The U-shaped relationship between parental age and the risk of bipolar disorder in the offspring: A systematic review and meta-analysis

Parenthood age may affect the risk for the development of different psychiatric disorders in the offspring, including bipolar disorder (BD). The present systematic review and meta-analysis aimed to appraise the relationship between paternal age and risk for BD and to explore the eventual relationship between paternal age and age at onset of BD. We searched the MEDLINE, ∗


Introduction
A basic tenet of psychiatry is that the disease phenotype results from a combination of genetic and environmental effects.Indeed, bipolar disorder (BD) is a severe psychiatric condition with a prevalence in the general population of 2% ( Carvalho et al., 2020 ) and an estimated family-and twinbased heritability of 60-85% ( McGuffin et al., 2003 ).However, genetics only explains a part of the etiology of BD, thereby a growing body of evidence has related diverse elements of the environment that may influence the risk for BD ( Bortolato et al., 2017 ).The broad range of environmental risk factors vary on time of exposure and may include both pre-or peri-natal factors, adverse childhood experiences, infections, as well as urbanization, migration, major stressful life events, and the misuse of drugs or alcohol ( Blanco et al., 2017 ;Grant et al., 2005 ;Kessing et al., 2004 ;Pacchiarotti et al., 2013 ;Vieta, 2014 ).Globally, environmental factors contribute up to 32% of the risk of BD ( Polderman et al., 2015 ), proving to be appealing potentially modifiable targets for preventive strategies.
Parental age is a well-known factor involving environmental and genetic contributions that impacts the risk of several disorders in the offspring ( Malaspina et al., 2015 ;Reichenberg et al., 2006 ).Both extremes of the reproductive age have been associated with worse outcomes, either regarding mother's age, father's age, or both ( McGrath et al., 2014 ).A notable example is the higher risk of Down's syndrome in older mothers ( Mai et al., 2013 ), while a younger childbearing age is associated with educational underachievement, juvenile crimes, substance misuse, and mental health problems ( D'Onofrio et al., 2014a ;Sujan et al., 2022 ).Advanced paternal age at birth has been associated with an increased incidence of several neurodevelopmental and psychiatric disorders, including schizophrenia and autism spectrum disorders ( Khachadourian et al., 2021 ;Reichenberg et al., 2006 ).Mechanisms underlying these relationships are not fully understood, with findings suggesting a role for de-novo mutations in the male germline ( Fromer et al., 2014 ;Malaspina, 2001 a) or even a greater likelihood for pregnancy complications which are more common with delayed parenthood ( Chudal et al., 2017 ;Garcia-Rizo and Bitanihirwe, 2020 ;Giménez et al., 2019 ;Talati et al., 2013 ).Also, advanced paternal age seems to increase the risk for earlyonset psychoses due to the possible role of accumulating age-related DNA mutations ( Wang et al., 2019 ).
However, there is still discordant evidence on the association between advanced parental age and risk of BD in the offspring, with some large studies proving the positive association between advanced paternal age and BD risk ( D'Onofrio et al., 2014 a;Frans et al., 2008 a;Weiser et al., 2020 ), whilst others have shown no significant association ( Buizer-Voskamp et al., 2011 ;McGrath et al., 2014 ).
Considering the increasing trend in average parental age (Bray et al., 1978), the elucidation of such an association might have societal and public health implications.Also, evidence of parental age effects on BD risk may provide insights into the etiology of this complex, multifactorial disorder.
Thus, the first aim of this systematic review and metaanalysis is to determine whether parental age is associated with an increased risk of BD in the offspring.Also, considering that early-onset BD displays homogeneous characteristics and that specific risk factors might operate in this subgroup ( Chengappa et al., 2003 ;Connor et al., 2017 ), a secondary aim is to assess whether advanced parental age is associated with an earlier onset of BD.

