The effect of lifestyle interventions in women with polyendocrine metabolic ovarian syndrome: A systematic review and meta-analysis

A total of 34 studies (1,871 participants) were included from the retrieved 1,798 publications. Women who participated in lifestyle interventions were more likely to achieve regular menstrual cycles OR 4.03 (95% CI [1.16, 14.05] I 2 = 73%, 5 studies, 237 women), and have reduced waist circumference mean difference (MD) −2.38 (95% CI [−4.68, −0.08] I 2 = 90%, 18 studies, 915 women), fasting glucose MD −0.13 (95% CI [−0.25, −0.02] I 2 = 86%, 22 studies, 1,045 women) and fasting insulin MD −6.08 (95% CI [−9.86, −2.30] I 2 = 71%, 18 studies, 824 women) compared to women who had minimal or no treatment. Subgroup analyses showed that diet interventions had greater effect on reducing fasting glucose levels while exercise was most effective at reducing waist circumference and fasting insulin. In women with PMOS who were overweight or obese, lifestyle interventions led to greater reduction in weight, waist/hip ratio, fasting glucose, fasting insulin, total testosterone, total cholesterol, and low-density lipoprotein. The main limitation of this review is the high between-study heterogeneity in pooled estimates across many of the outcomes.

We searched Medline, EMBASE, All EBM, PsycInfo, and CINAHL to June 2026 for studies examining lifestyle intervention compared with minimal or no treatment (usual care) in PMOS. Meta-analyses were conducted using random-effects model to estimate pooled odd ratios (OR) with 95% confidence intervals (CI). We conducted subgroup analyses to evaluate differences by intervention type and duration, and by baseline body mass index (BMI).

Funding: This study was supported by the National Institute for Health and Care Research (NIHR) Birmingham Biomedical Research Centre [ https://www.birminghambrc.nihr.ac.uk ] (to OO), the Australian National Health and Medical Research Council (NHMRC) fellowships [ https://www.nhmrc.gov.au/ ] (to AM, LM, and HT), and the NIHR Senior Investigator Award [ https://www.nihr.ac.uk/career-development/research-career-funding-programmes/professorships-and-senior-investigators/senior-investigators/award ] (to ST). The funders had no other role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Given the rapidly evolving nature of research in this field, this review aims to provide more comprehensive evaluation of the effectiveness of lifestyle interventions on reproductive, anthropometric, and metabolic outcomes in women with PMOS, and to explore whether these effects vary by intervention type, duration, or baseline body mass index (BMI).

Lifestyle interventions, including exercise and diet modification, are recommended as first-line management for symptomatic women with PMOS [ 1 ]. These strategies are commonly used for both weight management (including weight loss, weight maintenance or prevention of excessive weight gain) and overall general cardiometabolic health improvement [ 14 ]. Recent reviews [ 1 , 15 , 16 ] have reported beneficial effects of lifestyle interventions on reproductive, metabolic, and anthropometric outcomes in women with PMOS. However, existing evidence is limited by the number of studies, variation in intervention types, high risk of bias, and lack of granularity regarding optimal types and durations of interventions [ 17 ].

While the exact cause of PMOS remains unclear, it is thought to involve a combination of genetic, environmental and lifestyle factors [ 7 ]. Hallmark features of PMOS include hyperinsulinemia, hyperandrogenism and neuroendocrine hormone abnormalities, which contribute to downstream metabolic and reproductive complications. This complex hormonal milieu contributes to a high prevalence of cardiometabolic disorders in PMOS, including obesity, dyslipidaemia, hypertension, diabetes, gestational diabetes and cardiovascular disease [ 8 – 10 ]. Underpinning many of these metabolic and endocrine disturbances is insulin resistance, an intrinsic pathological factor which is exacerbated by obesity-related insulin resistance in PMOS [ 11 , 12 ]. Given its central role, interventions targeting insulin resistance are key to improving both clinical and metabolic outcomes in most women with PMOS [ 11 , 13 ].

