Background
Women with polycystic ovary syndrome are more likely to experience several pregnancy complications including hypertensive disorders, gestational diabetes mellitus, and preterm births than women without polycystic ovary syndrome. However, at present, there is limited research on whether polycystic ovary syndrome is associated with both anxiety and depression during pregnancy and whether this augments a woman’s risk of postpartum depression, particularly among high-risk populations who have limited access to care.
Objective
Our primary objective was to assess the association between prepregnancy polycystic ovary syndrome and postpartum depression, considering important baseline confounding factors. Our secondary objective was to evaluate the mediating role of prenatal depression and anxiety on the association between polycystic ovary syndrome and postpartum depression.
Study Design
This study involved a population-based sample of 3906 postpartum (2–6 months) women who completed the Utah Pregnancy Risk Assessment Monitoring System Phase 8 questionnaire (2016–2018). Weighted adjusted prevalence ratios were used to assess the association between polycystic ovary syndrome and postpartum depression, considering potential confounding factors and assessing mediating effects of depression and anxiety experienced during pregnancy.
Results
Following the exclusion criteria, 8.2% of women reported clinical polycystic ovary syndrome and 19.1%, 6.2%, and 4.4% reported irregular periods and acne, irregular periods and hirsutism, and all 3 symptoms, respectively. Moreover, 17.7% and 23.5% reported experiencing prenatal depression and anxiety and 9.5% and 10.2% reported experiencing postpartum depressed mood and anhedonia, respectively. Clinical polycystic ovary syndrome was associated with a 1.76 higher adjusted prevalence ratio (95% confidence interval, 1.03–3.00) for postpartum depressed mood or anhedonia after taking into consideration age, prepregnancy body mass index, race/ethnicity, education, and marital status. A similar higher prevalence was seen for irregular periods and acne (adjusted prevalence ratio, 1.65; 95% confidence interval, 1.13–2.41), irregular periods and hirsutism (adjusted prevalence ratio, 1.40; 95% confidence interval, 0.82–2.40), and all 3 symptoms (adjusted prevalence ratio, 1.75; 95% confidence interval, 0.96–3.19) and postpartum depressed mood or anhedonia. Prenatal depression and anxiety mediated 20% and 32% of the effect of clinical polycystic ovary syndrome on postpartum depressed mood and anhedonia, respectively.
Conclusion
Clinical polycystic ovary syndrome is associated with postpartum depressed mood and symptoms among this population-based sample inclusive of high-risk mothers. Prenatal depression and anxiety mediate this association, emphasizing the importance of prenatal psychological screening among women with polycystic ovary syndrome. An additional important clinical and public health implication of this study lies in the finding that nearly 20% of women in this population-based sample who reported at least 2 polycystic ovary syndrome symptoms (including at-risk women who may not have access to care) had not received a clinical diagnosis for polycystic ovary syndrome.
Introduction
Polycystic ovary syndrome (PCOS) is the most common endocrine metabolic disorder in women of reproductive age worldwide. PCOS is characterized by anovulation, hyperandrogenism, and polycystic ovarian morphology, with a prevalence ranging between 4% and 12% (approximately 4 million women) in the United States. The yearly economic impact on the United States healthcare system to identify and manage PCOS among women totals more than $4 billion.
Why was this study conducted?
Women with polycystic ovary syndrome (PCOS) have increased susceptibility to depression (28%–64%) and anxiety (34%–57%) outside of pregnancy, yet there is limited population-based research on whether PCOS is associated with both anxiety and depression during pregnancy and how this may augment a woman’s risk of postpartum depression.
Key findings
Clinical PCOS was associated with a higher prevalence of postpartum depressed mood and anhedonia after adjusting for potential confounders. Prenatal depression and anxiety partially mediated the association between clinical PCOS and postpartum depressed mood and anhedonia.
What does this add to what is known?
Clinical PCOS and symptomology are associated with postpartum depressed mood among a population-based sample inclusive of high-risk mothers. Prenatal depression and anxiety mediate the association between PCOS and postpartum depressed mood, emphasizing the importance of prenatal psychological screening among women with PCOS.