Material and methods
The present Systematic Review and Meta-analysis was conducted according to the 2020 version of the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines ( Page et al., 2021 ) and the Meta-analysis of Observational Studies in Epidemiology (MOOSE) ( Stroup, et al., 2000 ) (Supplementary 1 and 2).The protocol of this systematic review and metaanalysis was registered on the International Prospective Register of Systematic Reviews (PROSPERO) ( https://www.crd.york.ac.uk/PROSPERO/ ; protocol CRD42021293319).

Eligibility criteria and study outcomes
Original studies were eligible for inclusion if: (i) the study population consisted of people with BD diagnosed according to the Diagnostic and Statistical Manual for Mental Disorders (DSM) any edition ( APA, 1994( APA, , 2013 ) ) or the International Classification of Diseases (ICD) any edition (WHO 2004) diagnostic criteria with validated diagnosis through structured interviews, and the same criteria had to be applied also to the non-BD group to exclude the presence of psychiatric diagnoses; (ii) they examined maternal and/or paternal age as the variable of interest; (iii) they assessed the association between advanced parental age and risk of BD; (iv) they used a standardized format for presentation of data, allowing for comparisons between studies and calculation of crude ORs.No restrictions about language were applied.Whenever multiple studies considered overlapping study populations, the largest one with the most complete data was included.Reviews, case reports, case series, and studies conducted on animals were excluded.Search strings for the other databases are available in the Supplementary material.The references of each included study, textbooks, and other material were hand searched to identify potential additional studies not captured by the original search-string.

Study selection and data extraction
Two independent reviewers (GF and MDP) independently screened each study for eligibility.When a consensus could not be achieved, a third author (VO) was consulted.
The following data were extracted (when applicable): author(s), publication year, study design, geographical region, country, diagnostic criteria considered, (semi)structured interview adopted, number of BD cases, number of non-cases, BD type, age of BD onset, birth years, estimates of relative risk of BD (odds ratios from casecontrol or incidence rate ratios or hazard ratios from cohort studies), maternal and paternal age and how it was modeled (e.g., categorically).In the case of studies fulfilling inclusion criteria without fully available raw data, authors were contacted up to 2 times to ask for data.For studies reporting data in figures only, WebPlotDigitizer ( https://automeris.io/WebPlotDigitizer/ ) was used to extract data from figures manually.

Methodological quality appraisal
The risk of bias in the included studies was independently assessed by two authors (GF and MDP), and any disagreement was resolved by a third author (VO).The Newcastle-Ottawa Scale (NOS) ( Stang, 2010 ) was adopted to grade studies' quality.

Statistical analyses
Analyses were performed using RStudio R version 4.1.2( RStudio Team, 2020 ) and the metafor R-package ( Viechtbauer 2010 ) using a random-effect model (restricted maximum-likelihood estimator) ( Harville, 1977 ).Effect sizes were calculated as odds ratios (ORs) with their confidence intervals (CIs) when parental age was considered as a categorical measure, and as standardized mean differences (SMDs) with their CI when parental age was considered as a continuous measure.Effect sizes adjusted for the age of the other parent and socioeconomic status -measured as paternal occupation, ethnicity, education, or income ( Goldberg et al., 2011 ) -were used when available.
Heterogeneity between studies was assessed by χ 2 test of fit (Cochrane Q test) and I 2 statistic.A χ 2 statistic having p < 0.05 and I 2 statistic > 50% were considered suggestive of heterogeneity ( Higgins et al., 2003 ).
Sensitivity analyses were conducted by removing one study at time from the analysis; cumulative analyses were performed to evaluate the repercussions of the follow-up length and the year of enrollment on the effect size.Publication bias was explored by visual inspection of funnel plots and using the Egger's test .If the latter was significant, the trim-and-fill procedure was adopted to estimate the impact of the publication bias on the results.

