Abstract
Objective:
To study the associations between maternal polycystic ovary syndrome (PCOS) and hirsutism with offspring attention-deficit/hyperactivity disorder (ADHD), anxiety, conduct disorder, and behavioral problems.
Design:
Prospective birth cohort study.
Setting:
Community sample in upstate New York.
Patient(s):
A total of 1915 mother-child dyads.
Intervention(s):
N/A
Main Outcome Measure(s):
Maternal report of offspring ADHD, anxiety, or conduct disorder diagnosis at 7–8 years; emotional symptoms, behavioral problems (including peer relationship, conduct, hyperactivity/inattention), and prosocial problems measured with the Strengths and Difficulties Questionnaire (SDQ) at 7 years.
Results:
Prevalence of PCOS and hirsutism were 12.0% and 3.9%; 84% of women with hirsutism had PCOS. After adjustment for sociodemographic covariates, pre-pregnancy body mass index, and parental history of affective disorders, children born to mothers with PCOS had higher risk of anxiety (aRR 1.62, 95% CI 1.02–2.57) and borderline emotional symptoms (aRR 1.66, 95% CI 1.18–2.33) compared with children born to mothers without PCOS. Associations between maternal PCOS and offspring ADHD were positive but imprecise. Maternal hirsutism was related to higher risk of children’s ADHD (aRR 2.33, 95% CI 1.28, 4.24) and conduct disorder (aRR 2.54; 95% CI 1.18–5.47) and borderline emotional symptoms, peer relationship problems, and conduct problems (aRRs 2.61, 95% CI 1.69–4.05; 1.92, 95% CI 1.16–3.17; and 2.22, 95% CI 1.30–3.79, respectively).
Conclusions:
Maternal PCOS was associated with offspring anxiety while hirsutism was related to other offspring behavioral problems. These findings should be interpreted with caution as replication is needed in prospective cohort studies which assess PCOS and hirsutism diagnoses using medical records.
Keywords: polycystic ovary syndrome, hirsutism, attention-deficit, hyperactivity disorder, anxiety disorder, behavioral problems
Capsule:
Maternal polycystic ovary syndrome is associated with offspring anxiety; maternal hirsutism is related to offspring ADHD, conduct disorder, and other behavioral problems. Despite some diagnostic overlap, offspring behavioral results vary.
INTRODUCTION
Depressive and conduct disorders are major causes of disability adjusted life years in children and adolescents worldwide (1), yet much remains unknown about their etiology. Recent studies that investigate the association between maternal polycystic ovary syndrome (PCOS) and offspring behavioral disorders may offer evidence on the developmental origins of these childhood conditions.
PCOS is a common endocrine condition known for its association with infertility. Diagnostic criteria for PCOS generally includes having at least two of the following: 1) excess androgens, 2) oligo-anovulation, or 3) polycystic ovaries confirmed by ultrasound (2–4). Hirsutism, a condition marked by excess body hair in a male sex pattern, is often used to indicate a subtype of PCOS with excess androgens. Rodent models of a maternal PCOS-like phenotype induced by injections of androgens find an increase in anxiety-like behavior in offspring (5, 6), potentially mediated by alterations to the androgenic, serotonergic, and gamma-Aminobutyric acidergic (GABA) metabolisms in the amygdala (5). Higher levels of circulating inflammatory markers associated with PCOS may also contribute to the behavioral changes seen in animal models (6–8).
Recent epidemiologic studies suggest that maternal PCOS and hirsutism are associated with offspring autism spectrum disorder (ASD) (9–11) and developmental delay (12, 13). Less is known about the associations of maternal PCOS or hirsutism with behavioral problems or mental disorders later in childhood. Using data from the Swedish registry, a population-based case-control study found that offspring of women with PCOS had higher odds of attention-deficit/hyperactivity disorder (ADHD) compared with offspring of women without PCOS (14). This finding was confirmed in a matched cohort study in the United Kingdom; however, the association between maternal PCOS and children’s ADHD diagnosis was attenuated upon controlling for maternal history of mental disorders (10). These findings have not been replicated in a well-characterized prospective cohort, nor has the association between maternal PCOS and other offspring behavioral problems such as anxiety been examined. Further, maternal hirsutism has not been examined as an exposure.
To this end, we examined the relations of maternal self-report of PCOS and hirsutism with children’s behavioral problems and mental disorders at 7 and 8 years old in the Upstate KIDS cohort. As our assessment of these associations is limited by maternal self-report of PCOS and hirsutism rather than clinical assessment, our findings should be interpreted as hypothesis-generating and will require replication in future studies.
MATERIALS AND METHODS
Study setting and participants
The Upstate KIDS study is a population-based birth cohort designed to evaluate the effects of infertility treatment on child health and development (15). At approximately 4 months postpartum, mothers of children born between July 2008 and May 2010 in New York State (excluding New York City) were recruited into the study. All infants conceived using fertility treatments and multiple births were invited to participate. Three singletons not conceived by fertility treatment were sampled for every child conceived using fertility treatment, frequency matched to region of birth. In total, 5,034 mothers and 6,171 children were recruited. We did not observe any significant differences in early childhood development by fertility treatment status (16). In this analysis, children with information on behavioral problems or mental disorders are included. Due to low numbers, triplets and quadruplets were excluded (n=134).
The New York State Department of Health and the University of Albany Institutional Review Board (IRB) approved of the study and served as the IRB designated by the National Institutes of Health for this study under a reliance agreement. Parents provided written informed consent prior to enrollment.
