Obesity but Not Polycystic Ovary Syndrome Associated with Decreased Breastfeeding Initiation Rates

Abstract

Objective:

To determine whether women with polycystic ovary syndrome (PCOS) were less likely to initiate breastfeeding or have shorter breastfeeding duration.

Materials and Methods:

Cross-sectional analysis was performed of the Pregnancy Risk Assessment Monitoring System dataset, a national questionnaire from the Centers for Disease Control and Prevention of postpartum women, from 2016 to 2018. Logistic regression assessed the odds of any breastfeeding initiation. Cox proportional hazards assessed duration of any breastfeeding.

Results:

Sample included 2,382,290 women (6.1% PCOS). In univariable analysis, PCOS was associated with increased odds of any breastfeeding initiation (89.9% versus 87.9%; odds ratio [OR]: 1.23 [95% confidence interval: 1.02, 1.47]; p = 0.03). This outcome remained significant after controlling for body mass index (BMI; OR_adj: 1.3 [1.1, 1.6]; p = 0.005) but not after controlling for education and prior live births (OR_adj: 1.10 [0.89, 1.37]; p = 0.37). With increasing BMI, the odds of any breastfeeding initiation decreased, with the lowest odds seen in women with class III obesity (OR_adj: 0.74 [0.60, 0.9]; p = 0.003). In a subanalysis of racial/ethnic groups, PCOS did not impact any breastfeeding initiation in White or Black non-Hispanic groups but increased odds of any breastfeeding initiation in Hispanic women (OR_adj: 2.0 [1.1, 3.7]; p = 0.03). In multivariable models, there was no difference in the duration of any breastfeeding in women with PCOS compared with those without.

Conclusions:

Understanding predictors of breastfeeding success is paramount. In this national survey measuring any breastfeeding, PCOS did not decrease breastfeeding initiation or duration, despite confirming the association between overweight/obesity and decreased breastfeeding. However, because the data did not distinguish between exclusive breastfeeding and supplementation, we cannot rule out the possibility that PCOS affects breastfeeding exclusivity or necessitates supplementation. This limitation suggests that important trends could be obscured, and therefore, our findings should be interpreted with caution regarding breastfeeding exclusivity. Interventions aimed at increasing breastfeeding should target populations that would benefit the most; our data support that PCOS-specific targeting is not needed. Additional prospective studies are necessary to fully understand the association between different PCOS phenotypes and breastfeeding.

Introduction

Breastfeeding has short- and long-term health benefits that are dose-related for both infants and those who give birth and is currently recommended by the Centers for Disease Control and Prevention (CDC), American College of Obstetricians and Gynecologists, American Academy of Pediatrics, and American Academy of Family Physicians.^1–4 A systematic review concluded that infants who were breastfed scored higher on intelligence tests and had a decreased risk for type 2 diabetes, overweight/obesity, hypercholesterolemia, hypertension, leukemia, and sudden infant death syndrome.^5–9 For postpartum women, breastfeeding has protective effects against type 2 diabetes,¹⁰ reduces rates of ovarian and breast cancer,¹¹ and decreases the risk of developing nonalcoholic fatty liver disease, hypertension, myocardial infarction, and stroke.^12–14 Given these significant health benefits for women and children, it is important to identify conditions that may lead to decreased breastfeeding so that targeted education and interventions can be made.

Polycystic ovary syndrome (PCOS) is the most common endocrine disorder in reproductive-aged women with ovaries, affecting 6–10% of women worldwide, and contributes to reproductive, psychological, metabolic, and pregnancy complications.¹⁵ Based on Rotterdam criteria, PCOS is clinically defined by two of the following three features: oligomenorrhea, hyperandrogenism, and polycystic appearing ovaries on ultrasound.¹⁶ Forty-six percent of women with PCOS are overweight (body mass index [BMI] 25–29.9 kg/m²) or obese (BMI class I: 30.0–34.9, II: 35.0–39.9, III: ≥40 kg/m²).^17,18 Maternal obesity is a detriment to breastfeeding success and is associated with decreased breastfeeding intention, initiation, and duration, impaired milk production, and delayed onset of lactogenesis II (copious milk production) when compared with normal-weight women.^19,20 This adverse relationship between breastfeeding and BMI is influenced by physiological, mechanical, and psychological factors.

Early observational studies raised concerns about decreased rates of exclusive breastfeeding²¹ and increased rates of low milk production in women with PCOS.²² Two small studies demonstrated that hyperandrogenism, a prominent feature of PCOS, is associated with decreased breastfeeding initiation.^21,23 They hypothesize that high androgen levels may inhibit glandular differentiation during pregnancy, directly inhibit lactation, and potentiate psychological effects that may reduce intent to breastfeed.²³ However, these studies did not adjust for the confounding effect of obesity. An association between PCOS and decreased breastfeeding duration was not supported by a large cross-sectional study in Australia.²⁴ Given the scarce and conflicting data on breastfeeding in women with PCOS and the importance of determining risk factors for decreased breastfeeding, our objective was to use a large dataset to evaluate whether women with PCOS in the United States have decreased rates of breastfeeding initiation and shorter breastfeeding duration.