Polyendocrine metabolic ovarian syndrome (PMOS) is a common, heterogeneous endocrine-metabolic disorder affecting 10% − 13% of women of reproductive age [ 1 , 2 ]. This important, yet often under recognised public health problem, confers a range of reproductive, cardiometabolic, dermatologic and psychological sequelae [ 1 ]. PMOS accounts for approximately 40% of women presenting with subfertility [ 3 ], and over 90% of women with anovulation attending infertility clinics [ 4 ]. Moreover, chronic oligo-anovulation results in prolonged exposure to unopposed oestrogen, thus increasing the risk of endometrial hyperplasia and cancer. Pregnancy in women with PMOS is independently associated with increased maternal and perinatal risks, which are further exacerbated by higher body mass index (BMI) and gestational weight gain. Consequently, PMOS is considered a high-risk pregnancy condition, where early and sustained lifestyle modifications may offer potential benefits across reproductive health [ 5 ]. An international, multidisciplinary consensus recently adopted the term polyendocrine metabolic ovarian syndrome (PMOS) to better reflect the complex endocrine, metabolic, and reproductive features of the condition. This terminology replaces polycystic ovary syndrome (PCOS) and aims to reduce the stigma and inaccuracies associated with the previous name [ 6 ].

We performed sensitivity analyses whereby data from studies assessed to have high risk of bias were excluded to gage their impact on the findings. Blinding was not used in the sensitivity analysis as it was not possible to blind participants and the intervention provider due to the interactive nature of the interventions.

To assess differences between the intervention and control (minimal treatment) groups, we calculated effect sizes for each dichotomous and continuous outcome, reported as odds ratios (OR) or weighted mean differences (MD), respectively, with corresponding 95% confidence intervals (CI). The studies were weighted based on the inverse of the variance for the evaluated measure with a random-effects model. Heterogeneity was summarised using I 2 values, defined as low, moderate, or high heterogeneity according to the cut-offs of 25%, 50%, and 75%, respectively. We determined statistical significance using a two-tailed threshold of P < 0.05. All analyses were conducted on Review Manager version 5 [ 23 ].

Two reviewers independently extracted data using a pre-designed data extraction sheet. Relevant study characteristics were extracted including year of publication, country, participant inclusion and exclusion criteria, study intervention, number of recruited participants, and details of comparator group(s). Dichotomous data were extracted as 2 × 2 tables, while means and standard deviations (SD) were extracted for continuous data. Where medians and interquartile ranges were reported instead of means and SD, we used medians in place of means and calculated the SD using the formula of SD = (third quartile-first quartile)/1.35. Disagreements were resolved by consensus.

We used the Cochrane Risk of Bias (ROB) 2.0 tool [ 22 ] to evaluate study-level risk of bias across five domains: randomisation process, deviations from intended intervention, incomplete outcome data, outcome measurement, and selective reporting. Two reviewers independently assessed study quality, and any disagreements were resolved through consultation with a third, independent reviewer. We considered a study as having a low risk of bias if the trial was awarded low risk of bias in all domains, trials with some concerns of bias in one domain were considered as having a moderate risk of bias, and trials with a high risk of bias in at least one domain or some concerns of bias in multiple domains were considered as having a high risk of bias.

We included all randomised controlled trials (RCTs) investigating the effects of lifestyle interventions on fertility, hormonal, anthropometric and/or metabolic outcomes in women with PMOS and applied no language restrictions. The diagnosis of PMOS across the included studies was according to NIH 1990, Rotterdam 2003 [ 19 ], or AES-PCOS 2006 criteria [ 20 ]. Eligible lifestyle interventions included exercise, diet, behavioural management or a combination of these measures for a minimum duration of two weeks. We defined the comparator as minimal treatment which comprises no treatment/usual care or standard unstructured minimal dietary, exercise or behavioural advice. Following the CORE outcomes in PCOS [ 21 ], we planned to assess a broad range of reproductive outcomes (live birth, clinical pregnancy, miscarriage, biochemical pregnancy, ovulation, ovarian hyperstimulation, oocyte yield, oocyte maturation rate, duration of ovarian stimulation, fertilisation rate, ovarian hyperstimulation syndrome, preterm birth, gestational age at birth, gestational diabetes mellitus, pre‑eclampsia, intra‑uterine growth restriction, fetal macrosomia, stillbirth, neonatal death, perinatal death, NICU admission, cesarean section and shoulder dystocia). However, in the included trials, extractable reproductive data were almost entirely limited to regular and irregular menstrual cycles and, in one study, conception resulting in live birth. We therefore treated menstrual cycle regularity as the primary reproductive outcome and conception/live birth as exploratory outcome. Other non-reproductive outcomes include weight, BMI, waist circumference, waist-hip ratio, sex hormone-binding globulin, total testosterone, free androgen index, fasting glucose, fasting insulin, and lipid profile (total cholesterol, triglycerides, high-density and low-density lipoprotein cholesterol).

