ORIGINAL ARTICLE

Contemporary Reproductive Outcomes for Patients

With Polycystic Ovary Syndrome: A Retrospective
Observational Study

D. Aled Rees, Sara Jenkins-Jones, and Christopher L. Morgan

Pharmatelligence (S.J.-J.), Cardiff Medicentre, Heath Park, Cardiff CF14 4UJ, United Kingdom; and
Institute of Primary Care and Public Health (C.L.M.) and Neurosciences and Mental Health Research

Downloaded from https://academic.oup.com/jcem/article/101/4/1664/2804548 by guest on 24 March 2024

Institute (D.A.R.), School of Medicine, Cardiff University, Cardiff CF24 4HQ, United Kingdom

Context: Polycystic ovary syndrome (PCOS) is the most common cause of anovulatory infertility and
may be associated with adverse pregnancy and neonatal outcomes. However, it is difficult to
establish how much of this risk is due to PCOS and how much to obesity.

Objective: This study aimed to determine the effect of PCOS upon fertility, pregnancy, and neo-
natal outcomes.

Design and Setting: Data were extracted from the Clinical Practice Research Datalink (CPRD), a
longitudinal anonymized primary care research database in the United Kingdom. Patients with a
diagnosis of PCOS were matched to controls (1:2) by age (⫾1 y), body mass index (⫾ 3 U), and CPRD
practice. Standardized fertility ratios before and after diagnosis (index date) were calculated. Rates
of miscarriage, pre-eclampsia, gestational diabetes, premature delivery, delivery method, and
neonatal outcomes were compared.

Results: Nine thousand sixty-eight women with PCOS matched study criteria. Prior to index date
the standardized fertility ratio for patients with PCOS was 0.80 (95% confidence interval, 0.77–
0.83); following index date it was 1.16 (1.12–1.20). The adjusted odds ratios (95% CI) for miscarriage
(1.70; 1.56 –1.84), pre-eclampsia (1.32; 1.16 –1.49), gestational diabetes (1.41; 1.2–1.66), and pre-
mature delivery (1.25; 1.1–1.43) were all increased compared with controls. Of PCOS births, 27.7%
were by Caesarean section compared with 23.7% of controls (1.13; 1.05–1.21). Infants born to
mothers with PCOS had an increased risk of neonatal jaundice (1.20; 1.03–1.39) and respiratory
complications (1.20; 1.06 –1.37).

Conclusions: PCOS is associated with subfertility but fertility rates are restored to those of the
background population following diagnosis. Pregnancy complications and adverse neonatal out-
comes are more prevalent for women with PCOS independently of obesity. (J Clin Endocrinol Metab
101: 1664 –1672, 2016)

olycystic ovary syndrome (PCOS) is the most common women require assisted reproductive techniques to help
P endocrine condition in women of reproductive age,
affecting 5–10% of the premenopausal population (1, 2).
them conceive. Most commonly, this involves the use of
clomiphene citrate as an agent to induce ovulation. How-
The disorder is characterized by hyperandrogenism, insu- ever, two randomized, double blind, placebo-controlled
lin resistance, and glucose intolerance, which lead to an trials have also established that metformin may improve
increased risk of type 2 diabetes (3). PCOS is also a com- ovulation and conception rates in clomiphene-resistant
mon cause of anovulatory infertility (4, 5), such that many women with PCOS (6, 7). Previous studies have shown

ISSN Print 0021-972X ISSN Online 1945-7197 Abbreviations: BMI, body mass index; CI, confidence interval; CPRD, Clinical Practice Re-
Printed in USA search Datalink; GDM, gestational diabetes; HES, Hospital Episode Statistics; ICD, 10th
Copyright © 2016 by the Endocrine Society revision of the International Statistical Classification of Diseases and Related Health Prob-
Received June 25, 2015. Accepted February 2, 2016. lems; OR, odds ratio; PCOS, polycystic ovary syndrome.
First Published Online February 9, 2016

1664 press.endocrine.org/journal/jcem J Clin Endocrinol Metab, April 2016, 101(4):1664 –1672 doi: 10.1210/jc.2015-2682
doi: 10.1210/jc.2015-2682 press.endocrine.org/journal/jcem 1665

that the reproductive effects of the syndrome may also extend ized fertility ratios based on 10-year age bands using the non-
to a higher risk of many adverse outcomes in pregnancy, of PCOS population as the reference.
Miscarriages resulting in hospital admission were identified
which gestational diabetes (GDM), gestational hyperten-
through secondary care sources. Crude relative risk of miscar-
sion, premature birth, and early pregnancy loss are the most riage was calculated. Multivariate logistic regression predicting
studied (4, 5, 8, 9). Infants born to mothers with PCOS may miscarriage vs delivery was also performed adjusting for age,
also be at risk of adverse perinatal outcomes, including mac- BMI, number of previous births, and smoking history.
rosomia, low Apgar score, and meconium aspiration (9). Complications of pregnancy were defined as premature birth,
However, estimates of risk for these complications are dif- GDM, and pre-eclampsia, and were ascertained from both the
primary and HES data sources (Supplemental Table 3). Com-
ficult to ascertain due to heterogeneity in study design, small
plications were assigned to each delivery if they were recorded in
sample sizes, and inadequate matching for potential con- the preceding 295 days. The relative risks of each complication
founders such as obesity. In light of these uncertainties, we in the period, both before and after diagnosis of PCOS, were

