Systematic Review and Meta-Analysis on the Beneficial Effect of Exercise on Sexual Dysfunction in Women

Exercise interventions produced a mean 5.31-point FSFI improvement vs. controls (15 studies), but evidence certainty is rated low.

Journal: The Journal of Sexual Medicine | Published: 2026-05-09 | Type: Systematic Review, Meta-Analysis | PMID: 42114124 Authors: Ruiz-Pérez I, Ruiz-Rios M, Cuesta-Ibáñez S, Maldonado-Martín S (University of the Basque Country, Faculty of Education and Sport) Funding/COI: Not disclosed

Summary

A meta-analysis of 15 RCTs found a statistically significant 5.31-point improvement on the Female Sexual Function Index (FSFI) from exercise versus inactive controls (95% CI: 4.87–5.75), with gains across all six subdomains. The catch: heterogeneity was extreme (I² = 76–94%), and the authors' own GRADE assessment rated overall evidence certainty as low. The headline numbers are real; what they mean is genuinely unclear.

Claims

• Mean FSFI difference vs. control: +5.31 points (95% CI: 4.87–5.75) across 15 studies
• All six FSFI subdomains improved: desire, arousal, lubrication, orgasm, satisfaction, dyspareunia (subdomain range: 0.59–0.96 points)
• 18 of 23 included studies reported significant between-group differences; 20 of 23 showed within-group improvement after exercise
• Exercise types included: pelvic floor, aerobic, resistance, combined, yoga, aquatic, and core training (minimum intervention: 4 weeks)
• Prevalence figure cited: sexual dysfunction affects >43% of women worldwide
• PROSPERO pre-registered: CRD42023492467

Study Quality

The methodological scaffolding is solid: four databases searched through August 2025, pre-registration on PROSPERO, validated outcome measure (FSFI) required for inclusion, modified PEDro scale for trial quality, and GRADE for evidence certainty. Individual study quality was rated predominantly "good to excellent." That's the good news.

The bad news is the heterogeneity. I² of 76–94% across subdomains is not just "high"—it signals that these studies may be measuring fundamentally different things. Pelvic floor training in postoperative gynecological patients, yoga in perimenopausal women, and aerobic exercise in young healthy adults are not interchangeable interventions, yet they're pooled into a single 5.31-point estimate. The authors acknowledge this explicitly in their GRADE downgrade, but the abstract's framing ("significant benefits") buries the lede.

Red Flags

• Extreme heterogeneity (I² = 76–94%) undermines the pooled effect size—the confidence interval is tight, but the studies behind it are not measuring the same thing
• GRADE certainty: LOW—the paper's own evidence grading contradicts the enthusiastic conclusion
• Funding and COI not reported—standard disclosure, but the absence is notable for a journal article in 2026
• 23 exercise types collapsed into one category—pelvic floor exercises and aquatic therapy have distinct physiological mechanisms; treating them as one intervention inflates apparent generalizability
• No baseline FSFI values reported in the abstract—a 5.31-point improvement looks different depending on whether patients started at 15 (dysfunctional) or 22 (borderline); FSFI dysfunction threshold is typically ≤26.55
• Narrative synthesis for non-FSFI outcomes—heterogeneity prevented meta-analysis of non-FSFI validated measures, shrinking the quantitative base further

Strengths

• Pre-registered protocol (PROSPERO) reduces outcome-reporting bias
• Comprehensive search across four major databases
• Restricted meta-analysis to FSFI only—good discipline, avoids pooling incompatible scales
• Modified PEDro assessment adds transparency on trial-level risk of bias
• Large span of exercise modalities; if even heterogeneous programs improve outcomes, the signal may be robust
• Covers a clinically neglected area—female sexual dysfunction research is chronically underfunded relative to male ED

Verdict

The 5.31-point FSFI improvement sounds definitive until you notice the authors themselves rated their evidence as low certainty—and that I² topping 90% in some subdomains means the pooled number is partly a statistical artifact of averaging wildly different populations, conditions, and interventions. This paper is useful as a landscape survey: yes, exercise probably helps women's sexual function in some contexts; no, we can't say which exercise, for whom, or by how much. It's a well-constructed review of a messy literature. The conclusion oversells; the methods section tells you exactly why it shouldn't.