Study selection
A flow diagram showing the results of the literature search and selection of studies is presented in Fig. 1 .The literature search involving the electronic databases MEDLINE, Psych-Info, Embase, and Scopus identified a total of 714 records.Of these, 634 articles, including duplicates not fulfilling eligibility criteria, were excluded.Thus, 80 full-text articles were further evaluated for eligibility; of these, 64 were excluded for: (1) not including patients with BD, (2) not reporting data on parental age, (3) not presenting original data, (4) being congress abstracts.A final number of 16 articles were included for their qualitative synthesis, and of these, 14 were included in the quantitative syntheses.

Outcomes
Most of the studies included in the meta-analyses reported effect size as adjusted OR.The minority of included stud-ies reported adjusted relative risk (RR), incidence rate ratio (IRR), or hazard ratio (HR).As the proportion of outcome was reported as rare (i.e., < 0.07%) in those studies, OR approximated well to other effect sizes, thus we assumed these measures as equal to OR ( Schmidt and Kohlmann, 2008 ;Zhang and Yu, 1998 ).

Meta-analyses of studies
As our main aim was to conduct a meta-analysis using data adjusted for the confounding variables, and since the reference category varied among studies, we conducted for each parental sex a first meta-analysis using the 25-29 age group as the reference category and only adjusted effect sizes, and a second one using the 20-24 age group as reference category after calculating ORs from raw data if not already available as adjusted OR in the study.

Sensitivity analyses
In sensitivity analyses of paternal age considered as a categorical variable (adjusted data, reference age 25-29) and risk of BD, the direction of the combined estimates did not significantly vary with the removal of each study in turn, except in the age category ≤ 20 in which, by removing McGrath et al. (2014 ), the overall effect become nonsignificant and in the age category 40-44, in which by removing Weiser et al. (2020), the overall effect become significant (OR = 1.27, 95% CI: 1.12-1.44).( Supplementary   (2014 ), the overall effect become non-significant (Supplementary 6).

Cumulative analyses
Cumulative analyses of the effect of the duration of the observation on paternal or maternal ages (both age of reference 20-24 and 25-29) and the risk of BD in the offspring showed no statistically significant changes in the effect sizes as the duration of the follow-up increased (data available on request).Cumulative analyses based on the year of sample enrollment across different studies showed no statistically significant changes in the effect sizes in almost every agerange considered, with exception to the highest paternal category ( ≥ 45), which effect size became non-significant (OR = 1.16, 95% CI: 0.97-1.38)as the year of enrollment was more recent (data available on request).

Publication bias
In the analyses of paternal age group 20-24 and the risk of BD (reference age 25-29), the Egger's tests and funnel plots were suggestive of publication bias (Egger's t = -2.76,p = 0.0057), with two possible missing studies estimated with the trim and fill method.In the analyses of paternal age group ≥45 and the risk of BD (reference age 20-24), the Egger's tests and funnel plots were suggestive of publication bias (Egger's t = 2.68, p = 0.007), with three possible missing studies estimated with the trim and fill method.In the analyses of maternal age groups 25-29, 30-34 and the risk of BD (reference age 20-24), the Egger's tests and funnel plots were suggestive of publication bias (Egger's t = -2.85,p = 0.004, and t = -3.30,p = 0.0009), with one possible missing study in each test estimated with the trim and fill method.All other analyses on publication bias using Egger's test and inspection of funnel plots were not suggestive of publication bias (data available on request).

Parental age effect on the age of onset of BD in the offspring
Only two studies included in the narrative synthesis reported the effect of parental age on the age of onset of BD ( Frans et al., 2008 a;Grigoroiu-Serbanescu et al., 2012 ).Advanced paternal age was associated with earlier age of BD onset in the first study ( Frans et al., 2008 a) (OR = 2.63; 95% CI, 1.19-5.81)and in the second study ( Grigoroiu-Serbanescu et al., 2012 ), showing a negative association between increasing parental ages and decreasing proband age of BD onset (OR = 0.54, 95% CI, 0.35-0.80).Maternal age had no significant effect on BD age of onset in both studies (see Table 1 ).