Methods

Study population and sample

We conducted a cross-sectional analysis of participants using the Pregnancy Risk Assessment Monitoring System (PRAMS) dataset, a national questionnaire from the CDC sent to postpartum women 2–9 months after delivery. PRAMS collects state-specific, population-based data on maternal attitudes and experiences before, during, and after pregnancy. PRAMS uses a standardized data collection system that includes mandated core questions and optional standard questions with the goal of improving the health of mothers and infants. Each state samples 1,300–3,400 women per year and covers approximately 83% of all U.S. births. Our analysis used the Phase 8 Core PRAMS Research File from 2016 to 2018, coinciding with the initial inclusion of a PCOS-status survey question. Data on PCOS were available from Puerto Rico and 17 states (Arkansas, Connecticut, Delaware, Hawaii, Maryland, Maine, Michigan, Minnesota, Mississippi, Missouri, New Jersey, New York and New York City, Oklahoma, Pennsylvania, Utah, Wisconsin, and West Virginia). Data are assigned a sampling weight to adjust for nonresponse to produce population-based estimates.²⁵

Exposure measures

PCOS status was self-reported by answering “yes” to the question “During the 3 months before you got pregnant with your new baby, did you have any of the following health conditions? For each one, check ‘No’ if you did not have the condition or ‘Yes’ if you did.”

a. Asthma

b. Anemia (poor blood, low iron)

c. Heart problems

d. Epilepsy (seizures)

e. Thyroid problems

f. PCOS (polycystic ovarian syndrome)

g. Anxiety

Outcome measures

The primary outcome was any breastfeeding initiation within the postpartum period defined by a “yes” in response to “Did you ever breastfeed or pump breast milk to feed your new baby after delivery, even for a short period of time?” To investigate the impact of race/ethnicity on the PCOS and any breastfeeding initiation relationship, a subanalysis was performed including only those women who self-identified as Black non-Hispanic, Hispanic, or White non-Hispanic. The secondary outcomes were (1) breastfeeding duration (weeks) and (2) “short breastfeeding duration” defined as breastfeeding ≤4 weeks. For women who were still breastfeeding at the time of completion of their follow-up survey, their breastfeeding length was set to their current number of weeks postpartum and was censored in the Cox proportional hazards model.

Inclusion and exclusion

We included subjects with only term singleton deliveries of live infants from states that inquired about PCOS status. Subjects with missing outcome data (any breastfeeding initiation or duration) were excluded from their respective models. Those who responded to the survey ≥9 months from delivery were excluded (predefined response time limit set by PRAMS).

Covariates

PRAMS provided maternal demographic and medical information: age, race, ethnicity, BMI, gravidity, number of prepregnancy stressors, smoking status, history of diabetes, depression, anxiety, or infertility; and pregnancy/delivery history: Kotelchuck prenatal care index, plurality, mode of delivery, gestational age at delivery, length of hospital stay, and pregnancy intention.

Statistical analysis

Descriptive and inferential statistics used survey sample weights provided by PRAMS. Pearson χ² tests were used to describe associations between PCOS and each covariate. Subjects with unknown PCOS status were excluded from these χ² tests to conduct pairwise comparisons between those with and without PCOS. Binomial logistic regression models were used to estimate unadjusted and adjusted odds ratios (ORs). Length of breastfeeding was assessed using Cox proportional hazards with right censoring for women who were still breastfeeding at the time of follow-up. The adjusted primary and secondary models included all covariates. A p value of <0.05 was considered significant for all analyses. STATA SE version 15 was used for all summaries and analyses. The University of Wisconsin Institutional Review Board considered this study exempt.

Results

Data from 44,006 participants were included, representing 2,426,384 women due to sample weights. Less than 2% of participants were missing breastfeeding (1.8%) or PCOS (1.2%) status; however, those with PCOS were the least likely group to have missing breastfeeding status (PCOS: 0.6%, no PCOS: 1.9%, missing PCOS status: 1.9%; p = 0.001). Women with missing outcome data (breastfeeding status) were excluded. The remaining dataset (N = 2,382,290) represents 144,799 women with PCOS (6.1%), 2,207,906 without PCOS (92.7%), and 29,585 with missing PCOS status (1.2%). Median response time was 17.7 ± 5.4 weeks postpartum. Women with PCOS responded to the survey on average 1 week sooner than those without PCOS (16.9 ± 4.8 versus 17.8 ± 5.4 weeks, p < 0.001). The 2016 PRAMS weighted response rates ranged from 54.7% (Michigan) to 72.5% (New York City).²⁶

Table 1 shows baseline demographic characteristics in women with and without PCOS. Women with PCOS were more likely to be older, obese, and White non-Hispanic, have higher educational attainment, and for this to be their first pregnancy. They were also more likely to have metabolic comorbidities (diabetes and gestational diabetes) and mood disorders (depression and anxiety).

Table 1.