Measurements
Baseline questionnaire
At the time of enrollment, 4,886 mothers (97.1%) completed a baseline questionnaire on their reproductive history, familial health status, and sociodemographic information. This survey included questions on parental age, height, weight, and history of affective disorders (i.e., anxiety, mood, or eating disorder), and maternal race/ethnicity, education, insurance type, marital status, parity, history of gynecological medical conditions, history and type of infertility treatment, and smoking use during pregnancy. With regards to maternal gynecological conditions, mothers were asked if they had ever received a diagnosis of PCOS or excessive body hair (hirsutism). If they had received a diagnosis, they specified if the condition was diagnosed but not treated, treated with medication only, or treated with surgery. In addition, we obtained information from birth certificates on the child’s sex and plurality, as well as maternal pre-pregnancy height and weight.
Follow-up study
When the children were 7 years old, participants were invited to participate in the Upstate KIDS Follow-Up Study. Mothers completed annual questionnaires pertaining to their children’s development. When the children were an average of 7 and 8 years old, mothers reported if their children had ever been diagnosed with attention-deficit disorder (ADD) or ADHD, anxiety, or conduct disorder, and if the child was taking medications for ADD/ADHD or anxiety. In addition, mothers rated their children’s behavior with the Strengths and Difficulties Questionnaire (SDQ) (17) at 7 years and a modified version of the Vanderbilt ADHD Diagnostic Parent Rating Scale (VADPRS) (18) at 8 years.
The use of the SDQ in this population has been detailed previously (19). In brief, the SDQ is a validated and reliable questionnaire (17) designed to measure children’s behavior in the five domains of emotional symptoms, peer relationship problems, conduct problems, hyperactivity/inattention, and prosocial behaviors. The SDQ consists of 25 statements that mothers rate as not true (0 points), somewhat true (1 point), or certainly true (2 points). From the mother’s responses, we obtained domain-specific scores by summing the scores from the five questions pertaining to each behavior domain. To identify children with borderline behavioral problems, we utilized cutoff points recommended by the SDQ developer (emotional symptoms ≥4, peer relationship problems ≥3, conduct problems ≥3, hyperactivity/inattention ≥6, and prosocial behaviors ≤5) (20). In a nationally representative sample in the United States (US), the subscale cutoff points have good specificity (71–91%) and adequate sensitivity (31–85%) (21).
When the children were approximately 8 years old, mothers completed a modified version of the VADPRS, a validated and reliable questionnaire used in the US to measure symptoms related to ADHD (18, 22, 23). The VADPRS consists of 45 statements on children’s behavioral symptoms and eight statements on school performance and interpersonal relationships. Of the 45 behavioral statements, nine correspond to DSM-IV behaviors for predominately inattentive subtype of ADHD and nine to the hyperactive/impulsive subtype.
Mothers rated how frequently their children have exhibited each behavior on a 4-point scale (never, occasionally, often, or very often). Mothers then rated their children’s scholastic and interpersonal performance on a 5-point scale (excellent, above average, average, somewhat of a problem, or problematic). The VADPRS was scored using the National Institute for Children’s Health Quality’s criteria (24). To meet criteria for screening for the predominately inattentive or hyperactive/impulsive subtype scales, children must have scored “somewhat of a problem” or “problematic” on any of the eight performance items and “often” or “very often” on at least six behavioral items. To meet criteria to screen for ADHD, a child must screen positive for both the predominately inattentive and hyperactive/impulsive subtype scales.
Data analysis
Exposures assessment
The primary exposures of interest were maternal report of diagnosis of PCOS or hirsutism. To investigate if treatment status impacted the associations between maternal PCOS and child behavior, we also examined treated PCOS (with medication or surgery) and untreated PCOS as separate exposures.
Outcome assessment
Mothers of 1386 children (966 singletons, 420 twins) completed the SDQ at 7 years old and mothers of 1484 children (1067 singletons, 417 twins) responded to the VADPRS at 8 years; mothers of 1915 children (1333 singletons, 582 twins) reported on their child’s diagnosis of ADHD, anxiety, or conduct disorder or medication status at either time. We first considered maternal reported diagnosis of ADHD, anxiety, or conduct disorder as outcomes. Diagnosis of ADHD or anxiety was defined by reported diagnosis or use of medications prescribed for that condition at either 7 or 8 years. We also considered children who screened positive with the VADPRS or had ADHD diagnosis as screening for ADHD; children with a positive VADPRS screen were not analyzed as a separate group due to small numbers. We chose not to use the hyperactivity-inattention index of the SDQ to determine ADHD risk as the instrument did not reflect DSM criteria by also accounting for children’s performance. Diagnosis of conduct disorder was defined as maternal report of diagnosis. Finally, we considered the subscale borderline behavioral problems defined by the SDQ as behavioral outcomes.
Statistical analysis
Comparisons of participant characteristics by exposure or inclusion status were conducted within a primary cohort of singletons and one randomly selected twin. We first compared participant characteristics according to maternal PCOS and hirsutism status with those of all participants in the primary cohort. Similarly, characteristics of mothers with any annual questionnaire data (n=1651, 33.1%) were compared with those of all mothers at baseline (n=4989).