We updated our previous search for a 2022 technical report for the International Evidence-based Guideline for the Assessment and Management of PCOS [ 17 ] by searching Ovid Medline, EMBASE, All EBM, PsycInfo, and CINAHL from 1 January 2022 to 9 June 2026, using the same search terms ( S1 Appendix ). There were no material differences in the eligibility criteria, screening processes, or data synthesis methods between this review and the 2022 technical report. Two reviewers (OO and STq) independently identified eligible studies based on the predefined criteria by screening the titles and abstracts and then assessing the full text. We resolved disagreements through discussion and consensus.

There were overall subgroup differences by duration of intervention for waist circumference, fasting glucose or fasting insulin (all P < 0.05). Within subgroups, lifestyle interventions lasting for one to three months had greater reductions in, fasting glucose and fasting insulin compared to minimal treatment MD −0.25 mmol/L (95% CI [−0.43, −0.07] p < 0.01) and MD −7.03 pmol/L (95% CI [−12.47, −1.59] p < 0.01), respectively ( Table 4 ). Interventions lasting more than three months were effective in reducing total testosterone levels MD −0.18nmol/L (95% CI [−0.36, −0.01] p = 0.04). There were no significant differences by duration of lifestyle intervention for all other measured outcomes ( Table 4 , S5 Appendix ).

Diet modifications and combined interventions had significant reduction on fasting glucose levels compared to minimal treatment with MDs of −0.38 mmol/L (95% CI [−0.60, −0.17] P = 0.0005) and −0.42 mmol/L (95% CI [−0.44, −0.40] P < 0.00001), respectively. There were no differences in fasting glucose in women who had exercise or behaviour change interventions versus minimal treatment, with MDs of −0.01 mmol/L (95% CI [−0.18, 0.16] P = 0.92) and −0.22 mmol/L (95% CI [−0.45, 0.01] P = 0.06), respectively ( Fig 3e ).

There were no significant differences in BMI in those who had diet modification MD 0.43 kg/m 2 (95% CI [−1.42, 2.28] P = 0.65), behaviour change MD −0.44 kg/m 2 (95% CI [−1.68, 0.08] P = 0.49), or a combined intervention MD 0.95 kg/m 2 (95% CI [−4.09, 5.98] P = 0.71) versus those who had minimal treatment ( Fig 3a ). No differences in waist circumference were seen between women who had diet modification or combined interventions versus those who had minimal treatment, with a MD −1.77 cm (95% CI [−8.82, 5.28] P = 0.62) and MD 4.6 cm (95% CI [−12.03, 21.23] P = 0.59) ( Fig 3b ). Diet, behaviour change, or combined interventions did not show significant difference on fasting insulin when compared with minimal treatment ( Fig 3c ).

Following the exclusion of studies with an overall high risk of bias rating, only the beneficial effect of lifestyle interventions on fasting insulin MD −5.71 pmol/L (95%CI [−9.60, −1.83] P = 0.004) compared with minimal treatment persisted, while differences in fasting glucose and waist circumference were attenuated MD −0.07 (95% CI [−0.20, 0.06]) and MD −1.94 (95% CI [−3.93, 0.05]), respectively ( Table 2 ).