Demographics

	No PCOS (%)	PCOS (%)	Total	p value
Total	92.0	6.8
Maternal age (years)				p < 0.001
<17	1.1	0.4	10.6
18–19	2.9	1.8	28.7
20–24	18.1	11.3	17.7
25–29	29.4	29.7	29.5
30–34	30.0	35.3	30.3
35–39	15.1	17.5	15.2
40+	3.3	3.9	33.3
BMI (kg/m²)				p < 0.001
Underweight (<19.8)	8.9	6.4	8.8
Normal (19.8–25)	39.6	32.5	39.2
Overweight (25–30)	24.7	22.5	24.5
Obese class I (30–35)	12.1	17.7	12.5
Obese class II (35–40)	5.6	9.5	58.5
Obese class III (>40)	4.1	8.7	4.4
Maternal race				p < 0.001
Alaskan, Hawaiian, or Native American	0.9	0.9	0.9
Asian	6.3	7.1	6.4
Black	13.9	8.5	13.6
White	69.1	75.9	7.0
Other non-White	5.1	3.5	5.0
Mixed race	3.5	3.6	3.5
Maternal ethnicity				p < 0.001
Non-Hispanic	82.7	88.7	83.1
Hispanic	16.7	10.8	16.3
Maternal education (years)				p < 0.001
0–8	3.2	0.6	3.0
9–11	8.6	4.0	8.4
12	2.4	18.7	23.5
13–15	2.6	30.6	26.6
≥16	3.7	45.5	38.0
Number of previous live births				p < 0.001
0	37.9	45.4	38.3
1	33.7	34.3	33.7
2	16.6	13.1	16.3
3–5	10.5	6.7	10.3
6+	12.6	0.3	1.2
Pregnancy intent				p < 0.001
Then	45.1	39.3	44.7
Later	19.0	14.9	18.7
Sooner	13.6	26.7	14.4
Was not sure	15.0	12.0	14.8
Did not want then/anytime	5.6	4.9	5.5
History of depression				p < 0.001
None	83.6	53.8	81.7
Prenatal only	3.5	30.8	7.8
Postnatal onset	3.8	4.3	3.9
Persistent depression	7.7	9.0	5.1
Prenatal history of anxiety				p < 0.001
No	75.9	47.9	74.1
Yes	16.6	43.0	18.2
History of diabetes				p < 0.001
No	98.3	63.9	96.2
Yes	1.3	35.0	3.4
Smoker				p = 0.9
No	88.0	88.3	88.0
Yes	11.2	10.9	11.2
Stressors				p = 0.8
0	26.1	25.9	26.1
1–2	34.1	33.4	34.0
3–4	16.4	15.9	16.4
6+	4.1	4.4	4.2
Prenatal care index (Kotelchuck)				p < 0.001
Inadequate	11.0	7.8	10.8
Intermediate	10.8	8.8	10.7
Adequate	48.8	45.8	48.6
Adequate plus	26.8	35.8	27.3
Infertility treatment				p < 0.001
No	98.3	91.0	97.9
Yes	1.6	8.8	2.0
Gestational diabetes				p < 0.001
No	90.3	85.4	90.0
Yes	8.8	13.4	9.1
Mode of delivery				p < 0.001
Vaginal	67.7	61.7	67.3
C section	28.9	34.5	29.3
Forceps	0.5	0.9	0.6
Vacuum delivery	2.8	2.9	2.8
Length in hospital				p = 0.01
<1 day	2.0	21.8	20.1
1–2 days	59.4	57.0	59.3
3–5 days	32.7	35.0	32.8
6–14 days	2.8	3.7	2.9
>14 days	7.5	1.0	0.8
Not born in hospital	1.6	0.6	1.5
Still in hospital	0.05	0.03	0.05

All significant values (p < 0.05) are in bold.

BMI, body mass index; PCOS, polycystic ovary syndrome.

Breastfeeding initiation

Most women (87.9%) reported initiation of any breastfeeding. In univariable analysis, PCOS was associated with increased odds of any breastfeeding initiation (89.9% versus 87.9%; OR: 1.23 [95% confidence interval: 1.02, 1.47]; p = 0.03). After controlling for only BMI, there were still increased odds of any breastfeeding initiation in women with PCOS (OR_adj: 1.3 [1.1, 1.6]; p = 0.005). There was no interaction between PCOS and BMI (p > 0.05), meaning that the odds of any breastfeeding initiation among women with PCOS compared with controls did not differ based on BMI category.

We then performed forward selection to build our multivariable model. All variables were chosen a priori but were added selectively to investigate the individual impact of each variable (Table 2). When variables that represented baseline demographic or medical characteristics were added (BMI, age, depression, anxiety, diabetes, smoking, gestational diabetes, and delivery type), PCOS remained associated with increased odds of any breastfeeding initiation. When variables that had a social dimension (race, ethnicity, maternal education, number of prior live births, pregnancy intent, life stressors, prenatal care, infertility treatment, and length of hospital stay) were investigated one by one, maternal education and number of prior live births were the sole variables to change the PCOS-breastfeeding association from significantly positive to nonsignificant (Table 2). In the full multivariable model, women with PCOS did not have increased odds of any breastfeeding initiation (OR_adj: 1.10 [0.89, 1.37]; p = 0.37).

Table 2.

Odds of Breastfeeding Initiation in Women with PCOS

	Odds ratio (95% confidence interval)	p value
Forward selection
1: PCOS	1.23 (1.02, 1.47)	p = 0.029
2: PCOS, BMI	1.30 (1.08, 1.56)	p = 0.005
3: PCOS, BMI, age	1.24 (1.03, 1.49)	p = 0.023
4: PCOS, BMI, age, depression, anxiety, diabetes, smoking, gestational diabetes, delivery type	1.38 (1.12, 1.70)	p = 0.003
#4 plus race	1.34 (1.08, 1.65)	p = 0.007
#4 plus ethnicity	1.40 (1.13, 1.73)	p = 0.002
#4 plus maternal education	1.20 (0.97, 1.48)	p = 0.096
#4 plus number of prior live births	1.22 (0.99, 1.51)	p = 0.067
#4 plus pregnancy intent	1.31 (1.06, 1.61)	p = 0.013
#4 plus stressors	1.38 (1.12, 1.71)	p = 0.002
#4 plus prenatal care	1.37 (1.11, 1.67)	p = 0.003
#4 plus infertility treatment	1.34 (1.08, 1.65)	p = 0.008
#4 plus length of hospital stay	1.38 (1.12, 1.70)	p = 0.002
Backward selection
Full model^a	1.10 (0.89, 1.37)	p = 0.37
Full model, dropping maternal education	1.17 (0.94, 1.45)	p = 0.17
Full model, dropping number of prior live births	1.14 (0.92, 1.41)	p = 0.24
Full model, dropping maternal education and number of prior live births	1.26 (1.02, 1.56)	p = 0.036

Full model adjusted for age, BMI, education, prior live birth, pregnancy intent, depression, anxiety, diabetes, hypertension, smoking, prenatal care, infertility treatment, stressors, delivery type, gestational diabetes, and length of hospital stay.

All significant values (p < 0.05) are in bold.

BMI, body mass index; PCOS, polycystic ovary syndrome.

