up (Raeder et al., 2019).
Some investigations (n=3) suggested that OC-use might be a potential long-term risk factor for mental disorders (Anderl et al., 2022(Anderl et al., , 2020;;Skovlund et al., 2016).Particularly, three studies reported a significant association between OC-use and depression, indicating that adolescent OC-use might be linked to a higher risk of depression in adulthood (Anderl et al., 2022(Anderl et al., , 2020;;Skovlund et al., 2016).Details about these findings will be explained in more detail in the following chapters.A few studies (n=3) indicated a potential beneficial effect association between OC-use on and mental health (Cheslack-Postava et al., 2015;Kim et al., 2022;Paoletti et al., 2004).Notably, mentally healthy women using OC during the third month of intake demonstrated significantly lower intensity of anxiety, phobia, and paranoid ideation compared to their natural menstrual cycle and to mentally healthy women in the control group (Paoletti et al., 2004).Moreover, the prevalence of depression, anxiety disorder, and panic disorders in women using OC in the past year were significantly lower compared to former users and never users (Cheslack-Postava et al., 2015).One study examined the risk of dementia in women with depression, depending on hormonal contraceptive use (Kim et al., 2022), and reported a decreased risk of developing Alzheimer's disease in those with lifetime OC-use.Svendal et al. (2012) reported both protective and adverse relations effects of between OC-use on and mood disorders, depending on the type of OC preparation.Further details on these findings will be presented in the next paragraph.

Meta-analysis
A binary meta-analysis was conducted to investigate the association between OCuse and symptoms of mental disorders.In conducting the meta-analysis, it is important to note that not all eligible studies could be included in the final analysis.Some studies were excluded due to the unavailability of binary data required for the meta-analysis (see Supplementary Material).Additionally, a subset of studies had missing values in key variables, which precluded their inclusion in the quantitative synthesis.The decision to exclude these studies was made to ensure the integrity and accuracy of the meta-analysis results.The analysis included a total of 16 studies, with a combined sample size of 10,642,840 observations (Akin et al., 2010;Anderl et al., 2022Anderl et al., , 2020;;Cheslack-Postava et al., 2015;Cinar et al., 2012;Duke et al., 2007;Kim et al., 2022;Larsen et al., 2022;Lundin et al., 2022;McKetta and Keyes, 2019;Morssinkhof et al., 2021;Neri et al., 2017;O'Connell et al., 2007;Paoletti et al., 2004;Raeder et al., 2019;Rapkin et al., 2006;Skovlund et al., 2016;Svendal et al., 2012;Toffol et al., 2012Toffol et al., , 2011;;Zethraeus et al., 2017).The forest plot in Figure 3 (Panel A) illustrates the individual study RR estimates and their corresponding 95% confidence intervals.The summary RR estimate, represented by the diamond at the bottom, indicates the combined effect with its confidence interval.The meta-analysis yielded a pooled RR of 1.44 (95% CI [1.43,1.46]) in the common effect model, suggesting a positive association.The pooled RR of 1.00 (95% CI [0.82, 1.22]) in the random effects model indicates a non-significant association.