Downloaded from https://academic.oup.com/jcem/article/101/4/1664/2804548 by guest on 24 March 2024

sought to describe the fertility outcomes and establish the calculated. In addition, the odds ratio (OR) was calculated using
relative risk of adverse pregnancy and neonatal outcomes in multivariate logistic regression adjusting for age, BMI, multiple
a large contemporary cohort of patients with PCOS in the gestation, number of previous births, and smoking history.
To consider the effect of metformin treatment on pregnancy
United Kingdom.
outcomes, date of conception was estimated as 280 days prior to
delivery date for those pregnancies resulting in live birth. For the
miscarriage outcome, it was assumed that conception occurred
Materials and Methods 84 days prior to the event. Metformin use was assumed based on
prescription at two defined time points: 1) 90 days prior to es-
The study used a retrospective cohort design using data from the timated conception date and, 2) within the estimated first tri-
Clinical Practice Research Database (CPRD), a longitudinal, mester of the pregnancy.
anonymized research database derived from more than 700 pri- Hospital births were identified from the HES dataset. Deliv-
mary-care practices in the United Kingdom (10). Diagnostic in- ery method was defined by procedural codes (OPCS-4) and clas-
formation in CPRD is recorded using the Read code classifica- sified as normal vaginal delivery, elective caesarean, emergency
tion, a United Kingdom general practice standard. In addition, caesarean, forceps delivery, and vacuum delivery (Supplemental
approximately 60% of those practices are linked to other English Table 4). Method of delivery and birth outcome was described
data sources including the Hospital Episode Statistics (HES) da- and compared between cohorts using the ␹2 test. In addition,
taset. This provides data on all inpatient and outpatient contacts multivariate logistic regression was used to examine the associ-
occurring within National Health Trusts. ation of PCOS status with caesarean compared with vaginal de-
livery and the likelihood of twin delivery. Additional covariates
Patient selection and matching of controls included in the model were age, BMI, GDM, history of pre-
Women age 15– 44 years with a diagnosis of PCOS defined by eclampsia, and number of previous births. Length of stay, ag-
the Read code classification (Supplemental Table 1) or 10th re- gregated by delivery method, was compared using the t test.
vision of the International Statistical Classification of Diseases Using the family number within CPRD, it was possible to link
and Related Health Problems (ICD-10) classification (E28.2) females with their children for a proportion of subjects. This
between 2000 and 2012 were selected and defined as cases. The allowed for diagnoses relating to neonatal admissions (defined as
earliest diagnosis date was selected as the index date. A minimum those occurring between birth and 7 d) to be compared. We
“wash-in” period of 12 months from the patient’s practice reg- selected the six most common groups of related complications on
istration date to index date was used to maximize the likelihood the neonate’s inpatient record: jaundice (ICD-10 P580 –P599),
that the case represented an incident case. Patients were followed respiratory (P200-P229, P285, P288, and P289), feeding issues
until either leaving their practice or until the last CPRD collection (P920 –P929), overweight (P080, P081), low birth weight
data for their practice. Female controls were matched at a ratio (P050 –P059, P070 –P072), and hypoglycemia (P703, P704).
of 2:1 by age (⫾1 y), body mass index (BMI) (⫾ 3 U), and same Analyses were adjusted for age, BMI, multiple gestation, number
primary care practice. All controls had to have remained at the of previous births, and smoking history.
same practice for at least the same duration from index date as Studies using the CPRD are covered by ethics approval
their respective case, and followup was limited in both groups to granted by Trent Multicenter Research Ethics Committee (Ref-
that of the case. erence 05/MRE04/87). CPRD Independent Scientific Advisory
Committee approval was granted for this study (ISAC 13–192).
Outcomes and analysis
Baseline characteristics for cases and controls were presented
and compared using univariate statistics (t test for continuous Results
variables and ␹2 for categorical variables).
Infertility was defined by relevant Read code (Supplemental Figure 1 shows the identification of PCOS patients, of
Table 2) recorded for consultations within the primary care da-
whom 9068 could be matched to two non-PCOS controls.
taset. To assess the relationship between PCOS and infertility,
time between first diagnosis of PCOS and first consultation for The baseline characteristics of these patients are shown in
infertility was described graphically and by summary statistics. Table 1. There were significant differences between the
Fertility was compared between the two cohorts using standard- PCOS and control cohorts for the number of primary care
1666 Rees et al Fertility, Pregnancy and Neonatal Outcomes in PCOS J Clin Endocrinol Metab, April 2016, 101(4):1664 –1672

Age-specific fertility rates

During the study period (before and after index date)
the overall fertility rate for women with PCOS was 67.3
per 1000 years compared with 70.6 for those without. The
respective fertility rates pre- and postindex dates were
57.5 vs 71.8 and 79.8 vs 68.7. The age standardized ratios
were 0.80 (95% CI, 0.77– 0.83) prior to index date and
1.16 (95% CI, 1.12–1.20) after. Figure 2 (lower panel)
shows the age-specific fertility rates by PCOS status before
and after the index date.