We then performed backward selection, removing variables individually and in pairs. We found that removal of both maternal education and number of prior live births had the most impact on the PCOS-breastfeeding association, again suggesting that these were the largest confounders (Table 2).

Table 3 shows the results of the multivariable model. Women with overweight/obesity had decreased odds of any breastfeeding initiation in comparison with normal-weight controls. Moreover, with increasing BMI categories, the odds of any breastfeeding initiation decreased, with the lowest odds seen in those with class III obesity (overweight OR_adj: 0.88 [0.78, 0.99], p = 0.03; class I OR_adj: 0.76 [0.66, 0.88], p < 0.001; class II OR_adj: 0.78 [0.65, 0.94], p = 0.008; class III OR_adj: 0.74 [0.60, 0.91], p = 0.003). Other variables associated with decreased odds of any breastfeeding initiation included age ≤17 years (versus 25–29), Black race (versus White), between 9 and 11 years of education (versus 12), ≥1 prior live births, undesired or unsure pregnancy intent, persistent depression, current smoking, and forceps delivery. Variables associated with increased odds included Alaska, Hawaii native, or Native American race (versus White), Hispanic ethnicity, ≥13 years of education, ≥3 life stressors, and out-of-hospital delivery (Table 3).

Table 3.

Adjusted Odds of Breastfeeding Initiation

	Odds ratio (95% confidence interval)	p value
PCOS
No	1.10 (0.89–1.37)	p = 0.37
Yes	Reference
Maternal age (years)
<17	0.61 (0.43–0.85)	p < 0.001
18–19	0.89 (0.69–1.15)	p = 0.37
20–24	0.96 (0.85–1.09)	p = 0.55
25–29	Reference	Reference
30–34	1.09 (0.97–1.23)	p = 0.14
35–39	1.08 (0.93–1.26)	p = 0.30
40	1.00 (0.76–1.31)	p = 0.99
BMI (kg/m²)
Underweight (<19.8)	0.89 (0.76–1.05)	p = 0.18
Normal (19.8–25)	Reference	Reference
Overweight (25–30)	0.88 (0.78–0.99)	p = 0.03
Obese class I (30–35)	0.76 (0.66–0.88)	p < 0.001
Obese class II (35–40)	0.78 (0.65–0.94)	p = 0.01
Obese class III (>40)	0.74 (0.60–0.91)	p < 0.001
Maternal race
Alaskan, Hawaiian or Native American	1.48 (1.19–1.84)	p < 0.001
Asian	0.95 (0.79–1.14)	p = 0.58
Black	0.83 (0.74–0.93)	p < 0.001
White	Reference	Reference
Other non-White	1.27 (0.99–1.64)	p = 0.06
Mixed race	1.24 (0.98–1.56)	p = 0.07
Maternal ethnicity
Non-Hispanic	Reference	Reference
Hispanic	1.95 (1.67–2.28)	p < 0.001
Maternal education
0–8 years	1.04 (0.81–1.34)	p = 0.75
9–11 years	0.77 (0.67–0.89)	p < 0.001
12 years	Reference	Reference
13–15 years	1.60 (1.43–1.79)	p < 0.001
≥16 years	3.84 (3.32–4.44)	p < 0.001
Number of previous live births
0	Reference	Reference
1	0.63 (0.56–0.71)	p < 0.001
2	0.62 (0.54–0.72)	p < 0.001
3–5	0.59 (0.51–0.70)	p < 0.001
6+	0.69 (0.48–1.00)	p = 0.05
Pregnancy intent
Then	Reference	Reference
Later	1.08 (0.95–1.22)	p = 0.27
Sooner	1.01 (0.87–1.18)	p = 0.87
Was not sure	0.75 (0.66–0.85)	p < 0.001
Did not want then/anytime	0.70 (0.59–0.82)	p < 0.001
History of depression
None	Reference	Reference
Prenatal only	1.20 (0.95–1.50)	p = 0.12
Postnatal onset	1.01 (0.82–1.25)	p = 0.89
Persistent depression	0.81 (0.68–0.96)	p = 0.01
Prenatal history of anxiety
No	Reference	Reference
Yes	0.89 (0.78–1.03)	p = 0.12
History of diabetes
No	Reference	Reference
Yes	0.88 (0.68–1.13)	p = 0.32
Currently smoke
No	Reference	Reference
Yes	0.48 (0.42–0.54)	p < 0.001
Stressors
0	Reference	Reference
1–2	1.02 (0.90–1.15)	p = 0.76
3–4	1.22 (1.05–1.41)	p = 0.01
6+	1.57 (1.25–1.98)	p < 0.001
Prenatal care index (Kotelchuck)
Inadequate	0.93 (0.82–1.07)	p = 0.32
Intermediate	0.97 (0.84–1.13)	p = 0.73
Adequate	Reference	Reference
Adequate plus	1.00 (0.90–1.11)	p = 1.00
Infertility treatment
No	Reference	Reference
Yes	1.10 (0.74–1.63)	p = 0.64
Gestational diabetes
No	Reference	Reference
Yes	1.00 (0.86–1.16)	p = 0.98
Mode of delivery
Vaginal	Reference	Reference
C section	0.91 (0.81–1.01)	p = 0.09
Forceps delivery	0.50 (0.30–0.82)	p = 0.01
Vacuum delivery	0.85 (0.64–1.11)	p = 0.22
Length in hospital
<1 day	0.83 (0.63–1.10)	p = 0.20
1–2 days	1.02 (0.91–1.13)	p = 0.76
3–5 days	Reference	Reference
6–14 days	1.09 (0.86–1.38)	p = 0.49
>14 days	0.72 (0.49–1.06)	p = 0.09
Not born in hospital	2.70 (1.53–4.77)	p < 0.001
Still in hospital	0.37 (0.11–1.26)	p = 0.11

All significant values (p < 0.05) are in bold.