Women with PCOS are more likely to experience several pregnancy complications including hypertensive disorders, gestational diabetes, and preterm births and are substantially more likely to experience infertility and seek related treatment than those without the syndrome. In addition, existing research has demonstrated that nonpregnant women with PCOS have increased susceptibility to depression (28%–64%) and anxiety (34%–57%), namely, infertile women. Although the risk of depression and anxiety in pregnant women with PCOS is expected to be higher, previous research has reported higher prenatal and postpartum depression (PPD), but not postnatal depression, or postnatal depression or anxiety in women with PCOS compared with those without the syndrome. Whether PCOS is associated with both anxiety and depression during pregnancy and whether this augments a woman’s risk of PPD, particularly among high-risk populations who have limited access to care, remain to be studied. To address this knowledge gap, our primary objective was to assess the association between a prepregnancy diagnosis and symptoms of PCOS and prevalence of depression and anxiety during pregnancy and in the postpartum period among a population-based sample. Our secondary objective was to assess whether prenatal depression, anxiety, and pregnancy complications mediate the association between PCOS and PPD.
Materials and Methods
Study participants and questionnaire
The study population included women who completed the Utah Pregnancy Risk Assessment Monitoring System (UT-PRAMS) Phase 8 questionnaire between 2016 and 2018. In collaboration with state health departments, PRAMS is conducted by the Centers for Disease Control and Prevention’s Division of Reproductive Health. One key aspect of PRAMS is the stratified systematic sampling, which oversamples on features related to high-risk women (eg, mothers of low-birth-weight infants, those living in high-risk geographic areas, and racial/ethnic minority groups). A detailed description of the PRAMS surveillance system methodology and protocols is found elsewhere.
UT-PRAMS Phase 8 (2016–2018) drew stratified (by maternal education and infant birthweight) samples of approximately 200 new mothers (2–6 months after delivery) every month. New mothers are contacted via mailed questionnaire (available in English and Spanish) multiple times and telephone follow-up. An informed consent document was included within each survey packet explaining the participants’ rights. Consent is implied if the survey is completed. Similarly, the informed consent document is read verbally on phone interviews, and the participant verbally agrees to proceed with the survey. No written consent was required. The data are analyzed and presented in aggregate, with no individual case data published. The expected national PRAMS response rate is 60%. UT-PRAMS response rates were 65%, 66%, and 62% for 2016, 2017, and 2018, respectively.
Mothers’ responses were linked to extracted birth certificate data items, including pregnancy complications for index birth. The PRAMS weighting process produces an analysis weight considering the stratified sampling along with nonresponse and noncoverage components. The analysis weight of the PRAMS data can be interpreted as the number of women like herself in the population that each respondent represents. This study and the use of PRAMS data (deidentified) have been acknowledged by the University of Utah as a nonhuman subject research (University of Utah Institutional Review Board # 00130386).
Exposure
The presence of PCOS before pregnancy was assessed based on clinical PCOS and common symptoms. The PCOS diagnosis question asked, “Have you ever been told that you have Polycystic Ovarian Syndrome or PCOS by a doctor, nurse, or other healthcare worker?”—requiring a response of “yes or no” or “do not know.” A PCOS symptomology question asked, “Have you ever experienced any of the following health problems ? ”—with the following choices: (1) “Irregular periods (menstruation)”; (2) “Skin condition that causes pimples (acne)”; (3) “Increased hair growth on the face, chest, or other parts of the body (hirsutism)”; and (4) “Being overweight or obese.” PCOS symptomology was defined in possible alternate ways as having (1) irregular periods and acne, (2) irregular periods and hirsutism, or (3) irregular periods, acne, and hirsutism.
Outcomes
To assess the presence of prenatal or pregnancy depression and anxiety, women were asked, “During your most recent pregnancy, did you have any of the following health conditions?”—where depression and anxiety were listed as possible choices, with responses “yes or no.” PPD was defined having answered “always” or “often” to either of the following 2 questions that captured PPD or a postpartum depressed mood (1) and anhedonia (2): (1) “Since your new baby was born, how often have you felt down, depressed, or hopeless?” and (2) “Since your new baby was born, how often have you had little interest or little pleasure in doing things you usually enjoyed?” Combined effect variables were created to include those who had (1) both pregnancy depression and pregnancy anxiety and (2) either pregnancy depression or pregnancy anxiety. Similarly, PPD effect variables were created for those who had (3) both postpartum depressed mood and anhedonia and (4) either postpartum depressed mood or anhedonia.