Downloaded from https://academic.oup.com/jcem/article/101/4/1664/2804548 by guest on 24 March 2024

Pregnancy risks
There were 6861 pregnancies resulting in hospital ad-
mission for miscarriage or delivery for women with PCOS
and 15 214 for those without (Table 2). Risk of miscar-
riage was increased for women with PCOS; the overall
crude risk was 1.56 (95% CI, 1.45–1.68): 1.77 (1.61–
1.95) before diagnosis and 1.35 (1.21–1.51) after. After
adjusting for age, BMI, smoking status, and prior births,
the OR was 1.70 (1.56 –1.84). The respective ORs before
and after index date were 1.99 (1.79 –2.24) and 1.47
Figure 1. Patient selection.
(1.29 –1.67) (Figure 3).
During those pregnancies resulting in live birth, the
contacts (9.0 vs 7.6, respectively) in the previous year. There overall crude risk of pre-eclampsia was 1.31 (1.16 –1.46).
was considerable missing data for ethnicity, although this In the adjusted analyses, the OR was 1.31 (1.16 –1.49);
was lower for patients with PCOS compared with controls 1.23 (1.03–1.46) prior to diagnosis and 1.40 (1.17–1.68)
(33.0 vs 36.9%). There was a larger proportion of patients of afterward. The overall crude risk of GDM was 1.43 (1.23–
Indian, Pakistani, or Other Asian origin compared with con- 1.67). In the adjusted analyses the OR was 1.42 (1.21–
trols. Patients with PCOS were less likely to have ever 1.67); 1.21 (0.94 –1.54) prior to diagnosis and 1.61 (1.29 –
smoked tobacco products (56.9 vs 52.9%). 2.01) afterward. Premature delivery was also increased for
women with PCOS. The overall crude risk was 1.27 (1.13–
Infertility consultations
1.44). After adjusting for other factors, the OR was 1.24
Of women identified with PCOS, 1529 (16.9%) had
(1.09 –1.41); 1.24 (1.03–1.49) prior to diagnosis and 1.26
previously consulted their primary care practice regarding
(1.05–1.52) afterward (Figure 3). In a sensitivity analysis
issues of infertility compared with 800 (4.4%) of controls;
restricted to singleton births there was little difference in
a crude rate ratio of 4.69 (95% confidence interval [CI],
the adjusted ORs compared with the main analysis (Sup-
4.30 –5.11). Respective figures following PCOS index
plemental Table 6). In a further sensitivity analysis re-
date were 796 (8.8%) vs 496 (2.7%); a crude rate ratio of
3.59 (3.21– 4.02). Figure 2 (upper panel) shows time from stricted to the 4355 patients with a diagnosis of either
index date to first consultation for infertility. Four-hun- PCOS or Stein-Leventhal syndrome and their respective
dred twenty-three (18.2%) women with PCOS consulting controls, slightly higher adjusted ORs were observed for
their primary care practice for issues relating to infertility all outcomes (Supplemental Figure 1).
did so within ⫾ 90 days of first PCOS diagnosis.
Of patients with PCOS, 413 (4.6%) had a history of Effect of metformin
assisted reproduction compared with 188 (1.0%) of con- For the analysis restricted to PCOS patients treated
trols. Following index date, 507 (5.6%) of cases had a with metformin relative to other nonmetformin treated
record associated with assisted reproduction compared patients, there was a significant increase in pre-eclampsia
with 110 (0.6%) of controls. Subsequent to assisted re- associated with metformin use in both the 90 days prior to
production, 286 (56.4%) of cases and 41 (37.3%) of con- estimated conception (OR, 1.54; 95% CI, 1.06 –2.23) and
trols had conceptions resulting in either hospital birth or the first trimester of pregnancy (OR, 1.54; 95% CI, 1.07–
miscarriage. Details of the fertility treatments provided are 2.22) but no significant difference in any other pregnancy
given in Supplemental Table 5. related outcomes (Supplemental Figure 2).
doi: 10.1210/jc.2015-2682 press.endocrine.org/journal/jcem 1667

