Exercise Training and Reproductive Outcomes in Women With Polycystic Ovary Syndrome

A Pilot Randomized Controlled Trial

Jamie L. Benham; Jane E. Booth; Bernard Corenblum; Steve Doucette; Christine M. Friedenreich; Doreen M. Rabi; Ronald J. Sigal


Clin Endocrinol. 2021;95(2):332-343. 

In This Article

Materials and Methods

Study Design

We conducted a 6-month, single-centre RCT in Calgary, Canada, between December 2017 and September 2019. The protocol was registered prospectively (NCT03362918). The trial had a parallel-group design with three phases: (1) 3-month run-in-phase; (2) 6-month intervention after randomization to HIIT, CAET or a no-exercise control group; and (3) 6-month post-intervention follow-up. Trial outcomes were assessed objectively by health professionals and laboratory technologists, blinded to participant group assignment. The protocol was approved by the University of Calgary Conjoint Health Research Ethics Board (REB17-1574). All participants provided written informed consent.


Untrained women aged 18–40 years with PCOS defined by Rotterdam criteria[16] were recruited by advertisements, media, physician referrals and word of mouth. Exclusion criteria included medical conditions restricting exercise, participation in >40 min of exercise training weekly and medications potentially affecting ovulation (glucocorticoids, metformin, gonadotropins, clomiphene, letrozole, oestrogens, progestins). Potential participants were screened for inclusion and exclusion criteria by telephone. If eligible, they were mailed requisitions for the following screening investigations: (1) electrocardiogram and (2) blood tests (serum beta-HCG, 17-hydroxyprogesterone, prolactin, thyroid-stimulating hormone, fasting plasma glucose and haemoglobin A1C (HbA1c)). Potential participants meeting trial participation criteria were invited for in-person assessment where a baseline history and physical examination were completed.

Run-in Phase

Prior to randomization, participants entered a 3-month run-in-phase to assess baseline reproductive function including ovulation rate and menstrual cycle length and frequency. Participants did not exercise during this phase. Participants were asked to track menstrual cycles and check for ovulation using an at-home ovulation prediction kit (OPK) (Verify Diagnostics) daily which measured luteinizing hormone (LH) in urine. As LH surge duration is typically just 24–48 h,[17] adherence to daily testing was important. Participants sent photographs of completed test strips to the research team for verification through a secure messaging application (WhatsApp Inc.). If positive, ovulation was confirmed with a serum progesterone level 1 week after positive OPK results.


Participants completing >75% of daily OPKs during run-in were randomized to control, HIIT or CAET. Groups were stratified by body mass index (BMI, kg/m2) (< or ≥28 kg/m2). Central randomization was done using a secure web application (Research Electronic Data Capture (REDCap)). Block sizes varied among two, four or six. Allocation concealment was used prior to randomization.

Intervention Phase

Throughout the intervention period, participants were asked to continue tracking menstrual cycles and completing OPK-testing daily. Physical activity was tracked for all participants using Polar A370 (Polar Electro Oy). Control group participants were asked to maintain their usual level of physical activity throughout the intervention and were offered three sessions with a personal trainer upon completion.

Exercise group participants completed three exercise sessions/week using the aerobic exercise equipment of their choice. Gym memberships and parking were provided free of charge. All exercise sessions included a five-minute warm-up and five-minute cool-down. HIIT participants completed 10 cycles of 30 s at high-intensity (90% of heart rate reserve (HRR), or 9/10 on a modified Borg scale[18]) alternating with 90 s of low-intensity aerobic exercise. CAET participants completed 40 min of moderate-intensity aerobic exercise (50%–60% HRR, or 4–6/10 on a modified Borg scale). For better precision, a Polar H10 heart rate sensor was synced to the Polar A370 watches.

Outcomes and Measurements

Four feasibility outcomes were determined a priori: (1) randomization of ≥36 participants in 15 months; (2) <25% attrition; (3) adherence to >90% of daily OPK-testing; and (4) adherence to >70% of prescribed exercise. OPK-testing adherence was calculated as the number of OPK digital photographs completed divided by the total number of days of requested OPK tests. Prescribed exercise session completion was assessed as the attendance at the twice-weekly supervised sessions and the number of unsupervised sessions recorded using the Polar Flow App and verified using the Polar Coach platform.[19] Exercise adherence was calculated as the number of exercise sessions completed divided by the number of prescribed sessions.