Quality of the included studies
Overall, among case-control studies, 5 had "good" quality and 2 were of "poor" quality overall, based on the Newcastle-Ottawa Scale (Supplementary 6).Overall, among cohort studies, 7 had "good" quality and 2 were of "poor" quality overall, based on the Newcastle-Ottawa Scale (Supplementary 7).

Main findings
The present study aimed at quantifying, through a metaanalytical approach, whether parental age is associated with an increased risk of BD in the offspring and whether advanced paternal age is associated with an early-onset of BD.
The main results of the quantitative analysis were: (i) younger maternal and paternal ( ≤ 20 years) or advanced age ( ≥ 35 or ≥ 45 years, respectively) were associated with an increased risk of BD in offspring when the reference age group was 25-29 years and data were adjusted for confounding variables; (ii) when the reference age group was 20-24 years and the data were not adjusted for confounding variables, only advanced paternal age ( ≥ 45 years) was associated with an increased risk of BD in offspring; (iii) no significant increase in the risk of BD in the offspring was found when parental age was taken as means.
Our results are in line with the ones of another recent meta-analysis on the topic ( Polga et al., 2022 ), in which advanced paternal and maternal age were associated with a higher risk of BD in the offspring.However, this previous quantitative analysis included a total of 7 studies and did not consider two large prospective cohort studies ( Liang et al., 2021 ;McGrath et al., 2014 ), which recruited over 6 million people and were included in the present meta-analysis.Furthermore, the authors adopted age categories of 10 years, possibly leaving residual confounding within categories ( Reijneveld, 2003 ), and used unadjusted effect sizes.Since the age of mothers and the age of fathers are highly correlated, studies looking only at paternal age may show stronger effects, although the risks are attributable to maternal age ( Reijneveld, 2003 ).Indeed, nonlinear models examining the joint influence of paternal or maternal age and controlling for possible confounders showed a better fit than traditional models ( Thompson, 2019 ).Our results must therefore be considered in this light, with the adjusted results for confounding variables being more reliable than the unadjusted ones.