Table 1. Baseline Characteristics for Patients With PCOS and Controls

Study Arm

Baseline Characteristic PCOS Non-PCOS P

No. of patients 9068 18 136
Followup (y), mean, SD 4.2 3.3 4.2 3.3 1.000
Observation period pre-index (y), mean, SD 5.3 3.7 6.5 3.8 ⬍.001
Age (y), mean, SD 27.3 6.4 27.3 6.4 .808
Ethnicity
Bangladeshi, n, % 27 0.3 60 0.3 ⬍.001
Black (African/Caribbean/Other), n, % 174 1.9 331 1.8
Chinese, n, % 13 0.1 22 0.1
Indian, n, % 173 1.9 173 1.0

Downloaded from https://academic.oup.com/jcem/article/101/4/1664/2804548 by guest on 24 March 2024

Mixed, n, % 52 0.6 100 0.6
Other Asian, n, % 77 0.8 83 0.5
Other, n, % 103 1.1 202 1.1
Pakistani, n, % 143 1.6 160 0.9
Unknown/not recorded, n, % 2990 33.0 6701 36.9
White, n, % 5316 58.6 10 304 56.8
Primary care contacts in previous year 9.0 7.8 7.6 7.0 ⬍.001
Prescribed oral contraception ⫾ 90 d of index date, n, % 2712 29.9 6508 35.9 ⬍.001
History of assisted reproduction, n, % 413 4.6 188 1.0 ⬍.001
BMI (kg/m2), mean, SD 27.7 6.4 27.5 6.3 .019
Underweight (⬍ 18 kg/m2), n, % 187 2.1 283 1.6 ⬍.001
Normal (18 –25 kg/m2), n, % 3543 39.1 7581 41.8
Overweight (⬎ 25–30 kg/m2), n, % 2330 25.7 4543 25.0
Obese (⬎ 30 – 40 kg/m2), n, % 2637 29.1 5013 27.6
Extremely obese (⬎ 40 kg/m2), n, % 371 4.1 716 3.9
BP
No. with BP recorded, n, % 5378 59.3 15 529 85.6
Systolic BP (mm Hg), mean, SD 118.7 13.5 118.3 13.0 .046
Diastolic BP (mm Hg), mean, SD 74.8 9.7 73.5 9.6 ⬍.001
Hypertensive, n, % 362 6.7 879 5.7 .001
Smoking status ⬍.001
Never smoked, n, % 5159 56.9 9590 52.9
Ex-smoker, n, % 1511 16.7 3439 19.0
Current smoker, n, % 2364 26.1 5065 27.9
Not recorded, n, % 34 0.4 42 0.2
Abbreviation: BP, blood pressure.

Delivery methods births, an increased relative rate for mothers with PCOS
Supplemental Table 7 shows the method of delivery for (OR, 1.54; 95% CI, 1.22–1.95).
women with and without PCOS. Of PCOS births, 1606 Apgar scores (11) were available for 1731 (30.1%) chil-
(27.7%) were by caesarean section compared with 3243 dren born to mothers with PCOS and 4217 (31.0%) born
(23.7%) of non-PCOS. In logistic regression, the OR of to mothers without PCOS. Respective Apgar scores were
caesarean delivery for PCOS was 1.13 (1.05–1.21) after 6.33 and 6.43 at 1 minute and 6.97 and 7.02 at 5 minutes,
adjustment for other covariates including pre-eclampsia and were not significantly different between groups.
(2.54; 2.14 –3.00) and GDM (2.63; 2.25–3.06). Mean Neonatal inpatient admissions records for children
length of stay for delivery was significantly greater for could be linked to 3634 (62.7%) of births for women with
women with PCOS (3.8 vs 3.5 d; P ⫽ .002; Supplemental PCOS and 8557 (62.4%) of those without. Table 3 shows
Table 7). After adjustment for age, BMI, GDM, pre-ec- the number of defined neonatal complications recorded on
lampsia, and premature delivery, the respective means these admissions. There were significant increases for
were 3.7 vs 3.5 days (P ⬍ .001). jaundice (1.20; 1.03–1.39) and respiratory complications
(1.20; 1.06 –1.37).
Neonatal outcomes
Delivery outcome was available for 5757 (99.3%) Discussion
PCOS births and 13 607 (99.3%) non-PCOS births (Sup-
plemental Table 8). Multiple births occurred in 146 In this large population-based study, a diagnosis of PCOS
(2.5%) of PCOS births and 226 (1.7%) of non-PCOS was associated with a lower fertility rate, increased risk of
1668 Rees et al Fertility, Pregnancy and Neonatal Outcomes in PCOS J Clin Endocrinol Metab, April 2016, 101(4):1664 –1672

Downloaded from https://academic.oup.com/jcem/article/101/4/1664/2804548 by guest on 24 March 2024

Figure 2. Upper panel: Time from index date (first diagnosis of PCOS) to consultation for infertility. Left upper panel: PCOS cases. Right upper
panel: Non-PCOS cases. Lower panel: Standardized specific fertility rates for women with and without PCOS, before and after index date.