BMI, body mass index; PCOS, polycystic ovary syndrome.

To investigate the impact of race/ethnicity on the relationship between PCOS and any breastfeeding initiation, we included a smaller dataset of the three largest race/ethnic groups (n = 2,114,102): White non-Hispanic (67.2%), Black non-Hispanic (14.3%), and Hispanic (18.5%). Compared with controls, women with PCOS did not have increased odds of breastfeeding in the White non-Hispanic group or Black non-Hispanic group but did have higher odds of breastfeeding in the Hispanic group (OR_adj: 2.0 [1.1, 3.7]; p = 0.03; Table 4).

Table 4.

Odds of Breastfeeding Initiation by Race

	Odds ratio (95% confidence interval)	p value
Unadjusted
PCOS	1.2 (1.0, 1.5)	p = 0.04
Unadjusted
Black non-Hispanic	0.6 (0.6, 0.7)	p < 0.001
Hispanic	1.5 (1.3, 1.7)	p < 0.001
White non-Hispanic	Reference
Adjusted: Model 1^a
PCOS	1.1 (0.9, 1.4)	p = 0.4
Black non-Hispanic	0.8 (0.7, 0.9)	p = 0.002
Hispanic	2.1 (1.8, 2.5)	p < 0.001
White non-Hispanic	Reference
Adjusted: Model 2^b
PCOS versus control stratified by race
Black non-Hispanic	1.3 (0.7, 2.3)	p = 0.3
Hispanic	2.0 (1.1, 3.6)^a	p = 0.03
White non-Hispanic	1.0 (0.8, 1.3)^a	p = 0.9

Model 1 is adjusted for age, BMI, education, prior live birth, pregnancy intent, depression, anxiety, diabetes, hypertension, smoking, prenatal care, infertility treatment, stressors, delivery type, gestational diabetes, and length of hospital stay.

Model 2 is adjusted for the same covariates as Model 1 and includes an interaction term between PCOS and race/ethnicity.

All significant values (p < 0.05) are in bold.

BMI, body mass index; PCOS, polycystic ovary syndrome.

To assess whether there was a social dimension to this difference, we examined sources of breastfeeding information. Black non-Hispanic and Hispanic women were more likely than their White non-Hispanic counterparts to report discussions regarding breastfeeding plans with health care providers and to receive breastfeeding information from family. There was no association with breastfeeding information and PCOS status. The only interaction between race and PCOS was seen in Hispanic subjects: among Hispanic women, those with PCOS were less likely to report being asked about breastfeeding plans than those without PCOS (OR: 0.46 [0.31, 0.70]; p < 0.001).

Short breastfeeding

A quarter (23.1%) of subjects breastfed for ≤1 month. There were no differences in rates of short breastfeeding between PCOS and controls in univariable or multivariable analysis (Table 5).

Table 5.

Length of Breastfeeding

	No PCOS	PCOS	p value	Adjusted analysis (95% confidence interval)^c
Breastfeeding duration ≤1 month (%)^a	76.9	77.8	0.1	OR_adj: 1.0 (0.8, 1.2)
Still breastfeeding at survey collection (%)^a	55.0	55.9	0.5	OR_adj: 1.0 (0.9, 1.2)^d
Mean breastfeeding duration [weeks: mean (SD)]^a,b	4.8 (5.4)	5.1 (5.3)	0.1	HR_adj: 1.0 (0.9–1.1)

Subjects with missing breastfeeding length were excluded.

Among those who had stopped breastfeeding at the time of survey collection.

Adjusted for same variables as Model 1 in Table 4.

Adjusted for survey response time.

HR, hazard ratio; OR, odds ratio; PCOS, polycystic ovary syndrome; SD, standard deviation.

Length of breastfeeding (survival analysis)

Half of participants (54.9%) were still breastfeeding at survey collection. Subjects with missing breastfeeding length (0.4%) were excluded. After adjusting for survey response time, women with PCOS were no more likely to be still breastfeeding (OR_adj: 1.0 [0.9, 1.2]; Table 5). Among those who had already stopped breastfeeding, mean breastfeeding time was 4.8 ± 5.4 weeks. There was no difference in mean breastfeeding weeks in those with PCOS compared with controls (5.1 versus 4.8 weeks, p = 0.11; Table 5). In multivariable Cox models, there was no difference in breastfeeding duration in PCOS women in comparison with controls (hazard ratio [HR]_adj: 1.0 [0.9–1.1]; p = 0.90; Table 5). Again, women with overweight/obesity did have shorter breastfeeding duration in a dose-dependent manner (HR_adj overweight: 1.2; class I: 1.3; class II: 1.5; class III: 1.7, p < 0.001).

Discussion

In this national survey, we observed that women with PCOS had higher rates of any breastfeeding initiation compared with controls; however, this difference was not significant after adjusting for confounders. Furthermore, we found that maternal education and number of prior live births, rather than BMI, were the main confounders that drove this difference, as women with PCOS are more likely to be nulliparous and have higher educational attainment—two factors associated with increased breastfeeding rates. Interestingly, women with PCOS did not have shorter breastfeeding duration in comparison with controls. Our study lends more data to refute early observational studies, which suggested women with PCOS are less likely to breastfeed in the early postpartum period.^21,27 This is encouraging for women with PCOS who hope to breastfeed after pregnancy and for physicians to promote breastfeeding to their PCOS patients. In our study, the medical and obstetric complications commonly seen in women with overweight/obesity, such as cesarean section and diabetes, did not play a major role in any breastfeeding initiation. Our study’s findings reaffirm that decreased initiation of any breastfeeding or shorter duration is related to increasing BMI in a dose-dependent manner.

