There is robust evidence for an association between early pubertal timing in girls and depressive symptoms; Reference Galvao, Silva, Zimmermann, Souza, Martins and Pereira1–Reference Ullsperger and Nikolas3 however, there continues to be substantial variation in the measurement of pubertal timing in the literature. Reference Hayward4,Reference Negriff and Susman5 As puberty is a developmental process and not a fixed event, its measurement can pose a methodological challenge. While age at menarche (AAM) is the most frequently used measure of pubertal timing, self-reported indicators such as the Tanner stages of development are widely used. Reference Marshall and Tanner6,Reference Tanner7 These measures are subject to potential misreporting and can be influenced by embarrassment or cultural limitations, particularly when illustrations that depict only Caucasian adolescents are used. Reference Dorn and Biro8 Despite these limitations, self-reported indicators show good agreement with objective measures of pubertal timing, such as age at peak height velocity (aPHV), age at peak weight velocity (aPWV) and age at peak bone mineral content volume (aPBMCV) velocity, which have been used to explore associations between puberty and mental health. Reference Doneray and Orbak9–Reference McCormack, Cousminer, Chesi, Mitchell, Roy and Kalkwarf12 These measures can be derived from longitudinal anthropometric data using parametric growth curve modes, such as Superimposition by Translation and Rotation (SITAR) analyses, which estimate the peak velocity of change during the pubertal growth spurt. Reference Cole, Pan and Butler13,Reference Cole, Donaldson and Ben-Shlomo14 Growth curve analysis can be applied to evaluate timing of secondary sexual characteristics, such as Tanner stages and axillary hair. Reference Cole, Pan and Butler13 This study, based on data from a large UK birth cohort, compares associations between a range of pubertal timing indicators – both objective and self-reported – with depressive symptoms and depression, and explores whether these associations persist into early adulthood. The primary analysis explores depressive symptoms in mid-adolescence, where there is a high degree of inter-individual variation in pubertal maturation and depressive symptoms/depression diagnosis in late adolescence/early adulthood, where most girls will have gone through the pubertal transition.
Method
Participants
The Avon Longitudinal Study of Parents and Children (ALSPAC) is a birth cohort that originally recruited pregnant women (n = 14 541) residing in Avon, UK, with expected dates of delivery from 1 April 1991 to 31 December 1992. Of the initial pregnancies, there were 14 676 foetuses, of which 14 062 were live births and 13,988 were alive at 1 year of age. When the oldest children were approximately 7 years of age, an attempt was made to increase the original sample by recruiting eligible individuals who did not join the study. This resulted in a total sample size of 15 454 pregnancies (15 658 foetuses, 14 901 alive at 1 year of age) when using data after the age of seven. Because of confidentiality reasons, data on 13 triplets/quads is not provided resulting in 15 645 cases. Reference Boyd, Golding, Macleod, Lawlor, Fraser and Henderson15–Reference Northstone, Lewcock, Groom, Boyd, Macleod and Timpson17 Of these, 50.8% were female (assigned gender at birth), resulting in a sample of 7961 (for participant flowchart, see Supplementary Material S1).
The authors assert that all procedures contributing to this work comply with the ethical standards of the relevant national and institutional committees on human experimentation and with the Helsinki Declaration of 1975, as revised in 2013. All procedures involving human participants/patients were approved by ALSPAC Law and Ethics Committee and local research ethics committees. Informed consent for the use of data collected via questionnaires and clinics was obtained from participants following the recommendations of the ALSPAC Ethics and Law Committee at the time (http://www.bristol.ac.uk/alspac/researchers/research-ethics/). Study data were collected and managed using REDCap electronic data capture tools hosted at the University of Bristol. Reference Harris, Taylor, Thielke, Payne, Gonzalez and Conde18 REDCap (Research Electronic Data Capture) is a secure, web-based software platform designed to support data capture for research studies. Please note that the study website contains details of all the data that is available through a fully searchable data dictionary and variable search tool (http://www.bristol.ac.uk/alspac/researchers/our-data/).