adverse pregnancy and neonatal outcomes, and of oper- phenotype presenting at this age, but is also consistent
ative delivery, which were not attributable to obesity. with a trend toward regularization of the menstrual cycle
We observed an approximate 4-fold increase in con- in women with PCOS with advancing age (12). After di-
sultations for infertility in women with PCOS compared agnosis, fertility rates were restored in all age groups to
with matched controls. This is reflected in lower fertility those of the background population, suggesting that in-
rates for women with PCOS prior to diagnosis, particu- fertility in women with PCOS is eminently treatable. How-
larly in younger patients. This may reflect a more severe ever, it is not possible from the data to compare the in-
doi: 10.1210/jc.2015-2682 press.endocrine.org/journal/jcem 1669

Table 2. Hospital Admissions for Births, Miscarriages, study adjusted for obesity and fertility treatment (13)
and Complications of Pregnancy for Women With and which are themselves known to be associated with an in-
Without PCOS creased risk (18). In our study, which matched patients for
BMI, PCOS was associated with a significant increase in
PCOS Non-PCOS
pregnancy loss both before and after PCOS diagnosis.
n % n % For those pregnancies resulting in live birth there was
Before index date also significant increases in pre-eclampsia, GDM, and pre-
Identified pregnancies 3369 9437 mature delivery. Previous studies have suggested that the
Births 2778 82.5 8503 90.1 risk of GDM is increased in women with PCOS, but in-
GDM 93 3.3 235 2.8
Pre-eclampsia 188 6.8 466 5.5 terpretation of these studies and estimates of risk are dif-
Premature 171 6.2 427 5.0 ficult because of differences in study design and small sam-

Downloaded from https://academic.oup.com/jcem/article/101/4/1664/2804548 by guest on 24 March 2024

Miscarriage 591 17.5 934 9.9 ple sizes. One population-based study estimated a 2.4-fold
After index date
Identified pregnancies 3492 5777 increased risk of GDM in women with PCOS compared
Births 3021 86.5 5200 90.0 with women without a diagnosis of PCOS or symptoms
GDM 160 5.3 183 3.5 (19). However, the PCOS group was significantly older
Pre-eclampsia 236 7.8 302 5.8
Premature 214 7.1 288 5.5 and had a higher prevalence of multiple gestation than
Miscarriage 471 13.5 577 10.0 controls, which themselves are risk factors for GDM. Fur-
Combined thermore, given that groups were not matched for obesity,
Identified pregnancies 6861 15 214
Births 5799 84.5 13 703 90.1 it is unclear how much of these risks were due to PCOS per
GDM 253 4.4 418 3.1 se and how much to obesity, which is common in this
Pre-eclampsia 424 7.3 768 5.6 patient population and is itself associated with a higher
Premature 385 6.6 715 5.2
Miscarriage 1062 15.5 1511 9.9 risk of GDM. Indeed, one previous study showed that the
prevalence of GDM did not differ among women with
PCOS and controls when weight was matched for (20). In
tention to conceive between females with and without contrast, Roos et al (9), in a large population-based study
PCOS. of singleton births in Sweden, found that the risk of GDM
We also confirmed an increased risk of miscarriage in was more than doubled in women with PCOS, even after
women with PCOS. Only a few studies have examined the adjustment for age, BMI, and use of assisted reproductive
association between PCOS and early pregnancy loss (13– technology. This resonates with our understanding of
17). Some of these suggested that the risk of spontaneous PCOS as a metabolic disorder underpinned by defects in
abortion was higher in women with PCOS but only one insulin secretion and sensitivity. Our observations are

Figure 3. Adjusted† ORs for GDM, pre-eclampsia, premature birth, and miscarriage for women with and without PCOS. *Denominator is all
pregnancies, adjusted for age, body mass index, number of previous births, and smoking history.
1670 Rees et al Fertility, Pregnancy and Neonatal Outcomes in PCOS J Clin Endocrinol Metab, April 2016, 101(4):1664 –1672

Table 3. Neonatal Complications Reported for Offspring of Women With PCOS and Matched Controls
PCOS Non-PCOS
(n ⴝ 3707) (n ⴝ 8656)

Complication n % n % Adjusted ORa (95% CI) P

Jaundice 326 8.8 588 6.5 1.20 (1.03–1.39) .016
Respiratory 420 11.3 785 8.7 1.20 (1.06 –1.37) .005
High birth weight 40 1.1 104 1.2 0.97 (0.67–1.41) .894
Low birth weight 242 6.5 439 4.9 1.19 (1.00 –1.42) .054
Hypoglycemia 124 3.3 218 2.4 1.31 (0.99 –1.74) .062
Feeding issues 85 2.3 139 1.5 1.21 (0.96 –1.52) .106
Abbreviation: OR, odds ratio.