Some studies have questioned whether insulin resistance could drive the decreased lactogenesis observed in women with overweight/obesity.²⁸ Given that women with PCOS have increased rates of impaired glucose tolerance and diabetes even after adjusting for BMI,²⁹ it would suggest that they would be at increased risk of decreased initiation of any breastfeeding or shorter duration. However, this was not found by our study. Our data are supported by a recent Australian cross-sectional study of 4,898 women (6.5% with PCOS), aged 31–36 years, from the Australian Longitudinal Study on Individual’s Health. They showed that PCOS was not associated with a decreased initiation of any breastfeeding but was associated with increased odds of short breastfeeding duration (≤6 months). This association was not significant after controlling for confounders (BMI) and socioeconomic variables (education and occupation). Unlike our study, they did not investigate these confounders individually with PCOS, thus which confounders had the largest contribution to their analysis is unknown.²⁴ A case–control study of 401 women presenting to a breastfeeding clinic found that PCOS was not associated with low milk production. However, women with diabetes in pregnancy did have increased odds of decreased milk production in both adjusted and unadjusted analyses.²²

Prior studies have shown decreased breastfeeding rates in Black non-Hispanic women compared with White non-Hispanic women and have highlighted the impact of systemic racism on breastfeeding rates.³⁰ Our study confirmed that Black non-Hispanic women had lower breastfeeding rates than their White non-Hispanic counterparts after adjusting for confounders. However, there was no impact on PCOS status on this relationship. Interestingly, in the overall group, those who identified as Hispanic had the highest breastfeeding rates, and this was more pronounced in those with PCOS. Hispanic women with PCOS had almost two-fold higher odds of any breastfeeding initiation compared with Hispanic women without PCOS. Prior studies have confirmed higher rates of breastfeeding in Hispanic subjects even after adjusting for maternal factors including socioeconomic status.³¹ It is likely that the impact of Hispanic race on breastfeeding in those with PCOS is related to unmeasured social or systemic confounders.

Clinical implications

In addition to an increased risk of overweight/obesity, women with PCOS have higher rates of hypertension, diabetes, and dyslipidemia.^32,33 One study showed that women with PCOS who experienced no breast growth during pregnancy had not only low milk production but also increased metabolic disturbances in early pregnancy (higher blood pressure, fasting insulin, and triglyceride levels) and were more likely to be obese in both early and late pregnancy.³⁴

Prolonged lactation has been shown to lower blood glucose and insulin concentrations, increase rates of glucose production and lipolysis, and improve metabolic profiles and insulin sensitivity.³⁵ These factors may contribute to the reduced diabetes risk after a pregnancy complicated by GDM^28,35 and a significant reduction of postpartum weight retention in all prepregnancy BMI categories.^36,37 Given the associations between PCOS and metabolic disturbances and the improvement of these metabolic abnormalities with breastfeeding, women with PCOS could benefit from early and prolonged breastfeeding to improve their future health.

Research implications

Although our study adds to the literature on PCOS and breastfeeding initiation and duration, these are only two factors that contribute to long-term breastfeeding success. Future research should look more closely at breastfeeding intention, timing of onset of lactogenesis, and ease/difficulty of breastfeeding in PCOS, in particular their association with insulin resistance and hyperandrogenism.

Strengths and Limitations

To our knowledge, this study is the first of its kind to use U.S. national data to evaluate breastfeeding initiation rates and breastfeeding duration in women with PCOS. Strengths of this study include a large population with representation from all regions of the United States, increasing generalizability and careful evaluation of demographic characteristics, including race, that could influence breastfeeding. Given that we found the expected baseline comorbidities in women with PCOS and similar risk factors for decreased breastfeeding as other studies, it is likely a representative sample of women with PCOS in the United States. Additionally, there was minimal missing data on PCOS status.

One limitation is self-reported PCOS status. However, genome-wide association studies confirm similar genetic architecture in self-reported PCOS and PCOS diagnosed by National Institutes of Health (NIH) or Rotterdam criteria.³⁸ In addition, other studies have confirmed the association between self-reported PCOS and increased risks of metabolic comorbidities.^39,40 We were unable to determine if different PCOS phenotypes have different risks. Another significant limitation is that the PRAMS dataset measures “any breastfeeding” without differentiating between exclusive breastfeeding and breastfeeding with supplementation. This lack of differentiation could obscure important trends related to breastfeeding exclusivity and duration associated with PCOS. Therefore, we cannot assess whether PCOS impacts the exclusivity of breastfeeding or the need for supplementation. Additionally, the timing of the PRAMS questionnaire with a median response time of 3.7 months postpartum limits our ability to evaluate women who breastfed longer. We also did not measure rates of insufficient milk production, so although there was no association between PCOS and either breastfeeding initiation or duration, we cannot evaluate whether impaired milk production was associated with PCOS.

Conclusion

As PCOS status is not associated with reduced initiation of any breastfeeding or shorter duration, our data support the fact that interventions aimed at increasing breastfeeding should be targeted at populations that would benefit the most and that PCOS-specific targeting is not needed. However, many women with PCOS are overweight/obese, and there is a clear association between increased BMI and decreased breastfeeding. Therefore, we advocate for increased breastfeeding support and education for women with overweight/obesity and additional prospective studies to fully understand the association between different PCOS phenotypes and breastfeeding.

Footnotes

Acknowledgments

The authors would like to acknowledge the Pregnancy Risk Assessment Monitoring System Working Group and the CDC for providing the data used in the analysis reported.

Authors’ Contributions

L.G.C. had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. L.G.C.: Concept and design, obtained funding, administrative, technical, or material support, and supervision. L.G.C., L.M.B., and J.Z.: Acquisition, analysis, or interpretation of data and statistical analysis. L.G.C. and L.M.B.: Drafting of the article. L.G.C., L.M.B., and A.E.: Critical review of the article for important intellectual content.