Measures
Pubertal timing measures
Data on seven indicators of pubertal development were collected from 7 to 17 years. These included aPHV, aPWV, aPBMCV, age at Tanner pubic hair stage 3, age at Tanner breast development stage 3, age at axillary hair (‘started growing hair’) and AAM. At research clinics, data on height and weight were collected by trained fieldworkers and data on bone mineral content (BMC) (total body less head) were collected using dual-energy X-ray absorptiometry (DXA) scans. Self-reported Tanner stages, menarche and axillary hair were collected in questionnaires. aPHV, aPWV, aPBMCV, age at Tanner pubic hair stage 3, age at Tanner breast development stage 3 and age at axillary hair were derived using SITAR analysis, Reference Cole, Donaldson and Ben-Shlomo14 described in detail elsewhere. Reference Elhakeem, Frysz, Goncalves Soares, Bell, Cole and Heron10 SITAR is a nonlinear mixed effects model that fits a mean growth curve to a study sample and uses random effects to describe individual variations in size, timing and intensity. This model is particularly well suited for capturing heterogeneity in changes that typically progress monotonically but follow a nonlinear pattern, such as growth or pubertal development. AAM was obtained from the first reported age at onset of menstruation from clinic reports and postal questionnaires. For distribution of pubertal timing measures, see Supplementary Material S2.
Depressive symptoms
Depressive symptoms were measured at 13.84 (s.d. = 0.21) and 17.84 (s.d. = 0.40) years old, hereafter referred to as 14 and 18 years, respectively, using the Short Moods and Feeling Questionnaire (SMFQ), Reference Angold, Costello, Messer, Pickles, Winder and Silver19 a widely used and validated tool for depressive symptoms in adolescents. Reference Turner, Joinson, Peters, Wiles and Lewis20 The SMFQ is a 13-item questionnaire measuring the occurrence of depressive symptoms over the past 2 weeks. The total scores on the questionnaire range from 0 to 26, with higher scores indicating greater depressive symptoms. Because of the positively skewed distribution of the SMFQ responses, we used SMFQ scores of 11 and greater to indicate high levels of depressive symptoms, because this threshold has previously been found to have good specificity for predicting clinical depression. Reference Turner, Joinson, Peters, Wiles and Lewis20,Reference Kwong21
Depression
Depression was measured at 17.82 (s.d. = 0.46) years old, hereafter referred to as 18 years, using a computerised version of the Revised Clinical Interview Schedule (CIS-R). Reference Lewis, Pelosi, Araya and Dunn22 The CIS-R is a widely used and validated self-report questionnaire that assesses depression in community samples. Reference Lewis, Pelosi, Araya and Dunn22 The CIS-R asks about a range of symptoms experienced over the past week that are used to generate a depression diagnosis according to the ICD-10 criteria. 23 The outcome used in this study was any ICD-10 diagnosis of depression (mild, moderate or severe) versus none.
Confounders
Socioeconomic status (SES) was measured by occupational social class (manual versus non-manual) during the antenatal period, home ownership status (renter versus owned/privately rented) assessed when the study child was 1.8 years, maternal educational attainment (Certificate of Secondary Education [CSE]/vocational qualifications/none; O-Levels; A-Levels and above) assessed when the study child was 5.1 years, major financial problems (experienced in the first 5 years of child’s life versus none) and father absence (before age 5, between age 5 and 10 or father present). Body mass index (BMI), measured at age 9 years, was calculated based on height and weight measurements obtained from research clinics (90.2%) and parent-reported height and weight provided in questionnaires at age 9 when clinic data were missing (9.8%). Confounder selection was guided by empirical research on factors related to both pubertal timing and depression. Reference Hoyt, Niu, Pachucki and Chaku2,Reference Roberts, Joinson, Gunnell, Fraser and Mars24
Primary analysis
Our primary aim was to examine associations between the seven pubertal timing indices and depressive symptoms in mid-adolescence and depressive symptoms/depression diagnosis in late adolescence/early adulthood. For this analysis, we used three outcomes – depressive symptoms (SMFQ) data at ages 14 and 18 years and depression diagnosis data (CIS-R) at age 18. We used logistic regression analysis, adjusted for the confounders. In a preliminary analysis, age at response was adjusted to account for slight variations in participants’ age at outcome assessment. This had no meaningful impact on the results and was therefore not included in the main analysis. The potential for nonlinear relationships was examined using quadratic and fractional polynomial regressions and we found no evidence of departure from linearity. All analyses were carried out using Stata 17 for Windows (StataCorp, College Station, TX, USA); see (https://download.stata.com/).