All singletons and twin pairs were included in multivariable analysis. We estimated risk ratios (RR) and 95% confidence intervals (CI) for dichotomous endpoints using Poisson regression. These analyses were conducted using generalized estimated equations with robust standard errors to account for the correlation among twins. For all multivariable analyses, we imputed missing covariate and exposure information by generating 50 imputed datasets. Among the 1915 children with outcome data at either assessment, mothers of 39 children were missing information on PCOS and hirsutism (2.04%). Covariates with missing data included maternal marital status (3.03%), history of affective disorder (1.78%), smoking during pregnancy (0.16%), body mass index (BMI) (0.05%), and paternal age (5.43%), history of affective disorders (1.78%), and BMI (9.09%). To account for non-response to the follow-up questionnaires, we used inverse probability weighting (IPW). IPW weights were calculated from a multivariable logistic regression model where the outcome was responding to any behavior questions on the 7 year questionnaire for the SDQ (n=1386) or the 7 or 8 year questionnaire for the maternal report of diagnosis or ADHD screening (n=1915). Covariates in the logistic regression model were maternal age, race/ethnicity, education, insurance status, marital status, history of affective disorder, smoking and drinking during pregnancy, BMI, report of gynecological medical conditions; paternal age, history of affective disorder, and BMI; and child sex, plurality, conception using fertility treatment, and parity. Although fertility treatment and parity were included when constructing IPW weights, they were not considered as covariates in the main models as they are likely consequences of PCOS or hirsutism (3) and thus may be mediators on the causal pathway. Adjustment for mediators does not reveal the direct effect of an exposure (e.g., PCOS) on an outcome (e.g., child behavior) and can introduce bias (25, 26).
For the associations of maternal PCOS and hirsutism, we first present an unadjusted model. Model 1 was adjusted for maternal age, race, education, insurance, marital/cohabitation status, and smoking during pregnancy. Model 2 was additionally adjusted for maternal BMI and parental history of affective disorders. Since there is evidence of a bidirectional association between maternal PCOS and BMI or affective disorders (3), we presented these results in a separate model. Finally, Model 3 was additionally adjusted for child’s sex to increase statistical efficiency. In supplemental analyses, we examined the relations of PCOS with child behavioral problems separately among women with PCOS who report hirsutism and women with PCOS who do not report hirsutism.
Since maternal PCOS was self-reported, we conducted a sensitivity analysis to assess the robustness of our findings to misclassification. Four different scenarios were simulated, two representing a lower prevalence of PCOS, i.e., overreporting, (7 and 9.5%) and two representing a higher prevalence (14.5 and 17%). For each scenario, we generated 1000 datasets, then calculated the unadjusted and fully adjusted RR in each dataset. We then compared the association based on the observed prevalence in the study with the estimated associations of the 1000 simulated samples for each of the four misclassification errors.
All analysis was conducted with SAS version 9.4 (SAS Institute Inc.).
RESULTS
Prevalence of PCOS and hirsutism was 12.0% and 3.9%, respectively. Twenty-seven percent of women with PCOS had hirsutism and 84% of women with hirsutism had PCOS. Maternal socioeconomic status (SES) markers, and parental weight status were positively associated with PCOS and hirsutism. In addition, maternal history of affective disorders was positively related to hirsutism (Table 1). Compared with all mothers at baseline, mothers who responded to the questionnaires when their child was 7–8 years old were older, had higher SES, were more likely to report a diagnosis of PCOS (Supplemental Table 1).
Table 1.
Participant characteristics according to maternal PCOSa and hirsutism status among all singletons and one randomly selected twin of each pair
| Sociodemographic characteristics | All | PCOS | Hirsutism |
|---|---|---|---|
| Overall, n (%) | 1651 (100) | 194 (11.8) | 63 (3.8) |
| Maternal Characteristics | |||
| Age, year, mean ± SD | 31.3 ± 5.9 | 31.5 ± 4.7 | 32.2 ± 4.5 |
| Non-Hispanic white, n (%) | 1412 (85.5) | 168 (86.6) | 57 (90.5) |
| Education, n (%) | |||
| Less than high school | 53 (3.2) | 1 (0.5) | 0 (0.0) |
| High school or GED equivalent | 158 (9.6) | 11 (5.7) | 4 (6.3) |
| Some college | 421 (25.5) | 60 (30.9) | 21 (33.3) |
| College | 414 (25.1) | 55 (28.4) | 18 (28.6) |
| Advanced degree | 605 (36.6) | 67 (34.5) | 20 (31.7) |
| Private insurance, n (%) | 1354 (82.0) | 175 (90.2) | 56 (88.9) |
| Married or living as married, n (%) | 1468 (91.4) | 184 (95.3) | 59 (93.7) |
| History of affective disorder, n (%) | 288 (17.7) | 43 (22.2) | 19 (30.2) |
| Smoking during pregnancy, n (%) | 162 (9.8) | 14 (7.2) | 3 (4.8) |
| BMI, kg/m2, mean ± SD | 26.9 ± 7.0 | 30.4 ± 8.2 | 32.2 ± 8.7 |
| Paternal characteristics | |||
| Age, year, mean ± SD | 33.6 ± 6.7 | 33.4 ± 5.5 | 34.1 ± 5.4 |
| History of affective disorder, n (%) | 127 (7.8) | 16 (8.2) | 6 (9.5) |
| BMI, kg/m2, mean ± SD | 28.3 ± 5.5 | 30.4 ± 6.1 | 31.5 ± 6.5 |
| Child characteristics | |||
| Child sex, male, n (%) | 876 (53.1) | 100 (51.5) | 29 (46.0) |
| Maternal reported diagnosis | |||
| ADHD, n (%) | 148 (9.1) | 22 (11.7) | 11 (19.0) |
| Anxiety, n (%) | 125 (7.7) | 23 (12.2) | 7 (12.1) |
| Conduct disorder, n (%) | 99 (6.1) | 15 (8.0) | 8 (13.8) |
| Maternal reported diagnosis or screen on VADPRS | |||
| ADHD, n (%) | 164 (10.1) | 24 (12.8) | 11 (19.0) |
| Borderline behavioral problems, SDQ | |||
| Emotional symptoms, n (%) | 167 (14.1) | 31 (21.2) | 17 (37.8) |
| Peer relationship problems, n (%) | 185 (15.7) | 25 (17.1) | 12 (26.7) |
| Conduct problems, n (%) | 172 (14.6) | 21 (14.4) | 12 (26.7) |
| Hyperactivity/inattention, n (%) | 188 (15.9) | 29 (19.9) | 11 (24.4) |
| Prosocial problems, n (%) | 73 (6.2) | 12 (8.2) | 3 (6.7) |
ADHD, attention-deficit/hyperactivity disorder; BMI, body mass index; PCOS, polycystic ovary syndrome; SDQ, Strengths and Difficulties Questionnaire; VADPRS, Vanderbilt ADHD Diagnostic Parent Rating Scale
Maternal PCOS.