Downloaded from https://academic.oup.com/jcem/article/101/4/1664/2804548 by guest on 24 March 2024

a
Adjusted for age, BMI, number of previous births, multiple gestation, and smoking history.

consistent with these findings, albeit that the OR was for those women identified as being prescribed metformin
slightly lower at 1.4. This is also in keeping with our pre- prior to estimated conception and during their first tri-
vious CPRD study, which showed that women with PCOS mester. It is difficult to be certain whether this represents
had a 50% increased risk of type 2 diabetes in the United a true adverse effect of metformin therapy or is more likely
Kingdom population (3). a reflection of residual confounding due to preferential
Our finding of an increased risk of pre-eclampsia is prescribing in higher risk pregnancies. A large, multicenter
consistent with many (4, 21), but not all (22) previous randomized clinical trial found no effect of metformin,
studies but the mechanisms by which this develops are administered late in the first trimester through to delivery,
unclear. One such pathway may involve hyperinsulin- on pregnancy outcome, with the exception of an apparent
emia, which is a common finding even in lean patients with reduction in later miscarriage and preterm delivery (28).
PCOS, and which has been implicated in hypertension We did not find an effect of metformin on early miscar-
developing in pregnancy (22, 23). riage rate, consistent with previous studies (29, 30).
Two meta-analyses assessing pregnancy outcomes in In our study we have also shown that caesarean section
women with PCOS have suggested that the risk of preterm was more common as a method of delivery in patients with
delivery is increased in patients with the syndrome (4, 24). PCOS, even after adjustment for possible confounders.
Of the studies that informed these analyses (5, 20, 22, 25, This is in agreement with some (9) but not all studies (4).
26), only one excluded patients with multiple gestation Indeed, in a meta-analysis of pregnancy outcomes in
(26), which is a major risk factor for preterm delivery in its women with PCOS where subgroup analysis was re-
own right. Mikola et al (22) found that PCOS per se lost stricted to higher validity studies, no increased risk was
its significance as a risk for preterm birth in a multivariate observed (4). An increased rate of operative delivery was
analysis which identified multiple gestation and nullipar- associated with a 6% increased length of stay in fully ad-
ity as independent predictors. This may be in keeping with justed analyses. This is likely to lead to adverse health
the observation that many women with PCOS need ovu- economic consequences, of a similar order of magnitude as
lation induction to conceive and that such treatment is that apparent in obesity (31). As anticipated, a diagnosis
more likely to result in multiple gestation pregnancies. We of PCOS was also associated with an increased risk of
found an increased risk of premature delivery in our pop- multiple births likely related to a greater use of assisted
ulation, which could not be explained by differences in conception, albeit that multiple births only accounted for
age, BMI, multiple gestation, or previous pregnancy. This a small fraction of the total.
observation is in keeping with that of Roos et al (9) in their Neonatal outcomes were generally worse for infants
study of singleton pregnancies. However, obesity may ex- born to mothers with PCOS although we did not find a
acerbate the risk of preterm birth. De Frène and colleagues difference in stillbirth nor in Apgar scores with controls.
(27) reported an almost seven-fold increase in the propor- This contrasts with Roos et al (9), who found an increased
tion of preterm deliveries when comparing obese and risk of low Apgar score at 5 minutes in infants born to
nonobese females with PCOS. mothers with PCOS but no increased risk of neonatal
The effect of metformin use for women with PCOS was death. However, we did note a significantly increased risk
inconclusive as this study was not powered to detect dif- for neonatal jaundice and respiratory distress. To our
ferences within the PCOS cohort and the effect of multiple knowledge, ours is the first study to report such associa-
testing should be considered in interpreting these results. tions. There was also a trend toward increased risk of low
However, there was a significant increase in pre-eclampsia birth weight in our study, a finding which has been shown
doi: 10.1210/jc.2015-2682 press.endocrine.org/journal/jcem 1671

in one (32) but not other previous studies (9, 20, 22, 33), References
and also hypoglycemia.
1. Azziz R, Woods KS, Reyna R, Key TJ, Knochenhauer ES, Yildiz BO.
Significant strengths of our study include the large sam- The prevalence and features of the polycystic ovary syndrome in an
ple size and careful adjustment for possible confounders, unselected population. J Clin Endocrinol Metab. 2004;89:2745–
but our study has certain limitations. In common with 2749.
2. Asunción M, Calvo RM, San Millán JL, Sancho J, Avila S, Escobar-
other studies of this type, the CPRD collects data from Morreale HF. A prospective study of the prevalence of the polycystic
routine practice; thus, there are missing data and coding ovary syndrome in unselected Caucasian women from Spain. J Clin
imperfections. However, whereas these may introduce Endocrinol Metab. 2000;85:2434 –2438.
3. Morgan C, Jenkins-Jones S, Currie CJ, Rees DA. Evaluation of ad-
noise into the study, they are unlikely to introduce bias. verse outcome in young women with polycystic ovary syndrome
We were dependent upon hospital data for the accurate versus matched, reference controls: A retrospective, observational
ascertainment of births and therefore births outside Na- study. J Clin Endocrinol Metab. 2012;97:3251–3260.
4. Boomsma CM, Eijkemans MJ, Hughes EG, Visser GH, Fauser BC,

