Response to Letters Regarding Article, “Erectile Dysfunction as an Independent Predictor of Future Cardiovascular Events: The Multi-Ethnic Study of Atherosclerosis”

In Response:

We thank Dr. Qiang Su et al and Dr. Li et al for their interest in our article.¹ In their letter, Dr. Qiang Su et al recommended further adjustment for potential confounders that could affect the association between erectile dysfunction (ED) and cardiovascular disease (CVD), including anti-ED drugs such as PDE5 inhibitors, antidepressants, hypoglycemic drugs, and additional adjustment for duration of diabetes mellitus. We repeated our analysis using each of the above-mentioned covariates, adding each to our fully adjusted Model 3, in order to examine their effects on the association between ED and CVD. The results of the Cox regression remained similar to our current results for each covariate [hazard ratio for ED (95% confidence interval)]: PDE5 inhibitors: hard CHD 1.84 (0.85–3.98), hard CVD 1.90 (1.07–3.39); anti-depressants, which included selective serotonin reuptake inhibitors, tricyclic antidepressants, and non-tricyclic antidepressants other than monoamine oxidase inhibitors: hard CHD 1.86 (0.86–4.02), hard CVD 1.91 (1.07–3.42); duration of diabetes mellitus (0–4 years, 4–8 years and >8 years): hard CHD 1.88 (0.86–4.12), hard CVD 1.90 (1.06–3.40); diabetic medications: hard CHD 1.84 (0.85–4.02), hard CVD 1.91 (1.07–3.42). With our existing published analysis, augmented by the above additions, we believe we have performed one of the most thorough adjustments for confounders of any ED/CVD analysis, thus making a strong case for the ‘independence’ of ED in the prospective association with CVD.

In their letter, Dr. Li et al mentioned that questionnaire-based diagnoses - including the single Massachusetts Male Aging Study (MMAS) question - does not discriminate the pathophysiological causes of ED. We agree with Dr. Li et al, and would point out that this limitation is detailed in our paper: “Although similar to the primary care assessment of ED, the single Massachusetts Male Aging Study question does not distinguish between vascular and nonvascular types of ED, which may have attenuated the association between ED and CVD.” We also note that O’Donnell et al² assessed the validity of the single MMAS ED question compared to the gold standard urologic examination. The urologic examination consisted of four main components: detailed sexual history, medical history, psychosocial history, and a physical examination. O’Donnell et al showed that the MMAS ED question correlated strongly (Spearman r = 0.80) with the clinical urologic examination. Though well correlated with the urologic exam, and strongly correlated with markers of subclinical vascular injury and dysfunction,³ the single question does not confer an ED diagnosis and cannot distinguish between different ED types. However, this limitation would be expected to bias the association of ED with CVD towards the null. Despite the limitation, a single question of ED can be useful for screening patients in a cardiology office setting or in population-based studies where clinical examination of ED might not be practical.