In unadjusted analysis, maternal PCOS was positively associated with anxiety disorder diagnosis and borderline emotional symptoms in children (Table 2). Associations were strengthened after adjustment for SES markers, then attenuated upon adjustment for parental history of affective disorders and maternal BMI. After full adjustment (Model 3), children born to mothers with PCOS had higher risk of anxiety disorder diagnosis (adjusted risk ratio [aRR] 1.62, 95% CI 1.02–2.57) and borderline emotional symptoms (1.66, 1.18–2.33) compared with children born to mothers with no reported PCOS. In a sensitivity analysis, the association between maternal PCOS and child anxiety disorder diagnosis was relatively robust to misclassification (Supplemental Figure 1). The relations between PCOS and offspring anxiety disorder diagnosis were similar between women reporting PCOS with and without hirsutism. However, the associations between PCOS and ADHD were stronger among women with PCOS and hirsutism (Supplemental Tables 2 and 3). Associations of maternal PCOS with offspring ADHD were positive but not statistically significant.
Table 2.
Childhood behavioral problems and disorders according to maternal polycystic ovary syndrome (PCOS) status in the Upstate KIDS Study
| n | n, cases | Reference | Unadjusted RR (95% CI)a | Model 1 RR (95% CI)b | Model 2 RR (95% CI) | Model 3 RR (95% CI) | |
|---|---|---|---|---|---|---|---|
| Maternal reported diagnosis | |||||||
| ADHDc | 1906 | 173 | 1 | 1.25 (0.77, 2.04) | 1.56 (1.00, 2.44) | 1.33 (0.84, 2.09) | 1.34 (0.86, 2.09) |
| Anxiety | 1909 | 139 | 1 | 1.67 (1.05, 2.64) | 1.87 (1.18, 2.97) | 1.63 (1.02, 2.59) | 1.62 (1.02, 2.57) |
| Conduct disorder | 1905 | 116 | 1 | 1.29 (0.71, 2.35) | 1.90 (1.10, 3.30) | 1.59 (0.92, 2.75) | 1.64 (0.96, 2.79) |
| Maternal reported diagnosis or screen on VADPRS | |||||||
| ADHD | 1915 | 193 | 1 | 1.18 (0.75, 1.86) | 1.52 (1.00, 2.32) | 1.31 (0.86, 2.01) | 1.33 (0.87, 2.03) |
| Borderline behavioral problems, SDQ | |||||||
| Emotional symptoms | 1386 | 193 | 1 | 1.61 (1.14, 2.28) | 1.89 (1.33, 2.68) | 1.68 (1.19, 2.36) | 1.66 (1.18, 2.33) |
| Peer relationship problems | 1386 | 211 | 1 | 0.95 (0.63, 1.44) | 1.15 (0.76, 1.72) | 1.08 (0.72, 1.62) | 1.08 (0.72, 1.62) |
| Conduct problems | 1386 | 201 | 1 | 0.98 (0.66, 1.47) | 1.17 (0.79, 1.74) | 1.11 (0.74, 1.66) | 1.10 (0.73, 1.66) |
| Hyperactivity/inattention | 1386 | 214 | 1 | 1.23 (0.85, 1.76) | 1.40 (0.98, 2.02) | 1.24 (0.85, 1.82) | 1.22 (0.83, 1.81) |
| Prosocial problems | 1386 | 86 | 1 | 1.21 (0.66, 2.23) | 1.35 (0.75, 2.43) | 1.21 (0.67, 2.21) | 1.18 (0.64, 2.18) |
Values are risk ratios and their 95% confidence intervals from a Poisson regression model.
Model 1 is adjusted for maternal age, race, education, insurance status, marital status, smoking, and father’s age difference; Model 2 is additionally adjusted for maternal and paternal history of affective disorder and maternal body mass index; and Model 3 is additionally adjusted for child’s sex.