Post hoc analysis
Based on the results from the primary analysis, we conducted post hoc analyses with additional timepoints to examine if associations between the pubertal timing indices and depressive symptoms/depression were consistent over time. For these analyses, we used four additional outcome variables available in the ALSPAC data-set. Depressive symptoms were measured using the SMFQ at 16.67 (s.d. = 0.23) and 23.87 (s.d. = 0.52) years old, hereafter referred to as 17 and 24 years, respectively. Depression was measured using the CIS-R at 24.45 (s.d. = 0.82) years old, hereafter referred to as 24 years. While there were no earlier timepoints for the CIS-R, depression was measured using Development and Well-Being Assessment (DAWBA) Reference Goodman, Ford, Richards, Gatward and Meltzer25 band predictions at 15.49 (s.d. = 0.38) years old, hereafter referred to as 15 years. The DAWBA is designed to generate DSM-IV diagnoses for young people aged 2–17 years old. Reference Goodman, Heiervang, Collishaw and Goodman26 The DAWBA was self-reported via computerised questionnaire by the child at a research clinic, generating the following probability bands: ‘<0.1%’, ‘∼0.5%’, ‘∼15%’, ‘∼50%’, ‘>70%’. We used a probability of 50% or greater to indicate depression, a binary distinction that has been previously validated. Reference Goodman, Heiervang, Collishaw and Goodman26 We used logistic regression analysis, adjusted for the confounders.
Missing data
Of the total cohort, 4607 participants had at least one puberty measure available and were eligible for inclusion in this study. Complete data on the puberty variables, depression outcomes and confounders were available for 1200 participants. Missing data were imputed using multiple imputation by chained equations (MICE). Reference Royston and White27 Fifty data-sets were imputed (25 iterations), with parameter estimated pooled according to Rubin’s rules. Reference Rubin28 The imputation model included all variables in the analyses and relevant auxiliary variables (Supplementary Material S7).
Results
Mean age at puberty varied from 11.8 years for Tanner breast development stage 3 to 12.7 years for AAM (Fig. 1). Descriptive data for all puberty measures, depression outcomes and confounders are shown in Table 1. In Figs. 2 and 3, results from the primary analysis and post hoc analysis are combined to show the pattern of associations at the different timepoints. Fig. 2 shows the adjusted associations between the seven pubertal timing indicators and depressive symptoms (SMFQ ≥ 11) at 14, 17, 18 and 24 years. Fig. 3 shows the adjusted associations between the seven pubertal timing indices and depression at 15, 18 and 24 years. Unadjusted, SES-adjusted and fully adjusted results are presented in full in the Supplementary Material (S3 and S4), alongside the complete case results (S5 and S6).
Timing of pubertal development for each pubertal timing measure in the imputed sample (N = 4607). BMC, bone mineral content.
Association between pubertal timing (1-year increase) and depressive symptoms (Short Moods and Feeling Questionnaire ≥ 11) at 14, 17, 18 and 24 years, adjusted for socioeconomic status and pre-pubertal body mass index, in the imputed sample (N = 4607). BMC, bone mineral content.
Association between pubertal timing (1-year increase) and depression at 15, 18 and 24 years, adjusted for socioeconomic status and pre-pubertal body mass index, in the imputed sample (N = 4607). BMC, bone mineral content.
Descriptive statistics in the imputed sample (up to N = 4607)
BMC, bone mineral content; BMI, body mass index; SMFQ, Short Moods and Feeling Questionnaire; DAWBA, Developmental and Well-Being Assessment; CIS-R, Revised Clinical Interview Schedule; CSE, Certificate of Secondary Education.
a. Depressive symptoms = SMFQ ≥ 11.
Depressive symptoms
Age 14
A 1-year increase in aPHV, aPWV and aPBMCV was associated with a decrease in the odds of depressive symptoms (adjusted odds ratio (AOR): 0.82; 95% CI 0.72–0.95, AOR: 0.90; 95% CI 0.82–1.00 and AOR: 0.85; 95% CI 0.76, 0.94, respectively). Age at Tanner pubic hair and breast stage 3 were associated with reduced odds of depressive symptoms (AOR: 0.89; 95% CI 0.80, 0.98 and AOR: 0.87; 95% CI 0.79, 0.97, respectively). AAM was also associated with reduced odds of depressive symptoms (AOR: 0.84; 95% CI 0.76, 0.92). There was no evidence that age at axillary hair development was associated with depressive symptoms at age 14 (AOR: 0.93; 95% CI 0.82, 1.04).