Potential mechanisms
The U-shaped association between BD risk in offspring and parental age showed in our results is not surprising, since the previous meta-analysis on schizophrenia or other neuropsychiatric disorders also reported U-shaped risk patterns ( Miller et al., 2011 ;Oldereid et al., 2018 ;Wu et al., 2017 ).
There are several plausible underlying mechanisms by which advanced parental age might increase the risk of BD in the offspring, that might derive from a complex interplay between genetic and psychosocial factors.Interestingly, a recent study investigating the association between polygenic risk score for BD in parents and offspring and the risk of lifetime diagnosis of BD in offspring found that the role of genetic risk of the offspring is at least partially independent from clinical parental diagnosis of BD, explaining a big proportion of variance (5%) ( Birmaher et al., 2022 ).
Our study is the first one that draws particular attention to the link between both younger paternal and maternal age and BD since previous meta-analyses on the matter failed to report this association as significant ( Polga et al., 2022 ).Younger parenthood has been associated with lower educational achievements, adverse social outcomes, violent or criminal behavior both in parents and offspring ( Mills et al., 2011 ).Indeed, adolescent parents may experience more psychosocial stress, stigmatization, anxiety and use fewer coping strategies ( Kaye, 2008 ).Stressful life events might also contribute to the increase of the risk of any mood disorders ( Hillegers et al., 2004 ;Kessing et al., 2003 ).Increased rates of substance abuse including alcohol, cocaine, marijuana, and methamphetamines, are more common in younger parents ( Carroll Chapman and Wu, 2013 ;Daldegan-Bueno et al., 2021 ).Also, adolescents might be more prone to engage in non-protected sex, eventually resulting in unplanned pregnancies and increased rates of abortion ( Davis and Beasley, 2009 ).Therefore, evidence highlighted a positive relationship between induced abortion, spontaneous abortion, and an increased risk of psychiatric disorders ( Jacob et al., 2019 ).Younger parents with a tendency toward impulsive behavior, mood fluctuations, risky behaviors ( Axelson et al., 2011 ;Birmaher et al., 2018 ) or specific affective temperaments ( Fico et al., 2019 ) might also be individuals with a prodrome affective disorder, thus, with a higher genetic load and higher chances to progress to full-threshold BD ( Birmaher et al., 2018 ).However, the recognized role of environmental factors as cues for BD risk should not overshadow the possible biological mechanism involved.Indeed, a stressful environment promotes inflammation ( Herbert and Cohen, 1993 ); therefore, inflammation might be a mediator in the relationship between psychosocial stress and offspring neuropsychiatric outcomes ( Csatlosova et al., 2021 ;Hantsoo et al., 2019 ;Khalfallah et al., 2022 ).Furthermore, stress exposure may accelerate telomere shortening as observed in several psychiatric disorders, including BD. Telomere length is heritable and correlates with paternal age ( Njajou et al., 2007 ), but its role in increasing BD risk is unknown ( Slagboom et al., 1994 ).Still, a stressful environment rather than vertical transmission may have a stronger influence on telomere shortening in BD ( Powell et al., 2018 ;Shalev et al., 2013 ;Valdes et al., 2005 ).Indeed, sperm telomere length increases with age in humans, and as a result offspring of older fathers inherit longer telomeres ( Eisenberg and Kuzawa, 2018 ).
Substance use is associated with depression in the offspring in pre-clinical studies ( Pacheco et al., 2021 ), and with birth defects and increasing de novo mutations in germlines in clinical studies ( Laubenthal et al., 2012 ;Linschooten et al., 2013 ;Shi et al., 2001 ), which may ultimately result in an increased risk for BD in the offspring of younger parents.Nonetheless, most de novo mutations occur in paternal germlines, while aneuploidies are mainly of maternal origin and are associated with increased maternal age ( Larroya et al., 2021 ), but doubtfully associated with an increased paternal age ( Steiner et al., 2014 ).
As regards advanced paternal age and the risk of BD, our results replicated previous evidence ( Polga et al., 2022 ).The mechanism underlying the advanced parental age effect has been mainly related to increased rates of de novo mutations ( Crow, 2000 ;Kong et al., 2012 ), epigenetic alternations ( Denomme et al., 2020 ;Malaspina, 2001 a), and the personality of older fathers ( Zammit et al., 2003 ).More likely, a multitude of interacting factors seems to mediate the impact of advanced paternal age on BD risk, including the role of aging, de novo mutations, epigenetic mechanisms, psychosocial environment, and selection into late fatherhood ( Vervoort et al., 2022 ).
For the mechanisms that underlie advanced maternal age effects, evidence differs between BD and other neuropsychiatric disorders.It is well known that advanced maternal age ( Cavazos-Rehg et al., 2015 )or a mother's history of a severe psychiatric disorder ( Solé et al., 2020 ) are associated with higher rates of obstetric complications that might increase the risk of neurodevelopmental or psychiatric disorders in the offspring ( Zhang et al., 2019 ).However, while for disorders such as schizophrenia and autism spectrum disorder perinatal the association is replicated in several studies ( Abel et al., 2013 ;Buchmayer et al., 2009 ;Davies et al., 2020 ), no significant association was found for BD, although evidence is still scarce ( Serati et al., 2020 ).Maternal immune activation at the time of gestation, not limited to infections but considering a spectrum which includes stress or traumatic experiences, epigenetic modification of stressrelated pathways or changes in the microbiota, has been linked to increased risk of schizophrenia or BD ( Brown and Conway, 2019 ).
A secondary aim of our study was to assess whether parental age is associated with earlier BD onset, based on the hypothesis that advanced parental age increased the risk of early-onset BD in offspring.BD showed a bimodal (early or late) or trimodal (early, mid-and late) age-ofonset distribution, probably due to a combined effect of environmental and genetic factors ( Bolton et al., 2021 ;Kessing, 2006 ;Preisig et al., 2016 ;Vedel Kessing et al., 2021 ), but to date, the paternal age effect on BD onset is still to be explored.We found only two studies exploring this association, not includable in quantitative analysis.Overall, both studies indicated that advanced paternal, but no maternal, age is associated with earlier age of BD onset.This result is in line with the hypothesis that advanced age is associated with an accumulation of DNA mutations, especially in paternal germ cells ( Taylor et al., 2019 ).