ADHD, attention-deficit/hyperactivity disorder; VADPRS, Vanderbilt ADHD Diagnostic Parent Rating Scale; and SDQ, Strengths and Difficulties Questionnaire
In most instances, maternal treatment for PCOS did not appear to modify the associations with behavioral problems (Supplemental Table 4). There were suggestive differences in effect estimates for conduct disorder, and borderline emotional symptoms and prosocial problems by treatment status, but confidence intervals generally overlap. For example, children of mothers with untreated PCOS had a higher risk of conduct disorder diagnosis (aRRuntreated 2.33, 1.23–4.42; aRRtreated 1.18, 0.55–2.53) and borderline prosocial problems (aRRuntreated 2.33, 1.24–4.39; aRRtreated 0.41, 0.14–1.21) compared with children of mothers with no PCOS while the risk of borderline emotional symptoms was higher in children of mothers with treated PCOS (aRRuntreated 1.18, 0.64–2.18; aRRtreated 1.96, 1.34–2.87).
Maternal hirsutism.
Maternal hirsutism was positively associated with ADHD diagnosis in children and borderline emotional symptoms and conduct problems in unadjusted analysis (Table 3). After adjustment for covariates (Model 3), maternal hirsutism was related to children’s ADHD and conduct disorder diagnosis (2.33, 1.28–4.24 and 2.54, 1.18–5.47, respectively), and ADHD screening (2.17, 1.21–3.90). In addition, compared with mothers with no hirsutism, children of mothers with hirsutism had higher risk of borderline emotional symptoms (2.61, 1.69–4.05), peer relationship problems (1.92, 1.16–3.17), and conduct problems (2.22, 1.30–3.79).
Table 3.
Childhood behavioral problems and disorders according to maternal hirsutism status in the Upstate KIDS Study
| n | n, cases | Reference | Unadjusted RR (95% CI)a | Model 1 RR (95% CI)b | Model 2 RR (95% CI) | Model 3 RR (95% CI) | |
|---|---|---|---|---|---|---|---|
| Maternal reported diagnosis | |||||||
| ADHDc | 1906 | 173 | 1 | 2.12 (1.08, 4.16) | 2.53 (1.40, 4.59) | 2.06 (1.11, 3.81) | 2.33 (1.28, 4.24) |
| Anxiety | 1909 | 139 | 1 | 1.45 (0.63, 3.38) | 1.59 (0.68, 3.72) | 1.28 (0.55, 2.94) | 1.30 (0.56, 3.01) |
| Conduct disorder | 1905 | 116 | 1 | 2.15 (0.90, 5.14) | 2.74 (1.26, 5.97) | 2.18 (1.00, 4.75) | 2.54 (1.18, 5.47) |
| Maternal reported diagnosis or screen on VADPRS | |||||||
| ADHD | 1915 | 193 | 1 | 1.88 (0.99, 3.57) | 2.31 (1.29, 4.16) | 1.91 (1.04, 3.50) | 2.17 (1.21, 3.90) |
| Borderline behavioral problems, SDQ | |||||||
| Emotional symptoms | 1386 | 193 | 1 | 2.86 (1.90, 4.28) | 3.42 (2.33, 5.01) | 2.72 (1.75, 4.23) | 2.61 (1.69, 4.05) |
| Peer relationship problems | 1386 | 211 | 1 | 1.71 (0.94, 3.10) | 1.98 (1.22, 3.21) | 1.81 (1.10, 2.97) | 1.92 (1.16, 3.17) |
| Conduct problems | 1386 | 201 | 1 | 1.86 (1.10, 3.12) | 2.25 (1.40, 3.62) | 2.06 (1.22, 3.48) | 2.22 (1.30, 3.79) |
| Hyperactivity/inattention | 1386 | 214 | 1 | 1.48 (0.87, 2.52) | 1.66 (0.97, 2.84) | 1.34 (0.74, 2.43) | 1.46 (0.81, 2.64) |
| Prosocial problems | 1386 | 86 | 1 | 1.11 (0.42, 2.93) | 1.23 (0.47, 3.28) | 1.01 (0.37, 2.70) | 1.08 (0.42, 2.83) |
Values are risk ratios (RR) and their 95% confidence intervals from a Poisson regression model.
Model 1 is adjusted for maternal age, race, education, insurance status, marital status, smoking, and father’s age difference; Model 2 is additionally adjusted for maternal and paternal history of affective disorder and maternal body mass index; and Model 3 is additionally adjusted for child’s sex.
ADHD, attention-deficit/hyperactivity disorder; VADPRS, Vanderbilt ADHD Diagnostic Parent Rating Scale; and SDQ, Strengths and Difficulties Questionnaire
DISCUSSION
In this prospective birth cohort of mothers and children in upstate New York, maternal PCOS and hirsutism were associated with behavioral problems and mental disorders at 7 or 8 years in offspring. Maternal PCOS was associated with offspring anxiety and emotional symptoms on the SDQ, though the associations with emotional symptoms may be driven by women reporting both PCOS and hirsutism. Hirsutism was related to offspring ADHD and conduct disorder as well as SDQ emotional symptoms, peer relationship problems, and conduct problems. These associations were independent of parental socioeconomic indicators and history of affective disorders, maternal weight status, and child’s sex.
An association between maternal PCOS and offspring anxiety has not been previously reported. One previous study found a relation between maternal PCOS and offspring hospitalization for psychological development (27). Abnormal psychological development can include affective, developmental, and psychotic disorders; therefore, their results may not be comparable with ours. Two European registry linkage studies with thousands of children (10, 14) found a positive association between maternal PCOS and offspring ADHD. Although we did not find a statistically significant association between PCOS and ADHD, our point estimate (aRR=1.34) is similar to the point estimates from these studies (aOR=1.42 (14) and 1.34 (10)), which also controlled for socioeconomic indicators and maternal history of affective disorders. These large studies, however, did not evaluate other behavioral outcomes and our study adds evidence to the broader potential intergenerational impact of maternal PCOS on offspring mental health.
