Skip main navigation

Prediction of Cardiovascular Events and All-Cause Mortality With Erectile Dysfunction

A Systematic Review and Meta-Analysis of Cohort Studies
Originally publishedhttps://doi.org/10.1161/CIRCOUTCOMES.112.966903Circulation: Cardiovascular Quality and Outcomes. 2013;6:99–109

Abstract

Background—

Erectile dysfunction (ED) carries an independent risk for cardiovascular (CV) events. We conducted a meta-analysis of all longitudinal studies for determining the ability of ED to predict risk of clinical events and to dissect factors influencing this ability.

Methods and Results—

We conducted a comprehensive search of electronic databases through July 2012. Longitudinal studies that reported relative risk (RR) estimates with 95% confidence intervals (CIs) were included. Of the 14 studies included (92 757 participants; mean follow-up, 6.1 years; 16 articles), 13 (14 articles) reported results on total CV events (91 831 individuals), 4 on CV mortality (34 761 individuals), 4 on myocardial infarction (35 523 individuals), 6 on cerebrovascular events (27 689 individuals), and 5 on all-cause mortality (17 869 individuals). The pooled RRs for the above-mentioned end points were 1.44 (95% CI, 1.27–1.63), 1.19 (95% CI, 0.97–1.46), 1.62 (95% CI, 1.34–1.96), 1.39 (95% CI, 1.23–1.57), and 1.25 (95% CI, 1.12–1.39), respectively, for men with versus without ED. The RR was higher in intermediate- compared with high- or low-CV-risk populations and with younger age. The RR for studies that diagnosed ED with the use of a questionnaire compared with a single question was higher (RR, 1.61; 95% CI, 1.38–1.86 versus RR, 1.27; 95% CI, 1.18–1.37, respectively; P=0.006).

Conclusions—

ED is associated with increased risk of CV events and all-cause mortality. RR is higher at younger ages, in intermediate-risk groups, and when a questionnaire is used instead of a single question.

Introduction

Erectile dysfunction (ED) is a common clinical problem worldwide with thousands of new cases each year.1,2 In the Massachusetts Male Aging Study, the prevalence of ED was 52% in men 40 to 70 years of age and 70% in men >70 years of age.3 Cardiovascular (CV) disease (CVD) and ED share common risk factors,4 whereas evidence-based studies have identified pathophysiological links between ED and other vascular diseases, such as endothelial dysfunction and inflammation,5,6 thus identifying ED as an appealing candidate marker for future CV events. Screening and diagnosing ED could be of great importance for primary prevention because ED assessment offers an easy, low-cost alternative to several investigational CV biomarkers and could describe the risk over and beyond traditional risk factors, particularly for those patients belonging to the intermediate-CV-risk category. ED may precede clinically overt CVD by 2 to 5 years,6 providing a valuable time window for earlier modification of risk factors and potential improvement in outcomes. Importantly, studies have implicated that improvement of ED and ED-associated comorbidities by either pharmacological or lifestyle interventions might be beneficial in terms of prognosis.6

A number of studies examined the ability of ED to predict the risk of future fatal and nonfatal CV events and all-cause mortality.729 Although there is a general impression that ED has an important predictive role based on the results of individual studies,729 the overall quantitative estimate of this role has not been fully clarified,30,31 and implementation in clinical practice is suboptimal. In addition, because most studies yielded positive results, publication bias may have been involved. Finally, an important issue is whether the predictive ability of ED extends beyond CV events. Accordingly, we conducted the present study to calculate robust quantitative estimates of the predictive value of ED for different outcomes. Second, we investigated whether publication bias could have affected the true predictive ability of ED. Third, we evaluated clinically meaningful issues such as the effect of different baseline CV risk factors and of different diagnostic methods for ED (eg, questionnaire versus single question) on the predictive ability of ED. Toward this end, we included new landmark studies20,27 that have advanced our knowledge of the predictive value of ED for CVD. The robustness of our findings is ensured by the overall population size of the present meta-analysis (92 757 subjects) and by the lengthy follow-up.