Clinical and preventative implications
The combination of estimates from all previously published studies allowed a more complete and precise assessment of the parental as a risk factor for BD, supporting the hypothesis that both very young and old parental age is associated with an increased offspring's risk for BD.In older parents, the effect on BD risk might have a stronger genetic or epigenetic influence, while in younger fathers or mothers the effect might result from a complex interaction between psychosocial environment, activation of stress-immune related pathways, and epigenetic and genetic factors.
For both clinical practice and public health programs, our results underline the importance of applying prevention strategies to the patient population diagnosed with BD to minimize the risk of BD in the offspring, and to offspring that would be identified as a BD risk population, along with other known environmental risk factors ( Marangoni et al., 2016 ).However, due to the limitations of our study, such considerations should be taken into account cautiously before being translated into clinical practice.( Laurenzi et al., 2020 )The progressively increased age of parenthood worldwide ( Waldenström, 2016 ;Eurostat, 2020) is a public health concern, and future studies that demonstrate paternal age correlates with specific psychiatric disorders may guide individual and societal interventions.Lastly, although our data were controlled for the presence of parental history of psychiatric disorders, it should be noted that offspring of parents with BD have an increased risk of developing any mood disorder ( Mesman et al., 2013 ), and specific preventive strategies in this at-risk population should be also improved.

Strengths and limitations
Our study comes with some limitations.To begin, still few studies investigated the association between parental age and BD risk in offspring.Then, only observational studies were included in the analysis.Although there was moderate-to-high heterogeneity in some of the analyses, it was mostly limited to analysis with unadjusted data.Additional information about non-psychotropic drugs, the course of pregnancy, the history of BD in first-degree relatives, the history of other mood disorders in the parents, especially with younger age and possible not-yet diagnosed with BD or other mood disorders, deserves additional primary studies to allow for control of such moderators.Although most studies presented adjusted data for socioeconomic status, they considered it as a single socioeconomic variable rather than a complex and multifactorial one ( Braveman et al., 2005 ).Also, although the majority of the pooled effect sizes in our quantitative analyses have a narrow confidence interval demonstrating a greater degree of precision, these effect sizes are quite small, calling for caution.Finally, in the included studies the authors used a range of parental age as a reference group, being able to assess the risk of BD in the offspring only in lower or higher parental age categories compared to the reference points.However, several strengths should be addressed: we included a larger number of the studies on the topic confirming but, most importantly, significantly extending previous evidence; the majority of the included studies followed a longitudinal design, with long follow-up periods; the pooled effect sizes of risk for BD were largely consistent in several different sensitivity analyses; there was no publication bias in the majority of the analyses, so the probability that our findings are a result of selective publication seems to be minimal.

Conclusion
In conclusion, there is evidence that young and advanced parental age is associated with an increased risk of BD in the offspring.Recognition of this risk is of great importance in both clinical psychiatric practice and as a public health issue.It is necessary to inform individuals on the risk of early or delayed reproductive age and to implement interventions to reduce psychosocial stressors in adolescent parents, as well as programs to follow-up offspring at risk of developing BD.Further studies are needed to elucidate the cause-effect relationship and mechanism between parental age and increased risk of BD in the offspring.