Downloaded from https://academic.oup.com/jcem/article/101/4/1664/2804548 by guest on 24 March 2024

tional Health Service hospitals (private or home births) Macklon NS. A meta-analysis of pregnancy outcomes in women
could not be included. Home births represent approxi- with polycystic ovary syndrome. Hum Reprod Update. 2006;12:
mately 2.5% of all births in the United Kingdom and it is 673– 683.
5. Palomba S, Falbo A, Russo T, Tolino A, Orio F, Zullo F. Pregnancy
possible that there are differences in this proportion for in women with polycystic ovary syndrome: The effect of different
those with and without PCOS due the perceived risk of phenotypes and features on obstetric and neonatal outcomes. Fertil
complications. Similarly, we did not have access to data Steril. 2010;94:L1805–11.
6. Kocak M, Caliskan E, Simsir C, Haberal A. Metformin therapy
for infertility consultations that occurred in the private improves ovulatory rates, cervical scores, and pregnancy rates in
sector. We were unable to analyze the effect of PCOS phe- clomiphene citrate-resistant women with polycystic ovary syn-
notype on reproductive outcome given that these data are drome. Fertil Steril. 2002;77:101–106.
7. Vandermolen DT, Ratts VS, Evans WS, Stovall DW, Kauma SW,
not captured in routine primary care. Based on routine Nestler JE. Metformin increases the ovulatory rate and pregnancy
data, we were also unable to determine the proportion of rate from clomiphene citrate in patients with polycystic ovary syn-
women in each group who were seeking to become preg- drome who are resistant to clomiphene citrate alone. Fertil Steril.
2001;75:310 –315.
nant. Within the PCOS group, fewer women were pre- 8. Ghazeeri GS, Nassar AH, Younes Z, Awwad JT. Pregnancy out-
scribed contraceptives within ⫾ 90 days of index date. It comes and the effect of metformin treatment in women with poly-
is also noticeable (Figure 2Ai) that for women with PCOS, cystic ovary syndrome: An overview. Acta Obstet Gynecol. 2012;
91:658 – 678.
first consultations for infertility are not uniformly distrib- 9. Roos N, Kieler H, Sahlin L, Ekman-Ordeberg G, Falconer H,
uted but centered around first diagnosis of PCOS. This Stephansson O. Risk of adverse pregnancy outcomes in women with
may be due to PCOS being diagnosed during investiga- polycystic ovary syndrome: Population based cohort study. BMJ.
2011;343:d6309.
tions for infertility, which may be indicative that the pa- 10. MHRA. The Clinical Practice Research Datalink. Available from:
tient is seeking to become pregnant. http://www.cprd.com. Last accessed April 8, 2015.
In conclusion, we have shown that younger women 11. Apgar V. The newborn (Apgar) scoring system. Reflections and
advic. Pediatr Clin North Am. 1966;13:645– 650.
with PCOS have reduced fertility but that fertility rates 12. Hudecova M, Holte J, Olovsson M, Sundström Poromaa I. Long-
may be restored to those of the background population term follow-up of patients with polycystic ovary syndrome: Repro-
after diagnosis. We have also confirmed an adverse effect ductive outcome and ovarian reserve. Hum Reprod. 2009;24:1176 –
1183.
of PCOS, independently of obesity, on several pregnancy 13. Glueck CJ, Wang P, Fontaine RN, Sieve-Smith L, Tracy T, Moore
outcomes, including GDM, pre-eclampsia, and preterm SK. Plasminogen activator inhibitor activity: An independent risk
birth. These observations suggest that women with PCOS factor for the high miscarriage rate during pregnancy in women with
polycystic ovary syndrome. Metabolism. 1999;48:1589 –1595.
require more intensive monitoring during pregnancy, and 14. Wang JX, Davies MJ, Norman RJ. Polycystic ovarian syndrome and
emphasize the need for additional studies to understand the risk of spontaneous abortion following assisted reproductive
technology treatment. Hum Reprod. 2001;16:2606 –2609.
the mechanisms by which these complications arise.
15. Glueck CJ, Phillips H, Cameron D, Sieve-Smith L, Wang P. Con-
tinuing metformin throughout pregnancy in women with polycystic
ovary syndrome appears to safely reduce first-trimester spontaneous
abortion: A pilot study. Fertil Steril. 2001;75:46 –52.
Acknowledgments 16. Glueck CJ, Wang P, Goldenberg N, Sieve-Smith L. Pregnancy out-
comes among women with polycystic ovary syndrome treated with
Address all correspondence and requests for reprints to: Aled metformin. Hum Reprod. 2002;17:2858 –2864.
Rees, Neurosciences and Mental Health Research Institute, 17. Jakubowicz DJ, Iuorno MJ, Jakubowicz S, Roberts KA, Nestler JE.
School of Medicine, Cardiff University, Heath Park, Cardiff Effects of metformin on early pregnancy loss in the polycystic ovary
CF24 4HQ, United Kingdom. E-mail: reesda@cf.ac.uk. syndrome. J Clin Endocrinol Metab. 2002;87:524 –529.
18. Homburg R. Pregnancy complications in PCOS. Best Pract Res Clin
This work was supported by the Centre for Endocrine and
Endocrinol Metabol. 2006;20:281–292.
Diabetes Sciences, Cardiff University. 19. Lo JC, Feigenbaum SL, Escobar GJ, Yang J, Crites YM, Ferrara A.
Disclosure Summary: The authors have nothing to disclose. Increased prevalence of gestational diabetes mellitus among women
1672 Rees et al Fertility, Pregnancy and Neonatal Outcomes in PCOS J Clin Endocrinol Metab, April 2016, 101(4):1664 –1672