Heterogeneity Assessment
Heterogeneity among the included studies was assessed using the I-squared statistic, which indicated substantial heterogeneity (I² = 99.8%.The value of H is estimated as 22.02, which represents the heterogeneity factor.The test of heterogeneity was statistically significant (Q = 7275.42,df = 15, p < 0.001), suggesting that variability beyond chance likely exists.The quantification of heterogeneity revealed a τ 2 estimate of 0.1179, indicating the variability in true effects among studies.The restricted maximumlikelihood estimator for τ 2 and τ suggested a between-study variance of τ 2 = 0.1179 (95 CI [0.0467, 0.3097]), corresponding to τ = 0.3434 (95% CI [0.2161, 0.5565]).3.4.3.Publication Bias Assessment Publication bias was not indicated by the Egger's test (p = 0.988), suggesting that the results are unlikely to be heavily influenced by publication bias (Figure 3, Panel B and C).
Among the four studies focusing solely on COC-use, two found no significant effect relationship between COC-use and on mental symptoms (Neri et al., 2017;Toffol et al., 2011).Notably, Neri et al. (2017) observed that women treated with COC for 12 months did not experience changes in mental symptoms, but non-pathological symptom scores significantly decreased during the first cycle of COC treatment.Toffol et al. (2011) found no significant differences in mental health effects between second and third generation COCs.Conversely, Paoletti et al. (2004) reported significantly lower levels of anxiety, phobia, and paranoid ideation in the OC group during the third month of COCintake compared to the natural menstrual cycle and a control group.Cheslack-Postava et al. ( 2015) reported both reduced and increased risks for COC-use depending on the preparation.Monophasic COC-use was associated with a reduced risk for anxiety and depressive disorders, while multiphasic COC-use showed an increased risk for these disorders.The study also revealed significant differences in the association of monophasic versus multiphasic COC with subthreshold generalized anxiety disorder and major depressive disorder diagnoses.
Regarding POC-use, four studies compared COC-and POC-use (Larsen et al., 2022;Lundin et al., 2022;Skovlund et al., 2016;Svendal et al., 2012).The overall findings suggest that POC-use is associated with a higher risk of mental symptoms compared to COC-use.However, the differences in risks were marginal in some studies (Lundin et al., 2022;Skovlund et al., 2016).In the study of Svendal et al., 2012.P participants currently using POC showed an increased likelihood of a current mood disorder compared to non-progestin using women, while COC-use was associated with a decreased risk of a current mood disorder compared to non-users.An analysis of COC and POC-users in Larsen et al. (2022) could not be performed due to the limited participation of only six women using POC.
3.6.Influence of other variables: Age of first-use, duration of use, and recent or previous mental disorders A total of nine studies were included in the systematic review, investigating the impact of age of first-use and duration of use on the association between oral OC-use and symptoms of mental disorders (Anderl et al., 2022(Anderl et al., , 2020;;de Wit et al., 2020;Duke et al., 2007;Lundin et al., 2022;McKetta and Keyes, 2019;O'Connell et al., 2007;Skovlund et al., 2016;Toffol et al., 2011).
Regarding age of first-use, four studies found no significant association between age of first-use and mental symptoms (de Wit et al., 2020;Lundin et al., 2022;McKetta and Keyes, 2019;O'Connell et al., 2007), while four studies reported an increased risk of experiencing symptoms of mental disorders when using OC in adolescence or early adulthood (Anderl et al., 2022(Anderl et al., , 2020;;Lundin et al., 2022;Skovlund et al., 2016).For instance, Skovlund et al. (2016) reported an association between depression diagnosis and subsequent antidepressant use, particularly among adolescents aged 15 to 19 years, while Anderl et al. (2020Anderl et al. ( , 2022) ) found that women who used OC during adolescence were more likely to experience major depression as adults.Lundin et al. (2022) also observed a minor increased risk of depression in adolescents using POC.
Regarding duration of OC-intake, Duke et al. (2007) reported a statistically significant inverse relationship between years of OC-use and depressive symptoms, suggesting a decline in depressive symptoms with increasing years of OC-intake, potentially plateauing after five years.Additionally, Toffol et al. (2011) found a beneficial effects relationship with on mood depending on the duration of current OC-use, with specific associations observed for certain depression-related items.However, the crosssectional designs, with partly retrospective data collection, preclude any causal conclusions.Furthermore, the authors clarify that there is no independent effect of OCuse on depressive symptoms.
Two studies investigated the influence of previous diagnoses of mental disorders.Morssinkhof et al. (2021) found no relationship between OC-use and mental symptoms, irrespective of previous mental disorder history.However, interestingly, Anderl et al. (2022) suggested an increased risk of experiencing depressive symptoms among adolescents with no history of major depressive disorder.These findings suggest that previous mental disorder history is not significantly associated with the development of current mental disorders when using OC.