Only one of the included studies [ 58 ] reported on conception resulting in live birth. This follow-up study on pregnancy outcomes from a RCT assessing lifestyle interventions (a combination of normo-caloric diet, exercise according to the World Health Organization’s Global Recommendations, and cognitive behavioural therapy) in overweight/obese women with PMOS found that 39.8% (49/123) of those in the intervention group achieved pregnancy which resulted in live birth, compared to 38.3% (23/60) of women in the usual care group [ 58 ]. The finding was not statistically significant.

Using the Cochrane ROB2 tool [ 22 ], 79% (27/34) of the included studies had a low risk of bias for the randomisation process, 56% (19/34) for effect of assignment to intervention, 59% (20/34) for effect of adhering to intervention, 79% (27/34) for missing outcome data, 88% (30/34) for meassurement of outcome, and 91% (31/34) for selection of reported data. Thirty-five percent (12/34) of the included studies had an overall high risk bias ( Fig 2 , S3 Appendix ).

Discussion

This systematic review and meta-analysis provide quantitative estimates of the effects of lifestyle interventions on reproductive, anthropometric and metabolic outcomes in women with PMOS. Overall, lifestyle intervention led to more than 4-fold increase in the odds of achieving regular menstrual cycles. Significant reductions in waist circumference, fasting glucose and fasting insulin were also observed when compared with minimal treatment. Subgroup analyses revealed variation in effect by intervention type, duration, and baseline BMI status, with beneficial effects on other outcomes seen within certain subgroups. Exercise-based interventions improved anthropometric outcomes and fasting insulin compared with minimal treatment. Similarly, lifestyle interventions lasting one to three months led to significant reductions in fasting glucose and fasting insulin, while interventions lasting more than three months improved total testosterone level. Among the subgroup of participants with baseline overweight or obesity, lifestyle intervention improved weight, waist-hip ratio, total testosterone, and lipid profile compared with minimal treatment.

The main strength of this review is the robust methodology, incorporation of the most recent evidence to date and building on prior systematic reviews. With the inclusion of 34 studies, making it the most comprehensive analysis to date on the impact of diet, exercise, and behaviour changes on a range of outcomes in PMOS. This review expands on current knowledge and improves generalisability of the findings by including studies from high-, upper-middle-, and lower-middle-income countries.

The main limitation of this review was that over a third of the included studies had overall high risk of bias, reflecting methodological weaknesses which may lead to misrepresentation of the true effects of lifestyle interventions. To mitigate this, we conducted sensitivity analysis by excluding these studies to establish their influence on the findings. Another limitation is the broad definition of lifestyle interventions, which encompass a wide variety of diet, exercise, behavioural, or combined interventions and contribute to heterogeneity in our findings. In addition, we included studies that used different PMOS diagnostic criteria (NIH 1990, Rotterdam 2003, and AES‑PCOS 2006), which may have resulted in variation in baseline phenotype and outcome responses across trials and represent a further important source of clinical heterogeneity. Due to substantial statistical heterogeneity across most outcomes, we did not formally assess publication bias using funnel plots or Egger’s test as they may be difficult to interpret in these conditions. Participant characteristics and baseline risk profiles can also differ and affect intervention responses. We addressed this to some extent with the inclusion of subgroup analyses by type and duration of lifestyle intervention and by baseline BMI categories of study participants. However, while these sub-analyses can provide valuable insights into the differential effects caused by these factors, statistical power is limited, particularly for some subgroups with small numbers of studies. Lastly, although many of the included studies adjusted for confounders, the variability in intervention protocols and outcome measurements across the included studies may have introduced additional confounders such as cultural dietary patterns and variations in adherence to interventions which are difficult to standardise across studies and regions. In addition, a formal assessment of study integrity was not undertaken because this review focussed on trial-level risk of bias and evidence certainty using the Cochrane RoB 2 tool and GRADE.