Age 18
We found little evidence that a later age at puberty was associated with increased odds of depressive symptoms. Contrary to the findings at age 14, a 1-year increase in age at Tanner pubic hair stage 3 was associated with an increased odds of depressive symptoms (AOR: 1.12; 95% CI 1.03, 1.22). In the model adjusting only for SES and not pre-pubertal BMI, a later Tanner pubic hair stage 3 was not associated with depressive symptoms at age 18 (SES-AOR: 1.07; 95% CI 0.98, 1.17). When adjusting only for SES, a 1-year increase in aPWV and Tanner breast stage 3 was associated with a decrease in the odds of depressive symptoms (SES-AOR: 0.91; 95% CI 0.85, 0.99 and odds ratio: 0.91; 95% CI 0.84, 0.98, respectively), but not when adjusting for both SES and pre-pubertal BMI (AOR: 1.00; 95% CI 0.92, 1.09 and AOR: 0.97; 95% CI 0.89, 1.06, respectively). No associations were found with AAM, aPBMCV or age at axillary hair and depressive symptoms at 18.
Depression
Age 18
A later age at Tanner breast stage 3 and a later AAM were associated with decreased odds of depression (AOR: 0.83; 95% 0.72, 0.95 and AOR: 0.85; 95% CI 0.75, 0.97, respectively). No associations were found between the remaining pubertal timing measures and depression at 18 years.
Post hoc analyses
Based on the results of the primary analyses, we conducted post hoc analyses using further timepoints to explore the consistency of associations between pubertal timing and depressive outcomes over time.
Depressive symptoms
At age 17, a 1-year increase in AAM and age at Tanner pubic hair stage 3 was associated with a decrease in the odds of depressive symptoms (AOR: 0.92; 95% CI 0.86, 0.99 and AOR: 0.90; 95% CI 0.82, 0.98, respectively). For the remaining five pubertal timing measures, there was weak evidence of an association with decreased odds of depressive symptoms at age 17, indicating imprecision in the estimate. At age 24, we found no evidence of an association between pubertal timing and depressive symptoms.
Depression
At age 15, there was weak evidence (p = 0.067) for an association between pubic hair status and depression as measured by the DAWBA; a 1-year increase in age at Tanner pubic hair stage 3 was associated with approximately 20% lower odds of depression (AOR: 0.79; 95% CI 0.62, 1.01). There was no evidence of an association between the remaining pubertal timing indices and depression at age 15. At age 24, we found no evidence of an association between pubertal timing and depression, as measured by the CIS-R.
Discussion
Overview of findings
Later pubertal timing was consistently associated with reduced odds of depressive symptoms at age 14 across six of the seven indicators of pubertal timing, including self-reported (Tanner stages, AAM) and objective measures (aPHV, aPWV, aPBMCV). There were no associations with axillary hair. At age 18, later AAM and age at Tanner breast stage 3 were associated with reduced odds of depression (CIS-R); however, in contrast to earlier findings at age 14, a later Tanner pubic hair stage 3 was associated with increased odds of depressive symptoms at age 18. Post hoc analyses were carried out to check for consistency of the results over time. At age 15, there was weak evidence for an association between later Tanner pubic hair stage 3 and lower odds of depression (measured by the DAWBA probability bands). At age 17, a later AAM and later age at Tanner pubic hair stage 3 were associated with reduced odds of depressive symptoms. By age 24, there was no clear evidence of an association between pubertal timing and depressive symptoms or depression. The post hoc analyses confirmed the consistency of these findings with the primary analysis.
The consistent findings in mid-adolescence across both objective and self-reported measures of pubertal timing support the robustness of the association between early pubertal timing and depressive symptoms at this stage. Reference Hoyt, Niu, Pachucki and Chaku2,Reference Ullsperger and Nikolas3,Reference Negriff and Susman5 The repeated anthropometric measures (e.g. aPHV) have several notable strengths: they offer reliable, non-intrusive alternatives to self-reported questionnaires. Participants might find it embarrassing to answer questions on menstruation and breast/pubic hair development, Reference Walker, Smith Inskip, Davies and Baird29 particularly when there are cultural differences in the discussion of these topics. Reference Chandra-Mouli and Patel30 In addition, retrospective reports of menarche may suffer from recall error, with single-item reports recorded years after the event being particularly prone to inaccuracy. Reference Dorn and Biro8 In contrast, longitudinal studies that regularly collect data on menarche are more likely to capture an accurate AAM, reducing the potential for recall error. aPHV shows a strong correlation with AAM, Reference Elhakeem, Frysz, Goncalves Soares, Bell, Cole and Heron10 highlighting its potential as an alternative measure that can be applied to both genders, expanding its applicability. Anthropometric measures provide an opportunity for cohort studies that already collect height and weight data to derive objective indicators of pubertal timing.