Maternal PCOS is associated with higher circulating androgens and inflammatory cytokines, which may influence the fetal brain development. In rodent models, a maternal PCOS-like phenotype has been related to more anxiety-like behavior (5, 6) and less sociability (28) in offspring. In the amygdala, elevated maternal androgens alter androgen and serotonin receptor expression in female rodents, as well as GABA receptor expression in both sexes (5). Further, corticotropin-releasing hormone and its receptors are altered in the hypothalamus (6). These areas of the brain are associated with regulation of mood and response to stress. Excess inflammation associated with PCOS may also relate to the development of anxiety; in rodent models, chronic and acute inflammation result in more anxiety-like behavior in offspring as well as more activated microglia in the hippocampus (7, 8). Another potential mechanism is through infertility treatment. Women with PCOS are more likely to use fertility treatments, which have been hypothesized to independently relate to child behavioral problems (29, 30).
Metformin, a common treatment for PCOS, may be continued into pregnancy. Metformin increases fetal concentration of sex hormone binding globulin (SHBG) when taken in pregnancy (31) and reduces secretion of inflammatory cytokines from trophoblast cells in vitro (32). This may mitigate metabolic irregularities in women with PCOS during pregnancy. However, we did not see consistent differences in the associations between treated and untreated maternal PCOS and offspring behavior. Lack of consistency in these results may reflect the small number of children with behavioral problems or disorders in each group. Alternatively, these varying results could indicate heterogenous treatment regimens or a discontinuation of treatment during pregnancy. For example, oral contraceptives are also commonly prescribed to treat PCOS and would be discontinued prior to attempted pregnancy. We did not have information on which medications women were taking or use during pregnancy, and therefore are not able to investigate these associations further.
Approximately one-third of women with PCOS had hirsutism. Maternal hirsutism was related to offspring ADHD and conduct disorder, as well as borderline emotional symptoms, peer relationship problems, and conduct problems on the SDQ. One previous study found an association between maternal hirsutism and offspring ASD (11). PCOS with hirsutism is often conceptualized as a subtype of PCOS indicative of high androgens. However, hirsutism develops from an interaction between androgens and the sensitivity of the women’s hair follicle to androgens (33), and can therefore be present in women with low androgens. Indeed, a recent meta-analysis among women with PCOS found that Ferriman-Gallwey scores, the “gold standard” diagnosis for hirsutism, were positively related to the testosterone and estrogen precursors androstenedione and dehydroepiandrosterone sulfate, but not to other biochemical markers of hyperandrogenism (34). Given this, further evidence is needed to test if the associations seen here between maternal hirsutism and offspring behavioral problems and disorders are mediated by elevated androgens. Replication of these findings in a prospective cohort which uses Ferriman-Gallwey scores or clinical diagnosis to identify hirsutism is necessary as well.
This study has several strengths. Upstate KIDS is a well characterized, prospective birth cohort with information collected on socioeconomic and health indicators. As such, we were able to control for many potential confounders. Child behavior was measured by maternal report of child diagnosis, the SDQ, and the VADPRS. Use of multiple indicators allowed us to compare associations across different outcome measures to see if the results are consistent.
There are limitations as well. Foremost, diagnosis of maternal gynecological conditions is self-reported. The prevalence of hirsutism among women with PCOS is low in our study, which may be indicative of misclassification of PCOS and/or hirsutism. Although self-report is likely to have high specificity, there may be low sensitivity since women with the condition may not remember their diagnosis; indeed, in a Dutch population the sensitivity and specificity of ovarian factors related to infertility, including ovarian disorders, PCOS, and premature ovarian failure, among others, were 65% and 88%, respectively (35). As such, we addressed the potential impact of measurement error in self-reported diagnoses through a sensitivity analysis which indicated our findings were robust to misclassification. Further, multiple diagnostic criteria were likely used to diagnose women with PCOS in our sample which may result in non-overlapping phenotypes among women who report having PCOS. Mother’s report of child behavioral diagnoses may be influenced by recall bias. In addition, non-response to follow-up questionnaires was large. We used IPW to attempt to account for this non-response. We also performed multiple imputations of the exposures and covariates with the assumption that data were missing at random. Finally, generalizability is limited since our sample is primarily Caucasian.
In conclusion, we report here novel associations of maternal PCOS with offspring anxiety, and hirsutism with various offspring behavioral problems and mental disorders. Our findings are limited by self-report of PCOS and hirsutism and, until replication in a study with clinical diagnosis of these conditions, should be interpreted with caution. Treatment for PCOS did not appear to have a consistent effect on offspring behavioral outcomes; however, we were unable to ascertain treatment regimens during pregnancy. Future studies are needed primarily to replicate these findings in well-characterized prospective cohorts which assessed maternal diagnosis of PCOS and/or hirsutism using medical records, and secondarily to explore the role of maternal androgens and inflammatory cytokines during pregnancy, clarify the impact of medication use during pregnancy on child behavior among women with PCOS, and determine if this association varies by diagnostic criteria used to diagnosis PCOS. If replicated, our findings suggest that children of mothers with PCOS should be monitored for the development of behavioral problems to quickly mitigate their potential impact with behavioral or pharmaceutical treatments.
