Pooled data from 3,648 hypertensive African men found 65% report erectile dysfunction, but studies barely agree
Journal: Sexual Medicine Reviews | Published: 2026-06-30 | Type: Systematic Review, Meta-Analysis | PMID: 42447425 Authors: Gebrie Eshet, Chane Elias, Mengistie Berihun A, Kibralew Getasew, Worku Henok, Mulugeta Birhan, Zayede Amanuale (Ethiopian university departments of clinical chemistry, midwifery, psychiatry, immunology and medical microbiology) Funding/COI: Not listed
This meta-analysis pooled 18 studies covering 3,648 hypertensive men across three African regions and calculated a pooled ED prevalence of 65.05% (95% CI: 54.23-75.87%). Older age, longer hypertension duration, stage II hypertension, taking multiple blood pressure drugs, comorbidities, and depression were all associated with higher odds of ED. The headline number comes with a massive caveat: heterogeneity between studies was almost total (I2 = 98.39%), meaning the underlying studies measured wildly different things.
The authors followed a standard meta-analysis workflow: independent dual extraction, Joanna Briggs Institute critical appraisal, random-effects pooling, and funnel plot/Egger's test for publication bias. Subgroup and sensitivity analyses were run to probe sources of variation. That's a reasonable skeleton for a review of this kind.
The problem is what that skeleton is built on. An I2 of 98.39% is about as high as heterogeneity gets in meta-analysis - it means almost none of the variance in ED prevalence across the 18 studies is due to sampling error, and almost all of it is due to real differences between the studies themselves (different populations, different tools, different sampling methods, different hypertension severity mixes). A random-effects model still produces a pooled number in that situation, but the number describes an average of incompatible populations rather than a stable underlying rate. The subgroup findings (Central Africa highest, SHIM highest, convenience sampling highest) are essentially an admission of this: prevalence tracks methodology, not just biology. Convenience sampling in particular is prone to selecting men who already sought care for sexual complaints, which inflates prevalence independent of the true population rate.
This is a competently executed systematic review that answers a real question - there was no continent-wide estimate before - but the near-total heterogeneity (I2 = 98.39%) means the 65% headline number is more a summary of methodological chaos across 18 studies than a trustworthy population statistic. The associated-factor odds ratios (age, hypertension severity and duration, polytherapy, depression) are more interpretable since they're less sensitive to the prevalence-inflating effects of convenience sampling, but they still rest on cross-sectional data that can't establish direction of causation. Worth reading for the associated-factor patterns and as a map of where better primary research is needed; not worth citing for a single prevalence number.