with diagnosed polycystic ovary syndrome: A population-based of the pregnancy, delivery and neonatal outcome in overweight ver-
study. Diabetes Care. 2006;29:1915–1917. sus normal weight women with polycystic ovary syndrome. Hum
20. Haakova L, Cibula D, Rezabek K, Hill M, Fanta M, Zivny J. Preg- Reprod. 2014;29:2333–2338.
nancy outcome in women with PCOS and in controls matched by age 28. Vanky E, Stridsklev S, Heimstad R, et al. Metformin versus placebo
and weight. Hum Reprod. 2003;18:1438 –1441. from first trimester to delivery in polycystic ovary syndrome: a ran-
21. Fridström M, Nisell H, Sjöblom P, Hillensjö T. Are women with domized, controlled multicenter study. J Clin Endocrinol Metab.
polycystic ovary syndrome at an increased risk of pregnancy-in- 2010;95:E448 –E455.
duced hypertension and/or preeclampsia? Hypertens Pregnancy. 29. Palomba S, Falbao A, Orio F Jr, Zullo F. Effect of preconceptional
1999;18:73– 80. metformin on abortion risk in polycystic ovary syndrome: A sys-
22. Mikola M, Hiilesmaa V, Halttunen M, Suhonen L, Tiitinen A. Ob- tematic review and meta-analysis of randomized controlled trials.
stetric outcome in women with polycystic ovarian syndrome. Hum Fertil Steril. 2009;92:1646 –1658.
Reprod. 2001;16:226 –229. 30. Morin-Papunen L, Rantala AS, Unkila-Kallio L, et al. Metformin
23. de Vries MJ, Dekker GA, Schoemaker J.Higher risk of preeclampsia improves pregnancy and live-birth rates in women with polycystic
in the polycystic ovary syndrome: A case control study. Eur J Obstet ovary syndrome (PCOS): A multicenter, double-blind, placebo-con-
Gynecol Reprod Biol. 1998;1:91–95. trolled randomized trial. J Clin Endocrinol Metab. 2012;97:1492–

Downloaded from https://academic.oup.com/jcem/article/101/4/1664/2804548 by guest on 24 March 2024

24. Kjerulff LE, Sanchez-Ramos L, Duffy D.Pregnancy outcomes in 1500.
women with polycystic ovary syndrome: A meta-analysis. Am J Ob- 31. Denison FC, Norwood P, Bhattacharya S, et al. Association between
stet Gynecol. 2011;204:558.e1– e6. maternal body mass index during pregnancy, short-term morbidity,
25. Altieri P, Gambineri A, Prontera O, Cionci G, Franchina M, Pas- and increased health service costs: A population-based study.
quali R. Maternal polycystic ovary syndrome may be associated BJOG. 2014;121:72– 81.
with adverse pregnancy outcomes. Eur J Obstet Gynecol Reprod 32. Sir-Petermann T, Hitchsfeld C, Maliqueo M, et al. Birth weight in
Biol. 2010;149:31–36. offspring of mothers with polycystic ovarian syndrome. Hum Re-
26. Weerakiet S, Srisombut C, Rojanasakul A, Panburana P, Thakkin- prod. 2005;20:2122–2126.
stian A, Herabutya Y. Prevalence of gestational diabetes mellitus 33. Vollenhoven B, Clark S, Kovacs G, Burger H, Healy D. Prevalence
and pregnancy outcomes in Asian women with polycystic ovary of gestational diabetes mellitus in polycystic ovarian syndrome
syndrome. Gynecol Endocrinol. 2004;19:134 –140. (PCOS) patients pregnant after ovulation induction with gonado-
27. De Frène V, Vansteelandt S, T’Sjoen G, et al. A retrospective study trophins. Aust NZ J Obstet Gynaecol. 2000;40:54 –58.