Discussion
This review investigated the association between OC-use and symptoms of mental disorders.The investigation of other psychological, clinically non-relevant variables, such as mood or well-being, was not relevant in the current review (for reviews and further details on OC-use and mood please see Robakis et al. (2019) and Schaffir et al. (2016).A total number of 22 studies examining the association between OC-use and symptoms of mental disorders were identified with ranging sample sizes from n=22 to n=1,061,997 individuals.Most of the included studies (n=15) reported no statistically significant relationships between OC-use and mental symptoms, while some studies (n=5) suggested OC-use as a potential risk factor for mental disorders.Conversely, three studies indicated a potentially beneficial association with mental health.The results of the current review indicate that a relationship between symptoms of mental disorders due to and OC-use solely is very unlikely in the general population but cannot be completely disregarded.In accordance with this, the meta-analysis suggests a significant association between OC-use and symptoms of mental disorders based on the common effect model.However, the non-significant result from the random effects model, along with the substantial heterogeneity observed, underscores the need for cautious interpretation and consideration of study variability.The influence of other variables, such as the age of first-use and duration of use, on the association between OC-use and mental disorders was also examined.Some studies reported an increased risk of experiencing mental symptoms in adolescence or early adulthood when using OC early in life, while others found no significant effects relationships.Recent or previous diagnoses of mental disorders did not appear to have an impact on the relationship between OC-use and mental health.Subsequently, the review explored the association of COC vs. POC with symptoms of mental disorders.Eight studies provided information on OC preparations, with mixed results regarding the association between different preparations and mental health outcomes.The findings indicated both reduced and increased risks for symptoms of mental disorders depending on the OC type.