PRAMS Working Group

The PRAMS Working Group members include Izza Afgan, MPH (Alabama); Kathy Perham-Hester, MS, MPH (Alaska); Ruby D. Brown (Arkansas); Ashley Juhl, MSPH (Colorado); Jennifer Morin, MPH (Connecticut); George Yocher, MS (Delaware); Elizabeth C. Stewart, MSPH (Florida); Florence A. Kanu, MPH (Georgia); Wendy Nihoa, MPH (Hawaii); Julie Doetsch, MA (Illinois); Jessica Eagan (Iowa); Rosaria Trichilo, MPH (Louisiana); Tom Patenaude, MPH (Maine); Laurie Kettinger, MS (Maryland); Emily Lu, MPH (Massachusetts); Peterson Haak (Michigan); Mira Sheff (Minnesota); Brenda Hughes, MPPA (Mississippi); David McBride, PhD (Missouri); Emily Healy, MS (Montana); Jessica Seberger (Nebraska); Sara Riordan, MEd (New Hampshire); Sharon Smith Cooley, MPH (New Jersey); Eirian Coronado (New Mexico); Anne Radigan (New York State); Pricila Mullachery, MPH (New York City); Kathleen Jones-Vessey, MS (North Carolina); Sandra Anseth (North Dakota); Connie Geidenberger, PhD (Ohio); Ayesha Lampkins, MPH, CHES (Oklahoma); Tina Kent (Oregon); Sara Thuma, MPH (Pennsylvania); Karine Tolentino Monteiro, MPH (Rhode Island); Kristen Simpson (South Carolina); Uvonne Leverett (Tennessee); Tanya Guthrie, PhD (Texas); Barbara Algarin (Utah); Peggy Brozicevic (Vermont); Kenesha Smith (Virginia); Linda Lohdefinck (Washington); Melissa Baker, MA (West Virginia); Christopher Huard (Wisconsin); Lorie Chesnut, PhD (Wyoming); and the CDC PRAMS Team, Individual’s Health and Fertility Branch, Division of Reproductive Health.

Data Sharing Statement

The PRAMS data used for this study were provided by the CDC. The authors are not allowed to share the data according to the signed External Researcher Data Sharing Agreement with the CDC. Researchers who are interested in using the PRAMS data can submit a request to the CDC staff. More details on access to the PRAM data can be found on the website ().

Disclaimer

The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH.

Author Disclosure Statement

No competing financial interests exist.

Funding Information

The project was supported by the Clinical and Translational Science Award program, through the NIH National Center for Advancing Translational Sciences, grant UL1TR002373. Research support was provided by an award from the University of Wisconsin School of Medicine and Public Health Department of Obstetrics and Gynecology and the Shapiro Summer Research Program. The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the article; and decision to submit the article for publication.

References

US Department of Health and Human Services: Center of Disease Control. CDC’S Work to Support & Promote Breastfeeding in Hospitals, Worksites, and Communities. cdc.gov. Available from: https://www.cdc.gov/breastfeeding/about-breastfeeding/index.html

American College of Obstetricians and Gynecologists. Optimizing Support for Breastfeeding as Part of Obstetric Practice; 2018. Available from: https://www.acog.org/,https://www.acog.org/clinical/clinical-guidance/committee-opinion/articles/2018/10/optimizing-support-for-breastfeeding-as-part-of-obstetric-practice

Committee ABA. Breastfeeding, Family Physicians Supporting (Position Paper). American Academy of Family Medicine; 2015. Available from: https://www.aafp.org/about/policies/all/breastfeeding-position-paper.html

Meek

, Noble

, Feldman-Winter

, et al. Technical Report: Breastfeeding and the Use of Human Milk. Pediatrics, 2022; 150(1); doi: 10.1542/PEDS.2022-057989

Bernardo

, Cesar

, Organization

. Long-term effects of breastfeeding: A systematic review. Geneva: World Health Organization. Available from: https://apps.who.int/iris/handle/10665/79198

Hui

, Kwok

, Nelson

EAS

, et al. Breastfeeding in infancy and lipid profile in adolescence. Pediatrics, 2019; 143(5):e20183075.

Amitay

, Keinan-Boker

. Breastfeeding and Childhood Leukemia Incidence: A Meta-analysis and Systematic Review. JAMA Pediatr, 2015; 169(6):e151025–e151025.

Thompson

JMD

, Tanabe

, Moon

, et al. Duration of breastfeeding and risk of SIDS: An individual participant data meta-analysis. Pediatrics, 2017; 140(5):e20171324.

Miliku

, Moraes

, Becker

, et al. Breastfeeding in the first days of life is associated with lower blood pressure at 3 years of age. J Am Heart Assoc, 2021; 10:e019067.

10.

Schwarz

, Brown

, Creasman

, et al. Lactation and maternal risk of type 2 diabetes: A population-based study. Am J Med, 2010; 123(9):863.e1–6–863.e6.

11.

Breast cancer and breastfeeding: Collaborative reanalysis of individual data from 47 epidemiological studies in 30 countries, including 50302 women with breast cancer and 96973 women without the disease. Lancet, 2002; 360:187–195.

12.

Nguyen

, Jin

, Ding

. Breastfeeding and maternal cardiovascular risk factors and outcomes: A systematic review. PLoS One, 2017; 12(11):e0187923–e0187923.

13.

Rameez

, Sadana

, Kaur

, et al. Association of maternal lactation with diabetes and hypertension: A systematic review and meta-analysis. JAMA Netw Open, 2019; 2(10):e1913401–e1913401.

14.

Ajmera

, Terrault

, VanWagner

, et al. Longer lactation duration is associated with decreased prevalence of non-alcoholic fatty liver disease in women. J Hepatol, 2019; 70(1):126–132.

15.