Improving on a previous review [1] which identified only two studies reporting on reproductive outcomes, we included five studies reporting on the effects of lifestyle interventions on menstrual regularity; three studies assessed menstrual regularity at the end of the study without specifying the exact time to cycle normalisation, one study reported regular menses at 90–120 days, and one study assessed this outcome by 6 months. Overall, we found lifestyle interventions to be associated with an over 4-fold increase in the odds of achieving regular cycles. Of the five studies providing data for this analysis, three [24,30,47] included participants with baseline overweight or obesity while the baseline BMI of the participants in the remaining two [26,56] were unspecified. This finding aligns with that of another systematic review of two studies involving women with PMOS and obesity, which found similarly improved menstrual cycles with lifestyle intervention OR 4.34 (95% CI [1.75, 10.78] p = 0.02) [59]. This substantial improvement in ovulatory function offers women with PMOS a compelling non-pharmacological first-line option, particularly reassuring given widespread fertility concerns. Unlike pharmacological treatments (combined oral contraceptives, insulin sensitizers, anti-androgens) carrying adverse effect risks, lifestyle interventions demonstrated reproductive benefits without medications, with three of five studies specifically including overweight/obese participants where PMOS cardiometabolic risk is highest [59].

Previous reviews have reported different effects of lifestyle intervention on weight loss in women with PMOS, likely due to differences in the inclusion criteria, intervention types and duration in the included studies. The 2019 Cochrane review involving nine studies and 353 women suggested that there was low-quality evidence that lifestyle intervention leads to better weight loss compared to minimal treatment MD −1.68 kg (95% CI [−2.66, −0.70]) [14]. This finding aligns with another systematic review that also reported beneficial effects of lifestyle interventions on weight management in women with PMOS and obesity, although that review was based on only three studies [59]. More recent syntheses, including a 2025 review of randomised controlled trials published in the last 10 years [60] and another review of observational and interventional studies identified in 2022 [16] have also generally supported modest improvements in weight. Our review did not find a statistically significant overall effect of lifestyle interventions on weight compared to minimal treatment which is consistent with findings of our previous review MD −1.02 (95% CI [−2.08, 0.04] p = 0.058) [17].

Importantly, subgroup analysis restricted to women with overweight or obesity found a significant reduction in weight with lifestyle interventions, suggesting that the effects are more pronounced in this population. These findings highlight the need to consider the impact of participant characteristics, study design, intervention type and duration when interpreting available evidence. Exercise, particularly moderate-to-vigorous aerobic exercise and combined aerobic-resistance training, most effectively reduces waist circumference and visceral adipose tissue in overweight and obese adults, producing clinically meaningful improvements in metabolic outcomes including insulin sensitivity, glycaemic control, lipid profiles, and overall cardiometabolic risk. These reductions occur, in part, through preferential lipolysis and oxidation of visceral fat, driven by its heightened responsiveness to exercise-induced catecholamines and myokines (such as IL-6) relative to subcutaneous depots. Higher-intensity protocols often amplify this effect in a dose-dependent manner [61].

The earlier review reported that lifestyle interventions improved waist circumference MD −1.32 cm (95% CI [−2.46, −0.18] 12 studies), waist/hip ratio MD −0.03 (95% CI [−0.05, −0.01] 6 studies), total cholesterol MD −0.15 mmol/L (95% CI [−0.26, −0.03] 12 studies) and low density lipoprotein cholesterol MD −0.15 mmol/L (95% CI [−0.28, −0.02] 12 studies) compared to minimal treatment in women with PMOS [1]. In this review, the significant beneficial effects of lifestyle interventions on waist circumference were corroborated in a larger analysis involving pooled data from 17 studies. However, the significant effects on waist/hip ratio total cholesterol and low-density lipoprotein cholesterol were only observed among the subgroup of women with baseline overweight or obesity in subgroup analysis. Reductions in waist circumference, fasting glucose and fasting insulin translate to clinically meaningful improvements in central adiposity and insulin resistance, key drivers of PMOS cardiometabolic risk.