Mental health conditions in OC-use
Most studies reported no associations between developing symptoms of mental disorders and OC-use (Akin et al., 2010;Cinar et al., 2012;Duke et al., 2007;Larsen et al., 2022;Lundin et al., 2022;McKetta and Keyes, 2019;Morssinkhof et al., 2021;Neri et al., 2017;O'Connell et al., 2007;Raeder et al., 2019;Rapkin et al., 2006;Toffol et al., 2012Toffol et al., , 2011;;Wit et al., 2020;Zethraeus et al., 2017).This is consistent with other findings that reported also no clear general relationship between OCs and depression scores or incident diagnoses of depression (Böttcher et al., 2012;Worly et al., 2018).Some studies reported non-significant results.In general, null-results can be due to a lack of true effects or due to methodological issues such as lack of power or lack of reliable measures.Several studies, including those by Akin et al. (2010) 2007), and Rapkin et al. (2006), have reported no direct link between OC-use and the risk of depressive symptoms, thus suggesting the absence of an association between OC-use and depression symptoms in their findings.In contrast, Morssinkhof et al. (2021) conducted post-hoc analyses, revealing significant associations between depression and OC-use, implying that while OCs may not exhibit an overall association with depression, certain individuals might experience OC-related depressive symptoms.It's worth noting that some studies, like McKetta et al. (2019), had limited numbers of OC-users, potentially affecting the statistical power of their results.No clear pattern emerges in studies that failed to identify an association between OC-use and symptoms of mental disorders.Therefore, to this day, there is no sufficient explanation for null-results.However, some studies reported an increased risk (Anderl et al., 2022(Anderl et al., , 2020;;Kim et al., 2022;Paoletti et al., 2004;Skovlund et al., 2016;Svendal et al., 2012), and some studies reported a reduced risk for mental symptoms due to OC-use (Cheslack-Postava et al., 2015;Paoletti et al., 2004;Svendal et al., 2012).
The ambivalence in results may possibly be attributed to different sample sizes across the studies, sample characteristics (e.g.antidepressant-use), data collection methods (e.g.standardized questionnaires and diagnosis of a mental disorder), or study designs (e.g.longitudinal study, descriptive/observational study, or randomized controlled trial).Furthermore, most of the studies included in this review only compare current OC-use with current non-OC-use, without separating never users and former users.This analytical approach can lead to substantial underestimation of effects of OCuse on mental health outcomes.For this reason, interpretations that are based on studies using the common analytic approach combining never and former OC-users should be taken with caution.For a detailed discussion of the potential substantial underestimation of effects of OC-use on mental health based on the common approach described above please see Johansson et al., 2023.The inconsistent findings in the published literature may also be explained by the healthy user bias: Women encountering substantial depressive symptoms briefly after the beginning of HC-use are more likely to discontinue its usage, and they may not be well-represented in the group of long-term users (de Wit et al., 2020).Conversely, those who continue using HC for several years are more likely to experience positive outcomes while using it.For a detailed discussion why the healthy user bias may in fact contribute to the inconsistent findings please see Johansson et al., 2023.Additionally, the purpose of the data collection should be considered as a possible explanation for inconsistent results.Some results were reported from register-based cohorts (e.g.Lundin et al., 2022) whose original intent was not to investigate the effects of OC-use on symptoms of mental disorders.Therefore, important information (like menstrual cycle phases, history of OC-use etc.) could be missing leading to diverse results.Other studies (e.g.Paoletti et al., 2004;O'Connell et al., 2007) were designed for the purpose of investigating the impact of OC on mental health and data was collected according to that.However, these studies have smaller sample sizes compared to the register-based cohorts.Hence, the comparability of the studies is limited.Lundin et al. (2022) explain the discrepancies with previous longitudinal register-based studies by residual confounding, i.e. individual-level confounding that could not be captured by the registers used, rather than by an actual (positive or negative) association between OC exposure and outcome.The observed heterogeneity among the included studies, underscores the importance of considering sources of variability.While the common effect model assumes a single underlying effect size, the random effects model acknowledges the presence of heterogeneity and provides a more robust estimate by accounting for between-study variability.Further, in some studies a long-term association between previous OC-use and subsequent mental health symptoms was observed, which was not considered in other investigations (Anderl et al., 2022(Anderl et al., , 2020;;Raeder et al., 2019;Skovlund et al., 2016).
Furthermore, it is important to consider the distinction between relative risk and absolute risk in association with mental symptoms and OC-use, as elucidated in a rapid response by Kritsotakis et al. (2016).Absolute risk offers a direct measure of the actual risk, whereas relative risk serves to compare the risk between two cohorts, highlighting the difference in risk associated with OC-use in this context.Both measures are valuable in epidemiology and medical research, as they provide different perspectives on the impact of risk factors on health outcomes.However, a critical examination of the raw data presented by Skovlund et al. (2016), who reported a 23% elevated relative risk of antidepressant-use in association with OC-use, reveals a marginal absolute risk increase of merely 0.45 cases per 100 women per year, corresponding to a 0.45% overall risk.Notably, the absolute risk for a confirmed diagnosis of depression is even lower at 0.04% (Kritsotakis, 2016).
Thus, taken the results of this systematic review into consideration, emerging symptoms of mental disorders due to OC-intake alone seem unlikely, however, it remains possible that mental health relevant effects of taking OC may occur in the general population.This is in line with recently published results of a slightly, but significantly, increased risk of depression due to OC within the first two years of use.Ever OC-use was associated with a slightly increased lifetime risk of depression, but was not as pronounced after the first two years of use (Johansson et al., 2023).There might be a selected subset of women at higher risk for experiencing symptoms of mental disorders or showing an altered antidepressant drug treatment response.For instance, Larsen et al. (2022) speculated that women with depression using OC might represent a specific serotonin subtype of MDD, which might negatively influence the treatment response of antidepressants.The likelihood for first-use of antidepressants and first depression diagnosis in OC-users compared to non-users was slightly increased in the study by Skovlund et al. (2016).However, these findings might also be related to other factors: (1) women who tend to use medications in general may also use OC and have a higher propensity to use antidepressants as well (2) use of antidepressants could not be equated with having depressive symptoms.Antidepressants are taken for various reasons.In accordance with that, the two included studies on previous diagnoses of mental disorders (Anderl et al., 2022;Morssinkhof et al., 2021) reported no relationship between OC-use and current or previous mental disorders.

