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


Between December 2017 and November 2018, 200 women were screened for eligibility (Figure 1) and sixty were enrolled. Forty-seven participants completed run-in and were randomized to control (n = 17), HIIT (n = 16) or CAET (n = 14). Post-randomization attritions were 1 (6%), 3 (19%) and 3 (21%) among those randomized to control, HIIT and CAET, respectively.

Figure 1.

Study flow chart

Participant Characteristics

The groups were similar at baseline (Table 1). Twenty-eight participants were PCOS phenotype A (hyperandrogenism, ovulatory dysfunction, polycystic ovaries), 13 were phenotype B (hyperandrogenism, ovulatory dysfunction), two were phenotype C (hyperandrogenism, polycystic ovaries), and four were phenotype D (polycystic ovaries, ovulatory dysfunction). Twelve (26%) participants had previous pregnancies, with six (13%) having live births. The most common PCOS treatments attempted before trial enrolment were mechanical hair removal (89%), oral contraceptive (81%) and lifestyle intervention (53%).

Adherence to Prescribed Exercise

Participants assigned to exercise groups were prescribed 78 exercise sessions over 26 weeks. For those who completed the intervention, median overall adherence to exercise for both exercise groups was 68% (IQR 53%, 86%). There was no statistically significant difference in adherence to exercise between CAET (81%; IQR 56%, 85%) and HIIT (65%; IQR 51%, 85%; p = .91; Figure 2). In the first 3 months of the intervention, median adherence to prescribed exercise was 78% (IQR 55%, 97%) for HIIT and 88% (IQR 56%, 100%) for CAET. Median exercise adherences during the last 3 months of the intervention were 44% (IQR 41%, 69%) for HIIT and 64% (IQR 10%, 77%) for CAET.

Figure 2.

Adherence to prescribed exercise

Ovulation Prediction Kit Testing Adherence

During run-in, all 47 randomized participants completed ≥75% of the prescribed daily OPK tests, with 35/47 (74.5%) completing ≥90%; OPK-testing adherence was 97% (IQR 88%, 99%). During the first 3 months of the 6-month intervention, median adherence was 87% (IQR 61%, 97%) with 30/47 (63.8%) completing ≥75% of the OPK tests and 19/47 (40.4%) completing ≥90%. During the final 3 months of the intervention, 17/47 (36.2%) participants completed ≥90% of prescribed OPK tests, and median adherence was 65% (IQR 0%, 96%). There was no difference in OPK-testing adherence between-groups at any assessment interval (Figure 3).

Figure 3.

Adherence to ovulation prediction kit testing

Reproductive Outcomes

Given sub-optimal adherence to OPK-testing throughout the intervention, we were unable to analyse effects of the exercise interventions on ovulation using an intention-to-treat analysis. During the 3-month run-in-phase, 28/47 (59.6%) participants ovulated ≥once. In a per-protocol analysis of 33 participants completing ≥75% prescribed OPKs during the intervention, 22 (67%) had ≥1 documented ovulation. Further, there was no significant between-group difference in ovulation events during the intervention: 8/12 (67%) in control, 8/11 (73%) in HIIT and 6/10 (60%) in CAET.

Forty-two participants tracked menstrual cycles during run-in and intervention phases. Proportions of participants by group with regular menses (21–35 days) pre-intervention compared with the last three months of the intervention were 29% vs 47% (p = .31) in control, 50% vs 53% (p = .85) in HIIT and 29% vs 42% (p = .48) in CAET. There were no significant between-group differences in proportions with regular menstrual cycles pre-intervention (p = .51) or in the last three months of the intervention (p = .87). Eight participants with amenorrhoea or oligomenorrhoea at baseline had regular cycles during the intervention: four in control, three in HIIT and one in CAET (p = .32). These participants all had BMI > 25 kg/m2 at baseline and were PCOS phenotype A or B, and half lost weight during the intervention. One control participant and one CAET participant had regular menses at baseline that became irregular over the six-month intervention; both had baseline BMI > 40 kg/m2.

Twenty-nine participants had menstrual cycle lengths recorded during both run-in-phase and intervention. For ten control group participants, mean cycle length pre-intervention was 45.1 (SD 17.9) days compared with 37.7 (SD 13.9) days at the end of intervention (p = .23). For 12 HIIT participants, mean menstrual lengths pre-intervention and end of intervention were 35.9 (SD 11.9) days and 32.5 (SD 9.5) days, respectively (p = .23), compared with seven CAET group participants who had mean lengths of 46.8 (SD 20.0) days and 34.3 (SD 6.1) days, respectively (p = .09). There were no between-group differences pre-intervention (p = .53) or end of intervention (p = .57).

Twenty-seven participants had documented ovulation events and menstrual periods during the pre-intervention run-in-phase allowing estimation of the luteal phase length: nine in control, 12 in HIIT and six in CAET compared with 11 participants in the last three months of the intervention (three in control, five in HIIT and three in CAET). The mean luteal phase was 13.6 days (SD 2.5) compared with 14.2 days (SD 1.5) for the 23 participants during the intervention (p = .55). There were no between-group differences in the mean luteal phase length pre-intervention (p = .54) or end of intervention (p = .38).

Three participants achieved pregnancy during run-in and all resulted in live births.


The mean Ferriman-Gallwey hirsutism scores[20] were not significantly different pre-intervention and end of intervention in control (p = .31), HIIT (p = .44), or CAET (p = .86). There were no significant between-group differences: control vs HIIT (p = .91), control vs CAET (p = .58) and HIIT vs CAET (p = .68).

Anthropometric and Physical Fitness Outcomes

There was a statistically significant decrease of −0.8 kg/m2 in BMI in the CAET group (95%CI −1.4–0.2, p < .01; Table 2). Waist circumference decreased significantly within the controls (−4.5 cm, 95%CI −8.6–0.4; p = .03), CAET (−6.9 cm, 95%CI −11.5–2.3; p < .01) and HIIT groups (−7.3 cm, 95%CI −11.8–2.9; p < .01), with no statistically significant between-group difference. No within- or between-group differences were observed in VO2max (Table 2).

Cardiometabolic Outcomes

Mean fasting glucose increased in HIIT (0.3 mmol/L, 95% CI 0.1–0.6; p =. 02), which was significantly different from controls (p = .04) (Table 3). Mean fasting insulin levels increased significantly in controls by 19.5 mIU/L (95% CI 0.9–38.2; p = .04) with no changes within the CAET (p = .19) or HIIT group (p = .90), nor were between-group differences observed. LDL-C was reduced in the HIIT group compared with CAET (−0.3 mmol/L, 95%CI −0.6–0.1; p = .03). HDL-C increased in the HIIT group compared with control (0.2 mmol/L, 95%CI 0.0–0.4; p = .04), with no difference between CAET and control (p = .47). No within- or between-group differences were observed in blood pressure, HbA1c, HOMA2-IR, total cholesterol, triglycerides, ALT or GGT (Table 3).

Adverse Events

No trial-related adverse events were reported. One participant in the CAET group sustained a musculoskeletal injury unrelated to exercise training and withdrew from the trial.