Inconsistencies were observed in the findings when we used self-reported age at axillary hair and age at Tanner pubic hair stage 3 as indicators of pubertal timing. Age at axillary hair was the only measure to show no association with depressive symptoms at age 14 and no association with depressive symptoms or depression at any age. In this study, axillary hair data were collected as a dichotomous response (presence versus absence), which may have led to imprecise estimations. It is also possible that the lack of association between age at axillary hair and depressive symptoms may reflect the differing hormonal pathways driving adrenarche (e.g. axillary hair growth) and gonadarche (e.g. breast development). Gonadarche, which is primarily driven by increases in oestradiol, has been consistently linked to depression in females, whereas adrenarche may not be as strongly associated. Reference Lewis, Ioannidis, van Harmelen, Neufeld, Stochl and Lewis31,Reference Stumper and Alloy32 Inconsistencies were also found with Tanner pubic hair stage 3; a later age at Tanner pubic hair stage 3 was associated with reduced odds of depressive symptoms at age 14, but with increased odds of depression at age 18. Post hoc analyses were conducted to assess whether this result was anomalous or consistent over further timepoints. These analyses suggested that the conflicting results at age 18 were more likely caused by measurement error in the self-reported scale rather than a true reversal of the observed association. Concerns of validity when utilising self-reported or parent-reported Tanner stages have been noted by previous studies. Reference Rasmussen, Wohlfahrt-Veje, Tefre de Renzy-Martin, Hagen, Tinggaard and Mouritsen33,Reference Desmangles, Lappe, Lipaczewski and Haynatzki34 In this study, pubic hair development was assessed using line drawings, which could lead to misclassification of Tanner stage, particularly as the reporting shifted from parent/carer to self-reports. Over 75% of the first five puberty questionnaires were completed with the help of a parent or guardian, compared to less than 15% of the final four questionnaires, with this shift occurring at age 14.
Findings in context
Our findings align with the early maturation hypothesis, Reference Brooks-Gunn and Warren35 which suggests that a mismatch between physical development and emotional or cognitive development may increase vulnerability to depressive symptoms in early maturing girls. Early maturation may increase susceptibility to psychological distress through a combination of biological factors (e.g. hormonal effects), psychological factors (e.g. body image, self-efficacy) and social factors (e.g. unwanted attention, association with older peers resulting in risk-taking behaviour). Reference Ullsperger and Nikolas3 The association between later pubertal timing and lower odds of depressive symptoms is strongest in mid-adolescence, with the associations attenuating by early adulthood. This finding aligns with the attenuation hypothesis, suggesting that associations between pubertal timing and adverse mental health outcomes diminish over time, becoming negligible by the fourth decade of life. Reference Copeland, Shanahan, Miller, Costello, Angold and Maughan36,Reference Senia, Donnellan and Neppl37 It is notable that we observed differences in results when comparing depressive symptoms with depression diagnosis. This discrepancy may reflect true differences in the relationship depending on severity of depressive outcomes. Depression diagnoses require meeting a higher symptom threshold, which could attenuate associations observed with depressive symptoms. Alternatively, this discrepancy may be related to statistical power, as depression has a lower prevalence compared to depressive symptoms. In our study, one in four girls reported depressive symptoms (SMFQ ≥ 11) at age 18, compared to one in ten girls who met the criteria for ICD-10 depression (according to the CIS-R). These findings highlight the importance of distinguishing between clinical depression and depressive symptoms in research. Understanding the extent to which pubertal timing is associated with clinical outcomes, as opposed to depressive symptoms alone, is critical for informing clinical practice.
Strengths and limitations
A key strength of our study is the use of multiple measures of pubertal timing, including both subjective and objective indicators. By employing repeated measures of each pubertal timing indicator, we minimised recall error, as the proximity of data collection to the pubertal events themselves reduces the likelihood of inaccurate reporting. Reference Dorn and Biro8 This approach strengthens the validity and reliability of these measures compared to studies relying solely on retrospective recall. The longitudinal design enabled the assessment of associations between pubertal timing and depressive symptoms/depression in mid-adolescence and early adulthood, allowing us to examine if associations persisted over time.