Covariates
Covariates included maternal age, body mass index (BMI), race/ethnicity, education, and marital status. Prepregnancy BMI was calculated from self-reported height and weight from the birth certificate data. Maternal age and BMI were assessed both continuously and categorically. Although birth certificate data captures more detailed race/ethnicity information, UT-PRAMS is restricted to providing information on White or nonwhite and Hispanic or non-Hispanic regarding race/ethnicity owing to privacy issues. Marital status was defined as “married or other.” Finally, a previous preterm birth (<37 weeks) and infertility treatment for index pregnancy were reported as “yes or no.”
Variables available from the PRAMS questionnaire included smoking (yes or no, last 2 years) and alcohol (yes or no, last 2 years). Maternal education was recategorized from 8 to 5 categories (0–8, 9–11, 12, 13–15, and ≥16 years). Those with less than 9th-grade education or high school without a diploma, college and associate degrees, and master’s and professional degrees were combined for the analysis. Delivery modality was grouped into 2 categories: (1) cesarean delivery and (2) vaginal delivery (spontaneous vaginal, unsuccessful or successful forceps vaginal, or successful or unsuccessful vacuum vaginal). Preterm birth for index pregnancy was defined as less than 37 week of the gestational period. Small for gestational age (SGA) was defined as weight below the 10th percentile for gestational age. Finally, gestational diabetes mellitus (GDM), hypertensive diseases of pregnancy (HDP), and 3-month previous pregnancy depression questions (yes or no) were available from the PRAMS questionnaire.
Statistical analysis
Sociodemographic and health history characteristics among women with and without PCOS were compared using the chi-square test for categorical and t test for continuous variables, considering the complex sampling design. To test the association between PCOS and postpartum depressive symptoms, unadjusted and adjusted robust Poisson distribution models were used to estimate prevalence ratio (PR) and 95% confidence intervals (CIs). This was done using the Stata SVY: GLM (Stata-Corp LP, College Station, TX) function to account for weighted survey data.
The selection of potential covariates was informed by previous literature , , and assessed for confounding and mediation using directed acyclic graphs. The final multivariate regression models were adjusted for age, race/ethnicity, maternal education, marital status, and prepregnancy BMI. Based on a priori hypothesis that depression and anxiety, along with pregnancy complications, may mediate the relationship between PCOS and postpartum depressive symptoms, a counterfactual approach to mediation analysis was applied to estimate the PR for the natural direct effect (NDE) and the natural indirect effect (NIE) of clinical PCOS and symptoms on postpartum depressive symptoms mediated through clinical and psychological factors ( Figure 1 ). This was done by using the Stata PARAMED (Stata-Corp LP) function for mediating variables for clinical PCOS and PCOS defined as 2 or more symptoms. We also estimated proportion mediated to reflect the extent of mediation, where 100% indicates all of the total effect being mediated (no direct effect) and 0% indicates there is no mediation (all direct effect).
Given that preconception depression (up to 3 months before conception) and infertility treatment for index pregnancy may act as confounders (common causes of PCOS and postpartum depressive symptoms) or mediators (on the pathway between PCOS and postpartum depressive symptoms), they were assessed both ways. Multiple sensitivity analyses were conducted for those women with concordant answers for having had PCOS before pregnancy (3 months before) and having ever been diagnosed as having PCOS by a healthcare worker (n=3686).
Exclusion criteria
Among the total sample of 4101 women who completed the survey, 145 women whose infant was not living at the time they completed the survey and 152 women who had no information regarding the presence of PCOS were excluded. The 3906 women included in the analysis reflect an estimated population of 142,963 women as per the PRAMS sampling strategy. Analyses were performed in SAS 9.4 (SAS Institute, Cary, NC) and STATA 15.0 (Stata-Corp LP).
Results
In weighted analyses, of the 3906 women, 8.2% of women were diagnosed as having PCOS ( Table 1 ). Those with clinical PCOS were older (μ=30.4 years vs 28.6 years; P <.001), had higher BMI (28.7 vs 25.8; P <.001), and were more likely to report prenatal depression (21.2% vs 17.4%; P =.18), prenatal anxiety (30.8% vs 22.9%), and postpartum depressed mood (12.0% vs 9.2%) or anhedonia (13.6% vs 9.9%) than those without the syndrome. Similarly, women with clinical PCOS were more likely to report irregular periods and acne, irregular periods and hirsutism, and irregular periods, acne, and hirsutism (48% vs 19%, 43% vs 6%, and 30% vs 4%) ( Figure 2 ).