Supplementary Material
Supplemental Figure 1. Results of sensitivity analysis for four misclassification error scenarios on the association between maternal PCOS and child anxiety disorder diagnosis by simulations
Fully adjusted (black) and unadjusted (grey) risk ratios from Models 3 and 0 are plotted for reference. For each misclassification error scenario (PCOS prevalence of 7, 9.5, 14.5, and 17%), the 2.5th to 97.5th percentile of fully adjusted (black) and unadjusted (grey) risk ratios are plotted. Median risk ratios (circles) are included for comparison.
ACKNOWLEDGEMENTS
The authors thank the Upstate KIDS participants and staff for their important contributions.
ROLE OF THE FUNDING SOURCE
This work was supported by the Intramural Research Program of the Eunice Kennedy Shriver National Institute of Child Health and Human Development (NICHD; contracts #HHSN275201200005C, #HHSN267200700019C, #HHSN275201400013C, #HHSN275201300026I/27500004). The sponsor played no role in the study design, data collection, data analysis or interpretation, writing of the manuscript, or the decision to submit the article for publication.
Footnotes
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
CONFLICTS OF INTEREST
None of the authors have any conflicts of interest to report.
REFERNCES
- 1.Global Burden of Disease Pediatrics C, Kyu HH, Pinho C, Wagner JA, Brown JC, Bertozzi-Villa A et al. Global and national burden of diseases and injuries among children and adolescents between 1990 and 2013: Findings from the Global Burden of Disease 2013 Study. JAMA Pediatr 2016;170:267–87. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Conway G, Dewailly D, Diamanti-Kandarakis E, Escobar-Morreale HF, Franks S, Gambineri A et al. The polycystic ovary syndrome: a position statement from the European Society of Endocrinology. Eur J Endocrinol 2014;171:P1–29. [DOI] [PubMed] [Google Scholar]
- 3.Dumesic DA, Oberfield SE, Stener-Victorin E, Marshall JC, Laven JS, Legro RS. Scientific statement on the diagnostic criteria, epidemiology, pathophysiology, and molecular genetics of polycystic ovary syndrome. Endocr Rev 2015;36:487–525. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Skiba MA, Islam RM, Bell RJ, Davis SR. Understanding variation in prevalence estimates of polycystic ovary syndrome: a systematic review and meta-analysis. Hum Reprod Update 2018. [DOI] [PubMed] [Google Scholar]
- 5.Hu M, Richard JE, Maliqueo M, Kokosar M, Fornes R, Benrick A et al. Maternal testosterone exposure increases anxiety-like behavior and impacts the limbic system in the offspring. Proc Natl Acad Sci U S A 2015;112:14348–53. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Manti M, Fornes R, Qi X, Folmerz E, Linden Hirschberg A, de Castro Barbosa T et al. Maternal androgen excess and obesity induce sexually dimorphic anxiety-like behavior in the offspring. FASEB J 2018;32:4158–71. [DOI] [PubMed] [Google Scholar]
- 7.Bilbo SD, Tsang V. Enduring consequences of maternal obesity for brain inflammation and behavior of offspring. FASEB J 2010;24:2104–15. [DOI] [PubMed] [Google Scholar]
- 8.Makinson R, Lloyd K, Rayasam A, McKee S, Brown A, Barila G et al. Intrauterine inflammation induces sex-specific effects on neuroinflammation, white matter, and behavior. Brain Behav Immun 2017;66:277–88. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Kosidou K, Dalman C, Widman L, Arver S, Lee BK, Magnusson C et al. Maternal polycystic ovary syndrome and the risk of autism spectrum disorders in the offspring: a population-based nationwide study in Sweden. Mol Psychiatry 2016;21:1441–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Berni TR, Morgan CL, Berni ER, Rees DA. Polycystic ovary syndrome is associated with adverse mental health and neurodevelopmental outcomes. J Clin Endocrinol Metab 2018;103:2116–25. [DOI] [PubMed] [Google Scholar]
- 11.Lee BK, Arver S, Widman L, Gardner RM, Magnusson C, Dalman C et al. Maternal hirsutism and autism spectrum disorders in offspring. Autism Res 2017;10:1544–6. [DOI] [PubMed] [Google Scholar]
- 12.Bell GA, Sundaram R, Mumford SL, Park H, Mills J, Bell EM et al. Maternal polycystic ovarian syndrome and early offspring development. Hum Reprod 2018;33:1307–15. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Palomba S, Marotta R, Di Cello A, Russo T, Falbo A, Orio F et al. Pervasive developmental disorders in children of hyperandrogenic women with polycystic ovary syndrome: a longitudinal case-control study. Clin Endocrinol (Oxf) 2012;77:898–904. [DOI] [PubMed] [Google Scholar]
- 14.Kosidou K, Dalman C, Widman L, Arver S, Lee BK, Magnusson C et al. Maternal polycystic ovary syndrome and risk for attention-deficit/hyperactivity disorder in the offspring. Biol Psychiatry 2017;82:651–9. [DOI] [PubMed] [Google Scholar]
- 15.Buck Louis GM, Hediger ML, Bell EM, Kus CA, Sundaram R, McLain AC et al. Methodology for establishing a population-based birth cohort focusing on couple fertility and children’s development, the Upstate KIDS Study. Paediatr Perinat Epidemiol 2014;28:191–202. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Yeung EH, Sundaram R, Bell EM, Druschel C, Kus C, Ghassabian A et al. Examining infertility treatment and early childhood development in the Upstate KIDS Study. JAMA Pediatr 2016;170:251–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Goodman R Psychometric properties of the Strengths and Difficulties Questionnaire. J Am Acad Child Adolesc Psychiatry 2001;40:1337–45. [DOI] [PubMed] [Google Scholar]
- 18.Bard DE, Wolraich ML, Neas B, Doffing M, Laoma B. The psychometric properties of the Vanderbilt Attention-Deficit Hyperactivity Disorder Diagnostic Parent Rating Scale in a community population. J Dev Behav Pediatr 2013;34:72–82. [DOI] [PubMed] [Google Scholar]
- 19.Ghassabian A, Bell EM, Ma WL, Sundaram R, Kannan K, Buck Louis GM et al. Concentrations of perfluoroalkyl substances and bisphenol A in newborn dried blood spots and the association with child behavior. Environ Pollut 2018;243:1629–36. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Scoring the Strengths & Difficulties Questionnaire for age 4–17 or 18+. In. Vol. 2019 http://www.sdqinfo.org/py/sdqinfo/c0.py: youthinmind, 2016.