Limitations of this study include the use of different diagnostic tools (DAWBA in childhood, CIS-R in adolescence), which complicates the comparability of the depression outcomes. However, rates of depression are expected to be low in childhood, Reference Kessler, Avenevoli and Merikangas38 and thus a probable DAWBA-derived diagnosis of depression has been widely used and validated in adolescents. Reference Goodman, Heiervang, Collishaw and Goodman26 Another limitation is the reliance on self-reported data for some indicators, which may introduce measurement error. In addition, the underrepresentation of ethnic minorities in the ALSPAC cohort (95% White ethnicity) limits the generalisability of our findings. Missing data is a common challenge in cohort studies, which can lead to selection bias, particularly because excluded individuals tend to be more socioeconomically disadvantaged compared to the original cohort, and there is a higher prevalence of depression among individuals with a lower SES. Reference Freeman, Tyrovolas, Koyanagi, Chatterji, Leonardi and Ayuso-Mateos39 Multiple imputation was used to address potential bias from attrition and missing data under the assumption that data were missing at random (MAR), meaning that, conditional on observed data, the probability of an observation being missing is independent of unobserved data. Whilst the MAR assumption is not directly testable, the inclusion of extensive auxiliary variables strengthens the plausibility of this assumption, minimising potential biases associated with missing data. These auxiliary variables included measures of SES, maternal mental health and earlier indicators of mental health, such as previous measures of depressive symptoms. Multiple imputation has been shown to eliminate bias in MAR data regardless of the proportion of missing data. Reference Madley-Dowd, Hughes, Tilling and Heron40 Finally, we acknowledge that multiple testing increases the risk of Type I error. However, as our analyses examined related exposures (pubertal timing indices) and related outcomes (depressive symptoms and depression) over multiple timepoints, the tests are not fully independent. Traditional methods for correcting p-values, which assume independence, may be overly conservative in this context. We provide exact p-values and 95% confidence intervals for all analyses to provide transparency and allow readers to interpret the strength of evidence for each association. Reference Ranganathan, Pramesh and Buyse41
Implications
Our study provides robust evidence that early pubertal timing is associated with increased odds of depressive symptoms in girls during mid-adolescence, with associations diminishing by early adulthood. Both self-reported and objective measures of pubertal timing can be used as reliable indicators, but our findings advise against using Tanner pubic hair stage 3 and age at axillary hair. Each pubertal timing indicator captures specific aspects of pubertal development, for example, aPHV marks earlier and more visible physical changes, while others, such as AAM, reflect later, developmentally significant milestones. Researchers should carefully consider which pubertal indicators align with their study’s aims, taking into account the reliability, validity and feasibility of each measure, and ensuring that it accurately reflects the developmental processes or psychological outcomes under investigation.
Supplementary material
The supplementary material is available online at https://doi.org/10.1192/bjp.2025.88
Data availability
ALSPAC data access is through a system of managed open access. Information about access to ALSPAC data is given in the ALSPAC data management plan (http://www.bristol.ac.uk/alspac/researchers/data-access/documents/alspac-data-management-plan.pdf). Analysis code is available on request from the lead author.
Acknowledgements
We are extremely grateful to all the families who took part in this study, the midwives for their help in recruiting them and the whole ALSPAC team, which includes interviewers, computer and laboratory technicians, clerical workers, research scientists, volunteers, managers, receptionists and nurses.
Author contributions
D.T., A.F., J.H. and C.J. were involved in development of the project and interpreting results. D.T. wrote the manuscript and carried out the analysis, with support from C.J., J.H. and A.F. A.E. carried out the statistical modelling of the pubertal timing measures. A.F., J.H. and C.J. supervised the project. All authors provided feedback on the analysis and manuscript.
Funding
The UK Medical Research Council and Wellcome (Grant ref: 217065/Z/19/Z) and the University of Bristol provide core support for the ALSPAC. This publication is the work of the authors, and they will serve as guarantors for the contents of this article. A comprehensive list of grant funding is available on the ALSPAC website (http://www.bristol.ac.uk/alspac/external/documents/grant-acknowledgements.pdf). D.T. is supported by a personal scholarship from the Crown Prince International Scholarship Program.
Declaration of interest
None.
Open access
eLetters
No eLetters have been published for this article.