Bozdag

, Mumusoglu

, Zengin

, et al. The prevalence and phenotypic features of polycystic ovary syndrome: A systematic review and meta-analysis. Hum Reprod, 2016; 31(12):2841–2855.

16.

Revised 2003 consensus on diagnostic criteria and long-term health risks related to polycystic ovary syndrome. Fertil Steril, 2004; 81:19–25.

17.

Moran

, Brinkworth

, Norman

. Dietary therapy in polycystic ovary syndrome. Semin Reprod Med, 2008; 26(1):85–92.

18.

Lim

, Davies

, Norman

, et al. Overweight, obesity and central obesity in women with polycystic ovary syndrome: A systematic review and meta-analysis. Hum Reprod Update, 2012; 18(6):618–637.

19.

Turcksin

, Bel

, Galjaard

, et al. Maternal obesity and breastfeeding intention, initiation, intensity and duration: A systematic review. Matern Child Nutr, 2014; 10(2):166–183; doi: 10.1111/j.1740-8709.2012.00439.x

20.

Nommsen-Rivers

. Does insulin explain the relation between maternal obesity and poor lactation outcomes? An overview of the literature. Adv Nutr, 2016; 7(2):407–414.

21.

Vanky

, Isaksen

, Haase Moen

, et al. Breastfeeding in polycystic ovary syndrome. Acta Obstet Gynecol Scand, 2008; 87(5):531–535.

22.

Riddle

, Nommsen-Rivers

. A case control study of diabetes during pregnancy and low milk supply. Breastfeed Med, 2016; 11(2):80–85.

23.

Carlsen

, Jacobsen

, Vanky

. Mid-pregnancy androgen levels are negatively associated with breastfeeding. Acta Obstet Gynecol Scand, 2010; 89(1):87–94; doi: 10.3109/00016340903318006

24.

Joham

, Nanayakkara

, Ranasinha

, et al. Obesity, polycystic ovary syndrome and breastfeeding: An observational study. Acta Obstet Gynecol Scand, 2016; 95(4):458–466.

25.

Shulman

, D’Angelo

, Harrison

, et al. The Pregnancy Risk Assessment Monitoring System (PRAMS): Overview of design and methodology. Am J Public Health, 2018; 108(10):1305–1313.

26.

Centers for Disease Control and Prevention. 2016 PRAMS Response Rate Table. CDC; 2019. Available from: https://www.cdc.gov/prams/prams-data/response-rate-tables/2016-response-rate-table.html

27.

Marasco

, Marmet

, Shell

. Polycystic ovary syndrome: A connection to insufficient milk supply? J Hum Lact, 2000; 16(2):143–148.

28.

Nommsen-Rivers

, Dolan

, Huang

. Timing of stage II lactogenesis is predicted by antenatal metabolic health in a cohort of primiparas. Breastfeed Med, 2012; 7(1):43–49.

29.

Kakoly

, Khomami

, Joham

, et al. Ethnicity, obesity and the prevalence of impaired glucose tolerance and type 2 diabetes in PCOS: A systematic review and meta-regression. Hum Reprod Update, 2018; 24(4):455–467.

30.

Robinson

, Fial

, Hanson

. Racism, bias, and discrimination as modifiable barriers to breastfeeding for African American women: A scoping review of the literature. J Midwifery Womens Health, 2019; 64(6):734–742.

31.

Fryer

, Santos

, Pedersen

, et al. The hispanic paradox: Socioeconomic factors and race/ethnicity in breastfeeding outcomes. Breastfeed Med, 2018; 13(3):174–180.

32.

Moran

, Misso

, Wild

, et al. Impaired glucose tolerance, type 2 diabetes and metabolic syndrome in polycystic ovary syndrome: A systematic review and meta-analysis. Hum Reprod Update, 2010; 16(4):347–363; doi: 10.1093/humupd/dmq001

33.

Behboudi-Gandevani

, Amiri

, Bidhendi Yarandi

, et al. The risk of metabolic syndrome in polycystic ovary syndrome: A systematic review and meta-analysis. Clin Endocrinol (Oxf), 2018; 88(2):169–184; doi: 10.1111/cen.13477

34.

Vanky

, Nordskar

, Leithe

, et al. Breast size increment during pregnancy and breastfeeding in mothers with polycystic ovary syndrome: A follow-up study of a randomised controlled trial on metformin versus placebo. BJOG, 2012; 119(11):1403–1409.

35.

Gunderson

, Hedderson

, Chiang

, et al. Lactation intensity and postpartum maternal glucose tolerance and insulin resistance in women with recent GDM: The SWIFT cohort. Diabetes Care, 2012; 35(1):50–56.

36.

Baker

, Gamborg

, Heitmann

, et al. Breastfeeding reduces postpartum weight retention. Am J Clin Nutr, 2008; 88(6):1543–1551.

37.

Dewey

, Heinig

, Nommsen

. Maternal weight-loss patterns during prolonged lactation. Am J Clin Nutr, 1993; 58(2):162–166.

38.

Day

, Karaderi

, Jones

, et al.; 23andMe Research Team. Large-scale genome-wide meta-analysis of polycystic ovary syndrome suggests shared genetic architecture for different diagnosis criteria. PLoS Genet, 2018; 14(12):e1007813.

39.

Ollila

M-ME

, Kaikkonen

, Järvelin

M-R

, et al. Self-reported polycystic ovary syndrome is associated with hypertension: A Northern Finland Birth Cohort 1966 Study. J Clin Endocrinol Metab, 2019; 104(4):1221–1231.

40.

West

, Ollila

M-M

, Franks

, et al. Overweight, obesity and hyperandrogenemia are associated with gestational diabetes mellitus: A follow-up cohort study. Acta Obstet Gynecol Scand, 2020; 99(10):1311–1319.