Association between metabolic obesity phenotypes and erectile dysfunction: Insights from a nationally cross-sectional study

In 3,767 men, metabolically unhealthy obesity roughly tripled ED odds, but obesity alone with no metabolic issues didn't

Journal: Medicine | Published: 2026-07-03 | Type: Journal Article | PMID: 42410782 Authors: Fang Cheng, Zhou Tianyu, Fan Xiaoxiao, Hu Xiaowei, Xue Yiyang, Huang Jiawen, Xu Zhipeng (all Department of Urology, Ningbo Hospital of Integrated Traditional Chinese and Western Medicine) Funding/COI: No funding listed; authors report no conflicts of interest

Summary

Cheng et al., 2026 used NHANES 2001-2004 data on 3,767 men to sort them into four groups by BMI and metabolic health, then checked which combination predicted self-reported erectile dysfunction. Metabolically unhealthy obese (MUO) men had the highest ED prevalence at 36.1%, versus 13.6% in metabolically healthy nonobese men. The key finding: obesity by itself, without hyperglycemia, hypertension, or dyslipidemia, showed no significant bump in ED odds, whereas hyperglycemia and obesity combined did.

Claims

• Overall ED prevalence was 28.2% (1,062 of 3,767 men)
• ED prevalence by phenotype: MUO 36.1%, metabolically unhealthy nonobese (MUNO) 31.5%, metabolically healthy obese (MHO) 14.6%, metabolically healthy nonobese (MHNO) 13.6%
• Of individual risk components, only obesity (OR 1.457, 95% CI 1.092-1.944) and hyperglycemia (OR 1.795, 95% CI 1.404-2.294) were independently associated with ED after adjustment
• In men aged 41-60, all three non-reference phenotypes carried higher ED odds versus MHNO: MHO (OR 1.953, 95% CI 1.017-3.749), MUNO (OR 2.126, 95% CI 1.054-4.288), MUO (OR 2.784, 95% CI 1.519-5.104)
• ED prevalence rose progressively with the number of metabolic risk components in both obese and nonobese men (trend P < .001 for both), topping out at 45.3% in obese men with all three components
• No significant ED prevalence difference between MHO and MHNO in the overall sample

Study Quality

This is a cross-sectional analysis of a nationally representative, complex-sampled survey (NHANES), with weighted multivariable logistic regression adjusting for age, race/ethnicity, income, education, marital status, smoking, alcohol use, and CVD history. The metabolic phenotype definitions (BMI ≥30 for obesity, ≥2 of 3 MetS components for "unhealthy") are standard and clearly specified, and the dose-response gradient across risk component counts is a reasonably convincing internal consistency check.

But this is associational data from a single self-reported ED question, collected over 20 years ago (2001-2004, the only NHANES cycles with an erectile function questionnaire), and the paper draws no causal conclusion despite language throughout ("targeting both obesity and metabolic abnormalities could be critical in mitigating ED risk") that leans that direction. The age-stratified subgroup analysis (20-40, 41-60, presumably 60+) multiplies comparisons without a stated correction, and several of the significant findings in the 41-60 subgroup have confidence intervals that barely clear 1.0 (MHO: 1.017-3.749) — not a robust effect.

Red Flags

• Cross-sectional design cannot establish causality, but discussion language ("should be considered for individuals with ED") edges toward advice framing
• Data are 20+ years old; NHANES stopped asking the ED question after 2004, so this can't be updated with more recent cycles
• Single self-reported item defines ED status; despite being described as "previously validated," it's still one question, not a validated multi-item instrument like the IIEF
• No adjustment for testosterone levels, depression/psychological factors, or medications (e.g., antihypertensives, statins) that independently affect erectile function
• Multiple age-stratified subgroup comparisons without correction for multiple testing; several significant results have CIs that barely exclude 1.0
• PAF (population attributable fraction) analysis is referenced in the discussion but not detailed in the excerpted results

Strengths

• Nationally representative, weighted sampling design (NHANES)
• Reasonably large sample (3,767 men) for a subgroup-stratified analysis
• Clear, standard operational definitions for obesity and metabolic health components
• Demonstrates a dose-response relationship between number of metabolic risk components and ED prevalence, which strengthens the internal validity of the association
• Phenotype-based approach separates obesity's effect from metabolic dysfunction's effect, addressing a real gap where prior studies conflated the two

Verdict

A methodologically competent but unsurprising cross-sectional study: it confirms that metabolic dysfunction, not body size alone, tracks with self-reported ED, using old NHANES data that can't be refreshed. The dose-response pattern across risk components is its most credible contribution. The age-subgroup findings are underpowered and shouldn't be over-read given wide, borderline confidence intervals. Useful as a data point, not a foundation for anything beyond "obesity and glycemic dysfunction correlate with ED status in this 2001-2004 sample."