| Characteristics, %, μ±SD | Total | Polycystic ovary syndrome | P value b | |
|---|---|---|---|---|
| Yes (8.19%) | No (91.81%) | |||
| Age, y | 28.73±0.10 | 30.43±0.38 | 28.57±0.11 | <.001 |
| <18 | 0.72 | 0.14 | 0.78 | |
| 18–24 | 23.54 | 14.64 | 24.34 | |
| 25–29 | 33.14 | 30.24 | 33.40 | |
| 30+ | 42.60 | 54.98 | 41.48 | |
| BMI, kg/m 2 c | 26.04±0.12 | 28.65±0.54 | 25.80±0.12 | <.001 |
| <18 | 6.11 | 4.40 | 6.26 | |
| 18–24 | 53.53 | 42.67 | 54.49 | |
| 25–29 | 18.19 | 15.44 | 18.44 | |
| 30+ | 22.17 | 37.49 | 20.81 | |
| Ethnicity | ||||
| Non-Hispanic | 84.50 | 85.77 | 84.38 | .62 |
| Hispanic | 15.50 | 14.23 | 15.62 | |
| Race | ||||
| Mother White | 91.15 | 94.61 | 90.84 | .04 |
| Mother nonwhite | 8.85 | 5.39 | 9.16 | |
| Education c | ||||
| Up to 12th grade | 8.08 | 4.32 | 8.41 | <.001 |
| High school | 19.73 | 13.34 | 20.31 | |
| Some college | 34.45 | 32.87 | 34.59 | |
| Bachelor’s degree | 30.54 | 39.91 | 29.70 | |
| Higher education | 7.20 | 9.56 | 6.99 | |
| Married | 82.58 | 91.18 | 81.81 | <.001 |
| Smoker c , d | 2.83 | 2.99 | 2.82 | .88 |
| Drinker d | 33.51 | 31.92 | 33.66 | .62 |
| Hypertensive diseases of pregnancy c | 10.51 | 14.64 | 10.15 | .04 |
| Gestational diabetes mellitus for index pregnancy c | 7.49 | 12.20 | 7.07 | .01 |
| Preterm births (<37 wk) for index pregnancy | 8.12 | 11.81 | 7.79 | .01 |
| Previous preterm births (<37 wk) | 5.46 | 8.61 | 5.18 | .04 |
| Small for gestational age c | 2.04 | 2.89 | 1.96 | .02 |
| Infertility treatment | 6.74 | 32.43 | 4.44 | <.001 |
| Delivery modality | ||||
| Vaginal | 78.64 | 72.48 | 79.19 | |
| Cesarean | 21.36 | 27.52 | 20.81 | .03 |
| Previous depression (3 mo before pregnancy) | 17.18 | 18.23 | 17.09 | .67 |
| Prenatal psychological symptoms | ||||
| Prenatal depression c | 17.73 | 21.24 | 17.42 | .18 |
| Prenatal anxiety c | 23.52 | 30.81 | 22.88 | .02 |
| Prenatal depression and anxiety | 13.64 | 15.76 | 13.45 | .37 |
| Prenatal depression or anxiety | 27.51 | 35.78 | 26.80 | .01 |
| Postpartum psychological symptoms | ||||
| Postpartum depressed mood c | 9.46 | 11.97 | 9.24 | .20 |
| Postpartum anhedonia c | 10.16 | 13.55 | 9.86 | .10 |
| Postpartum depressed mood and anhedonia | 4.87 | 7.92 | 4.60 | .04 |
| Postpartum depressed mood or anhedonia | 14.80 | 17.72 | 14.54 | .24 |
a Reflecting an estimated population size of 142,963 women
c Frequencies do not add up to 100% because of missing values
d Smoker and drinker measured at the last 2 years before pregnancy.
A higher adjusted prevalence of prenatal anxiety (adjusted PR [aPR], 1.07; 95% CI, 1.01–1.13), prenatal depression or anxiety (aPR, 1.32; 95% CI, 1.06–1.63), postpartum anhedonia (aPR, 1.54; 95% CI, 1.04–2.27), and postpartum depressed mood and anhedonia (aPR, 1.76; 95% CI, 1.03– 3.00) was observed among those with clinical PCOS than those without the syndrome ( Table 2 ). Consistent with primary analysis, most prevalence estimates were similar in magnitude but differed in significance in sensitivity analyses ( Supplemental Tables 1 and 2 ).