Role of the funding source
The author(s) received no specific funding for this work.

Contributors
GF designed the study and wrote the protocol, managed the literature searches and analyses and wrote the first draft of the manuscript.Authors GF, MDP and VO undertook the statistical analysis.All authors contributed to and have approved the final manuscript.

Declaration of Competing Interest
EV has received grants and served as consultant, advisor, or CME speaker unrelated to the present work for the following entities: AB-Biotics, Abbott, Allergan, Angelini, Dainippon Sumitomo Pharma, Ferrer, Gedeon Richter, Janssen, Lundbeck, Otsuka, Sage, Sanofi-Aventis, Sunovion, and Takeda.GF has received CME-related honoraria, or consulting fees from Angelini, Janssen-Cilag and Lundbeck.MSV has received financial support for CME activities or travel funds from Janssen-Cilag and Lundbeck, and has served as a speaker for Casen Recordati.She reports no financial or other relationship relevant to the subject of this article.AM has received funding unrelated to the present work for research projects and/or honoraria as a consultant or speaker from the following entities: Angelini, Janssen, Lundbeck, Otsuka, Sanofi-Aventis and Spanish Ministry of Science and Innovation-Instituto de Salud Carlos III.SGC has received CME-related honoraria, or consulting fees from Janssen-Cilag, Italfarmaco, Angelini and Lundbeck and reports no financial or other relationship relevant to the subject of this article.AS is or has been a consultant/speaker for Abbott, Abbvie, Angelini, AstraZeneca, Clinical Data, Boehringer, Bristol-Myers Squibb, Eli Lilly, GlaxoSmithKline, Innovapharma, Italfarmaco, Janssen, Lundbeck, Naurex, Pfizer, Polifarma, Sanofi, and Servier.
EV thanks the support of the Spanish Ministry of Science and Innovation (PI15/00283, PI18/00805) integrated into the Plan Nacional de I + D + I and co-financed by the ISCIII-Subdirección General de Evaluación and the Fondo Europeo de Desarrollo Regional (FEDER); the Instituto de Salud Carlos III; the CIBER of Mental Health (CIBERSAM); the Secretaria d'Universitats i Recerca del Departament d'Economia i Coneixement (2017 SGR 1365), the CERCA Programme, and the Departament de Salut de la Generalitat de Catalunya for the PERIS Grant SLT006/17/00357.AM has received a grant (PI19/00672) from the Instituto de Salud Carlos III Subdirección General de Evaluación y Fomento de la investigación, Plan Nacional 2019-2022.The project that gave rise to these results received the support of a fellowship from "La Caixa" Foundation (ID 100010434).GF fellowship code is LCF/BQ/DR21/11880019. Thanks the support of the European Union Horizon 2020 research and innovation program (EU.3.1.1.Understanding health, wellbeing and disease: Grant No 754907 and EU.3.1.3.Treating and managing disease: Grant No 945151).

Fig. 1
Fig. 1 PRISMA flowchart of systematic review search process.

Fig. 2
Fig. 2 Pooled odds ratios (OR) with 95% CI for risk of BD by adjusted maternal and paternal age, with age of reference 25-29 years.

Fig. 3
Fig. 3 Pooled odds ratios (OR) with 95% CI for risk of BD by unadjusted paternal age, with age of reference 20-24 years.

Table 1
Demographic, clinical characteristics, and main results of the included studies.

Table 1 (
continued ) ( continued on next page )
category of parental age was chosen by considering the one appraised by most of the included studies.For studies that had other over-age categories (i.e., 45-49, > 50), the highest category of parental age was chosen by considering the one appraised by most of the included studies 45 years.Maternal age was categorized into the following age groups: < 20, 20-24, 25-29, 30-34, 35-39, 40.The high-est 6).In sensitivity analyses of maternal age considered as a categorical variable (adjusted data, reference age 25-29) and risk of BD in the age category 35-39, by removing McGrath et al.