While the 2019 Cochrane review found that lifestyle interventions may improve free androgen index (MD −1.11 (95% CI [−1.96, −0.26] 6 studies, 204 women, I2 = 71%, low-quality evidence)) [14], we found no significant overall effect on free androgen index or total testosterone. However, subgroup analyses demonstrate greater reductions in total testosterone (MD −0.25 nmol/L (95% CI [−0.37, −0.14] 9 studies, 475 women)) specifically among women with baseline overweight or obesity, indicating particular benefit for this population. Similarly, while the Cochrane review found lifestyle interventions to be effective in reducing total cholesterol (MD −0.14 mmol (95% CI [−0.25, −0.02] 9 studies, 331 women)) and low-density lipoprotein levels (MD −0.16 mmol (95% CI [−0.29, −0.03] 9 studies, 326 women)) [14], we found statistically significant beneficial effects only in the subgroup of participants with baseline overweight or obesity.

In our pooled analysis of 18 studies, lifestyle intervention significantly reduced fasting insulin levels. This finding aligns with, and expand on that of the earlier reviews that included 14 and 10 studies, which also reported significant reductions in fasting insulin MD of −1.8 pmol/L (95% CI [−3.10, −0.65]) [1] and MD −1.42 µU/mL (95% CI [−2.44, −0.39]) [14], respectively. In both the earlier review [1] and the 2019 Cochrane review [14], there were no differences in fasting glucose levels between the lifestyle intervention and minimal treatment groups in pooled analyses of 15 and 11 studies, respectively. With 22 included studies, we found that lifestyle intervention led to significant reductions in fasting glucose levels; however, this was not corroborated in a sensitivity analysis where studies with high risk of bias were excluded, suggesting caution in interpretation of this finding.

The findings of this review reinforce the importance of lifestyle interventions as a first-line approach for managing symptomatic women with PMOS. The observed improvements in menstrual regularity, central adiposity, fasting glucose and fasting insulin levels associated with lifestyle interventions highlight the potential benefits of incorporating these approaches as essential components of clinical PMOS care. These results support the 2023 International Evidence-based Guideline for the Assessment and Management of PCOS, which recommends lifestyle intervention as the initial step in management [1]. Clinicians should prioritise exercise-based programs for improving central adiposity and insulin resistance, and dietary interventions for glycaemic control, particularly in those with overweight or obesity where cardiometabolic benefits are most pronounced. Implementation requires multidisciplinary support (e.g., dietitians, physiotherapists) and digital tools for adherence, with shared decision-making to tailor strategies to patient preferences and barriers.

Our analyses indicate that different lifestyle interventions may be more effective for specific outcomes or population subgroups, underscoring the importance of personalised care. Diet modifications—encompassing a heterogeneous range of dietary approaches are suggested to be more appropriate in managing glucose levels, while exercise interventions, alongside which patients often adopted modest dietary restriction may be better suited for improving BMI and waist circumference and reducing insulin levels. Additionally, among PMOS subgroups with overweight and obesity, lifestyle interventions improved weight, fasting glucose, insulin, testosterone, and dyslipidaemia, suggesting particular benefit in this subgroup for managing cardiometabolic risk.

Taken together, these findings emphasise the need for future research into the mechanisms through which specific lifestyle strategies impact reproductive, anthropometric, and metabolic parameters in PMOS. Notably, 78% of the included studies in this review were assessed as having some concerns or high risk of bias. This underscores the need for better-designed and implemented trials, prioritise methodological rigour, including appropriate randomisation, blinding of treatment allocation, and standardised outcome reporting. Additional research is also needed to establish the effects of lifestyle interventions across the Rotterdam PCOS phenotypes (A: hyperandrogenism + ovulatory dysfunction + polycystic ovarian morphology; B: hyperandrogenism + ovulatory dysfunction; C: hyperandrogenism + polycystic ovarian morphology; D: ovulatory dysfunction + polycystic ovarian morphology), as responses may vary depending on underlying pathophysiology [19,62,63]. A more nuanced understanding of these differences can support the development of tailored and evidence-based approaches to care. Lifestyle interventions significantly improved menstrual regularity and reduced waist circumference, fasting glucose and insulin levels compared to minimal treatment, reinforcing their role as a first-line approach in PMOS management. These findings emphasise the importance of personalised care, with tailored lifestyle interventions to address the diverse needs of women with PMOS. Further research is needed to identify the most effective intervention components and combinations for specific outcomes, and their impacts on fertility and broader outcomes including psychological features in PMOS.