- 21.He JP, Burstein M, Schmitz A, Merikangas KR. The Strengths and Difficulties Questionnaire (SDQ): the factor structure and scale validation in U.S. adolescents. J Abnorm Child Psychol 2013;41:583–95. [DOI] [PubMed] [Google Scholar]
- 22.Becker SP, Langberg JM, Vaughn AJ, Epstein JN. Clinical utility of the Vanderbilt ADHD diagnostic parent rating scale comorbidity screening scales. J Dev Behav Pediatr 2012;33:221–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Wolraich ML. Psychometric properties of the Vanderbilt ADHD Diagnostic Parent Rating Scale in a Referred Population. Journal of Pediatric Psychology 2003;28:559–68. [DOI] [PubMed] [Google Scholar]
- 24.NICHQ Vanderbilt Assessment Scales: Used for diagnosing ADHD. In. National Institute for Children’s Health Quality: American Academy of Pediatrics and National Initiative for Children’s Healthcare Quality, 2002. [Google Scholar]
- 25.Hernandez-Diaz S, Wilcox AJ, Schisterman EF, Hernan MA. From causal diagrams to birth weight-specific curves of infant mortality. Eur J Epidemiol 2008;23:163–6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Glymour MM, Greenland S. Causal Diagrams. In: Seigafuse S, ed. Modern Epidemiology Philadelphia, PA: Lippincott Williams & Wilkins, 2008:183–212. [Google Scholar]
- 27.Doherty DA, Newnham JP, Bower C, Hart R. Implications of polycystic ovary syndrome for pregnancy and for the health of offspring. Obstet Gynecol 2015;125:1397–406. [DOI] [PubMed] [Google Scholar]
- 28.Xu XJ, Zhang HF, Shou XJ, Li J, Jing WL, Zhou Y et al. Prenatal hyperandrogenic environment induced autistic-like behavior in rat offspring. Physiol Behav 2015;138:13–20. [DOI] [PubMed] [Google Scholar]
- 29.Carson C, Redshaw M, Sacker A, Kelly Y, Kurinczuk JJ, Quigley MA. Effects of pregnancy planning, fertility, and assisted reproductive treatment on child behavioral problems at 5 and 7 years: evidence from the Millennium Cohort Study. Fertil Steril 2013;99:456–63. [DOI] [PubMed] [Google Scholar]
- 30.Punamaki RL, Tiitinen A, Lindblom J, Unkila-Kallio L, Flykt M, Vanska M et al. Mental health and developmental outcomes for children born after ART: a comparative prospective study on child gender and treatment type. Hum Reprod 2016;31:100–7. [DOI] [PubMed] [Google Scholar]
- 31.Carlsen SM, Vanky E. Metformin influence on hormone levels at birth, in PCOS mothers and their newborns. Hum Reprod 2010;25:786–90. [DOI] [PubMed] [Google Scholar]
- 32.Han CS, Herrin MA, Pitruzzello MC, Mulla MJ, Werner EF, Pettker CM et al. Glucose and metformin modulate human first trimester trophoblast function: a model and potential therapy for diabetes-associated uteroplacental insufficiency. Am J Reprod Immunol 2015;73:362–71. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Rosenfield RL. Hirsutism. N Engl J Med 2005;353:2578–88. [DOI] [PubMed] [Google Scholar]
- 34.Amiri M, Ramezani Tehrani F, Nahidi F, Bidhendi Yarandi R, Behboudi-Gandevani S, Azizi F. Association between biochemical hyperandrogenism parameters and Ferriman-Gallwey score in patients with polycystic ovary syndrome: A systematic review and meta-regression analysis. Clin Endocrinol (Oxf) 2017;87:217–30. [DOI] [PubMed] [Google Scholar]
- 35.de Boer EJ, den Tonkelaar I, Burger CW, van Leeuwen FE, Group OP. Validity of self-reported causes of subfertility. Am J Epidemiol 2005;161:978–86. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Supplemental Figure 1. Results of sensitivity analysis for four misclassification error scenarios on the association between maternal PCOS and child anxiety disorder diagnosis by simulations
Fully adjusted (black) and unadjusted (grey) risk ratios from Models 3 and 0 are plotted for reference. For each misclassification error scenario (PCOS prevalence of 7, 9.5, 14.5, and 17%), the 2.5th to 97.5th percentile of fully adjusted (black) and unadjusted (grey) risk ratios are plotted. Median risk ratios (circles) are included for comparison.