COC and POC
When it comes to the influence of OC-use on mental health, it is important to note the composition of the preparations.Here, we considered COC-and POC-preparations of OC.Four studies considered COC-preparations and reported a potentially favorable association with mental health (Cheslack-Postava et al., 2015;Lundin et al., 2022;Paoletti et al., 2004;Svendal et al., 2012).In this line, it should be noted, while some studies did not report changes in clinically relevant symptoms, a few reported an improvement in mental symptoms that did not meet diagnostic criteria during COC-intake (Neri et al., 2017;Paoletti et al., 2004;Svendal et al., 2012).While monophasic COCpreparations seem to have beneficial effects, multiphasic preparations (Cheslack-Postava et al., 2015) and preparations with EE more than 0.05 mg EE (Skovlund et al., 2016) showed higher rates in first antidepressant use and diagnosis of depression.An association between EE dosages of more than 0.03 mg and depressive symptoms was reported in the past (Herzberg et al., 1970).To minimize side effects of estrogen, the dosage of EE in most available COC is as low as possible (0.02 to 0.035 mg) (Lawrie et al., 2011).However, traditional COC with EE higher than 0.035 mg are still in use (Skovlund et al., 2016).POC-preparations were associated with an increased risk for mental symptoms (Lundin et al., 2022;Skovlund et al., 2016;Svendal et al., 2012).These results were consistent across all studies that considered POC-preparations.However, difference in decreased depression rates between COC-users and increased depression rates in POC-users were marginal (Lundin et al., 2022;Skovlund et al., 2016;Svendal et al., 2012).The evidence for an association between POC-use and a higher risk of mental symptoms compared to COC-use is inconclusive.However, emerging indications suggest that the formulation of OC might play a role in mental health outcomes, and therefore this matter deserves further investigation.Following this, conclusions about differences of specific OC-preparations and mental symptoms remain unclear.Furthermore, most studies did not consider the specific progestin-type contained in either COC or POC.It should be noted that different types of progestins may differ in their ways of action and, therefore, in their potential effects.Toffol et al. (2011) was the only study examining differences between second and third generation COC (Toffol et al., 2011).The generation of OC, and in this context also its androgenicity, could further represent an explanatory factor for the development of mental symptoms.