We evaluated menstrual cycle length, luteal phase length and numbers of ovulation events, pregnancies, abortions and live births. Menstrual cycle length was calculated from the first day of menses to the first day of the subsequent menses. Luteal phase length was calculated from the day of ovulation to the first day of the subsequent menses. An ovulation event was documented if a positive OPK result was confirmed (serum progesterone level ≥5.0 nmol/L). Pregnancy was confirmed by foetal cardiac activity on a first-trimester ultrasound. Ferriman-Gallwey score[20] was assessed pre-intervention and end of intervention.

Height and weight were assessed using a stadiometer (SECA-220, Seca GmbH & Co.) and weigh scale (SECA-703, Seca GmbH & Co.). Waist circumference was measured midway between the lowest rib and iliac crest in the horizontal plane using a tape measure to the nearest 0.5 cm. Blood pressure was measured using an automated device (Omron HEM-907, Omron Healthcare Inc.).

Blood work was drawn after fasting for eight hours. Glucose, total cholesterol, high-density lipoprotein cholesterol (HDL-C), triglycerides, alanine transferase (ALT) and gamma-glutamyl transferase (GGT) were measured using enzymatic methods on a Cobas c701 analyser (Roche Diagnostics). Insulin was measured using a chemiluminescent microparticle immunoassay on an Architect i2000SR analyser (Abbott). HbA1c was measured using a Tina-quant Hemoglobin A1cDx Gen.3 assay and a Cobas c513 analyser (Roche Diagnostics). Low-density lipoprotein cholesterol (LDL-C)[21] and homeostasis model assessment index of insulin resistance (HOMA2-IR)[22] were calculated.

Exercise testing was completed pre-intervention and end of intervention by Certified Exercise Physiologists. Maximal oxygen uptake (VO2max) was determined using a Vmax Encore 29 metabolic cart (SensorMedics Corporation). Breath-by-breath ventilatory volumes and expiratory gases were measured during a ramp exercise test (Balke-Ware protocol[23]) on a programmable treadmill (Trackmaster, Full Vision Inc.). Ventilatory thresholds were determined and verified by two independent investigators according to the V-slope method.[24]

Adverse Events

A standard form was created to document adverse events. Participants were asked to report adverse events to the research team at the mid- and end-of-intervention assessments.

Sample Size

In addition to assessing pre-specified feasibility outcomes, our goal for this feasibility trial was to enrol sufficient participants to evaluate the preliminary effectiveness of HIIT and CAET on ovulatory rate. Previous studies reported improvements of up to 65% in ovulation rate with exercise training.[7,25] We estimated that 36 randomized participants (12 per group) would allow for 81% power to detect a difference of 0.6 in the proportion of women with improvement in ovulation rate between the control group and each exercise group with an ɑ = 0.05. We estimated 40% attrition during run-in because of the daily OPK protocol and therefore enrolled sufficient participants to result in ≥36 randomized.

Statistical Analysis

For baseline characteristics, means were compared with one-way analysis of variance (ANOVA) and proportions were compared with Fisher's exact test. The reproductive analysis was conducted on an intention-to-treat basis and all randomized participants were included. We also performed per-protocol analyses including only participants completing end-of-study measures and ≥75% of daily menstrual cycle and OPK data collection. Tests of proportion were used to compare within- and between-group differences pre- and post-intervention. Mean menstrual cycle and luteal phase lengths were compared within-group pre- and post-intervention using paired t tests and between-groups using ANOVA.

For anthropometric and cardiometabolic outcomes, repeated measures mixed models were used with effects for time, group and time by group interaction, with age as a covariate and an unstructured covariance matrix. Within the mixed models, we estimated 95% confidence intervals (CI) and p-values for intergroup contrasts and for change in each variable over time. Only participants who had ≥1 postbaseline assessment were included.

Analyses were conducted using SAS version 9.4 (SAS Institute) and STATA version 15.1 (StataCorp.). Statistical significance was set as p < .05.