OC-use in adolescence
Mediating factors, such as age at first-use and duration of OC-use were considered in this systematic review.The indications of an increased risk for experiencing depressive symptoms in adolescent OC-users should be noted (Anderl et al., 2022(Anderl et al., , 2020;;Wit et al., 2020).Moreover, women who had used OC during adolescence were more likely to have a first depressive episode as an adult (Anderl et al., 2022).But there is also data showing the contrary (Doornweerd et al., 2022) .Adolescence is a sensitive period for the onset of mental disorders (Fuhrmann et al., 2015).Changes on a behavioral level during the time of adolescence, such as investing in peer bonding and involvement in high-risk behavior (McClure et al., 2014;Romer, 2010;Whelan et al., 2012), are associated with the rise in sex hormones (Forbes and Dahl, 2010) One explanation for this change in behavior can be linked to brain development.In this line, adolescence is a vulnerable time for brain development involving synaptic pruning and maturation of the brain, especially the prefrontal cortex which is involved in cognitive functioning (Herting et al., 2015).Studies reported associations of estradiol, testosterone, and LH with brain gray and white matter in adolescence (Koolschijn et al., 2014;Peper et al., 2009Peper et al., , 2008)).A suppression of these crucial sex hormones in those using OC during adolescence may, thus, interfere with normal brain development and behavior that is dependent on natural endogenous hormonal fluctuations during puberty.There are some studies that reported changes in brain volume and function in women taking OC compared to women with a natural menstrual cycle (Heller et al., 2022;Lisofsky et al., 2016;Rehbein et al., 2021).However, to the best of our knowledge, there are no studies specifically investigating the influence of sex hormones on brain structure and function in adolescents taking OC.To date, little is known about the extent to which adolescent OC-intake shapes/influences the development of the brain (Sharma et al., 2020).According to this systematic review, there is some evidence that adolescent-use of OC might impact the emergence of symptoms of mental disorders at the current age but also in adult life.This is consistent with a recently published population-based cohort study in 264.557 women from the UK Biobank (Johansson et al., 2023).In this context, the age of first-use of OC seems to have more impact on current or later mental symptoms than the duration of OC-intake.Results, furthermore, indicate rather beneficial effects on mood with increasing years of OCintake (Duke et al., 2007;Toffol et al., 2011).However, the misleading interpretation that long-enough duration of use, OC-use may eventually become protective is not warranted by the presented evidence.Longitudinal patterns like this are often driven by, or at least strongly affected by, the healthy user bias (Johansson et al., 2023).Women who experience substantial depressive symptoms within the first few months of initiating OCuse are disproportionately likely to discontinue use and are unlikely to be represented in the sample of long-term users.Conversely, women who are still in the group of OC-users after several years are disproportionately likely to be the ones who have "thrived" under OC-use.It is important to note, that current investigations are scarce, further investigations are needed and no conclusive statements of whether OC-use is beneficial or disadvantageous good or bad should be drawn at this point.Rather, the choice of using OC is an individual decision that should be made after careful consideration of the patient's medical history and developmental age.

Implications for clinical practice
In clinical practice, arguments for and against OC-use should be considered.OC provide an effective option for contraception and safe family planning (McKenna and Fogleman, 2021;Osayande and Mehulic, 2014).However, they are increasingly used by adolescent and young women not only to control fertility but for a number of other treatments (Upadhya et al., 2017), including endometriosis (Brown et al., 2018), polycystic ovary syndrome (PCOS) (Vrbíková and Cibula, 2005), acne, or hair loss (Eichenfield et al., 2021;Williams et al., 2021).Hence, the effects of OC go beyond their original purpose of contraception.The results of this systematic review demonstrate no clear proof that OC-use is accompanied by symptoms of mental disorders.Despite the uncertain effects OC-intake might have on mental disorders, women who are prescribed OC should be monitored for emerging and changes in psychological symptoms.Furthermore, women considering OC-use need to receive adequate counseling regarding potential beneficial as well as negative side effects.After careful consideration of individual risk factors, the decision for or against OC depends on the current life situation and should be re-evaluated regularly.

Limitations of evidence
The results of this systematic review depend on the available information of the conducted studies.Following this, every single study shows its own limitations.Most included studies preclude any causal relationships between effects of OC-use on mental symptoms due to their study design.Available evidence is provided by predominantly cross-sectional, cohort and case-control studies.The lack of randomized controlled trials examining the relevant association should be considered.Moreover, some studies contained missing or unclear information (e.g.OC preparation, duration of use).Incomplete or unclear data resulted in limited interpretations.Additionally, most included articles provided no detailed information on OC methods or duration of intake.It is suspected that this information was not assessed due to the high cost, retrospective analyses or due to the difficult assessment in studies with 1,000 or more participants.Moreover, some studies are also limited by their small sample sizes.Also, several included articles did not gather or did not analyze data regarding potential confounding factors, such as recent or previous diagnosis of mental disorders, duration of use, or specific OC types.Thus, variables functioning as moderator or mediator in the association of OC and mental symptoms were often not considered.Further, most of the data were largely analyzed retrospectively.As hormonal concentrations change across both the natural menstrual cycle as well as across OC-intake, longitudinal studies that track individuals over time are of particular importance.
Most included studies addressed depression (n=17).Three articles also included anxiety disorders in addition to depression.A lack of studies investigating other mental disorders such as anorexia nervosa or schizophrenia must be noted.Hence, no conclusion considering OC-use and the wide range of mental disorders can be drawn.For instance, women with anorexia nervosa are frequently put on OC because of irregular menstrual cycles to amenorrhea, the absence of the menstrual cycle (Milano et al., 2022).It remains to be explored whether OC-use might further influence mental symptoms in these individuals.4.6.Limitations of this systematic review and its process Risk of bias and level of evidence were assessed by only one author.No proofed risk of bias assessment tool was used due to the differences and uncertainty in study designs of included studies.Despite a conscientious assessment, assessment biases cannot be excluded.Moreover, included studies varied in several aspects such as study design, OC methods, examined study groups, relevant outcomes, and measurements.Therefore, only 16 out of the 22 studies could be included in the meta-analysis.

Conclusion
The majority of included studies suggest that OC-use is not significantly associated with the development of mental symptoms in the general population.However, findings indicate a potential increased risk for mental symptoms in certain subgroups, particularly in adolescent OC-users and those using progestin-only OC preparations.It is important to acknowledge the limitations of the available evidence, which primarily consist of cross-sectional and cohort studies.To establish more definitive conclusions, future research should incorporate randomized controlled trials and longitudinal studies.Additionally, further investigations into the influence of different OC preparations and progestin types on mental health outcomes are needed.In clinical practice, OC remain an effective and widely used contraceptive option.Healthcare providers should continue to discuss potential benefits and risks of OC-use with their patients, taking into account individual medical history and risk factors.Regular re-evaluation of the decision to use OC is crucial to ensure the most suitable contraceptive method for each individual.Ultimately, a balanced and informed approach is essential when considering OC-use in the context of mental health and overall well-being.and their corresponding 95% confidence intervals (95% CI).The summary RR estimate, represented by the diamond at the bottom, indicates the combined effect with its confidence interval.Panel B) Funnel plot displays the effect sizes (risk ratios) of individual studies against their standard error.Panel C) displays the results of the publication bias assessment using a regression analysis.It shows the observed effect sizes (log of the risk ratios) plotted against their standard errors.Please note that six studies were excluded from the meta-analysis due to the unavailability of binary data or missing values in key variables.The decision to exclude these studies was made to ensure the integrity and accuracy of the meta-analysis results.

Highlights
1. Most studies found no significant relationship between OC-use and mental symptoms.
2. Adolescent OC-users showed an increased risk for depressive symptoms, highlighting a subgroup vulnerability.
3. The link between progestin-only OC-preparations and a higher risk of mental symptoms is inconclusive.
4. The meta-analysis showed a pooled association between OC-use and mental symptoms.
5. The non-significant random effects model and substantial heterogeneity warrant cautious interpretation.

Figure 3 :
Figure 3: Panel A) Forest plot illustrates the individual study relative risk (RR) estimatesand their corresponding 95% confidence intervals (95% CI).The summary RR estimate, represented by the diamond at the bottom, indicates the combined effect with its confidence interval.Panel B) Funnel plot displays the effect sizes (risk ratios) of individual studies against their standard error.Panel C) displays the results of the publication bias assessment using a regression analysis.It shows the observed effect sizes (log of the risk ratios) plotted against their standard errors.Please note that six studies were excluded from the meta-analysis due to the unavailability of binary data or missing values in key variables.The decision to exclude these studies was made to ensure the integrity and accuracy of the meta-analysis results.
Note.M = mean.OC = oral contraception.COC = combined oral contraception.POC = progestin-only oral contraception.OR = odds ratio.RR = risk ratio.HR = hazard ratio.a↓ decreased likelihood of a current mood disorder; ↑ increased likelihood of a current mood disorder b risk of AD for OC use < 1 year / ≥ 1 year

Table and figure legendsTable 1 :
Included studies.

Table 2 :
Summary of included studies.

Table 2 .
Summary of included studies