WHAT IS KNOWN

  • Cardiovascular disease and erectile dysfunction share common risk factors, whereas evidence-based studies have identified pathophysiological links such as endothelial dysfunction and inflammation, thus identifying erectile dysfunction as an appealing candidate marker for future events.

  • Although there is a general impression that erectile dysfunction has an important predictive role based on the results of individual studies, the overall quantitative estimate of this role has not been fully clarified.

WHAT THE STUDY ADDS

  • The presence of erectile dysfunction increases the risk for future cardiovascular events, myocardial infarction, cerebrovascular events, and all-cause mortality, whereas it shows a trend to increase risk for cardiovascular mortality.

  • Relative risk is higher at younger ages, in intermediate-risk groups, and when a questionnaire is used instead of a single question.

Methods

The meta-analysis was conducted according to the checklist of the Meta-analysis of Observational Studies in Epidemiology.32 Outcomes of interest were (1) total CV events, (2) CV mortality, (3) myocardial infarction (MI), (4) cerebrovascular events (stroke, transient ischemic attacks, intracranial hemorrhage), and (5) all-cause mortality. Total CV events were defined as CV death, MI, revascularization, cerebrovascular events, peripheral vascular disease, angina, heart failure, and arrhythmia.

Data Sources and Searches

Studies were drawn from a systematic review of the literature in the PubMed, Cochrane, and Embase databases until July 2012. The search terms are cited in the online-only Data Supplement. Data sources were also identified through a manual search of the references of articles.

Study Selection

Studies were deemed eligible on the following conditions: (1) if they were full-length publications in peer-reviewed journals; (2) if they evaluated ED (for details, see Table I in the online-only Data Supplement); (3) if they reported a combined CV outcome or a separate outcome such as CV mortality, MI, cerebrovascular event and all-cause mortality; (4) if the follow-up period was at least 1 year; and (5) if they were longitudinal cohort studies, either retrospective or prospective. Studies with cancer patients were excluded from our meta-analysis. Otherwise, no restriction criteria were imposed with regard to the type of the population studied (general population or populations with risk factors or disease) or to the size of the population. All but 3 longitudinal studies (4 articles)7,8,11,26 included in the meta-analysis were prospective studies. Retrospective studies were intentionally included (see Discussion for rationale).

Data Extraction and Quality Assessment

The literature search, selection of studies, quality assessment (for details, see the online-only Data Supplement), and extraction of data were done independently by 2 reviewers (C.V.V., D.G.T-P.). Disagreements were resolved by consensus. For each study, we recorded a risk estimate for ED. Numeric data appearing in the articles were used.

Data Synthesis and Analysis

The risk estimates of each study were reported as hazard ratio, relative risk (RR), odds ratio, or dichotomous frequency data. We treated hazard ratios as RRs. When available, we used the adjusted-risk estimates from multivariate models. Of all available multivariate models, we chose the one that had adjusted for the largest number of possible clinically relevant confounding factors. We performed meta-analyses of studies investigating ED to obtain the pooled RRs separately for the following: (1) total CV events, (2) CV mortality, (3) MI, (4) cerebrovascular events, and (5) all-cause mortality. The proportion of inconsistency across studies not explained by chance was quantified with the I2 statistic. Heterogeneity between subgroups was calculated with the Cochran Q test.33 When significant heterogeneity existed among studies (P<0.05), the random-effects model was used to obtain the pooled RRs. We also calculated adjusted RRs of ED versus no ED groups in each study. We performed a sensitivity analysis to evaluate whether the strength of risk estimates differs between intermediate-risk groups (subjects with a 10-year risk of CVD between 10% and 20% based on Framingham Risk Score) and high-risk groups (subjects with CVD or diabetes mellitus and a 10-year risk of CVD >20% based on the Framingham Risk Score) or low-risk groups (subjects with a 10-year risk of CVD <10% based on the Framingham Risk Score) and between different methods of ED diagnosis (single question versus questionnaire). Risk estimates in subgroups were compared by use of a test of interaction.34 The RRs and confidence intervals (CIs) of individual studies were illustrated with forest plots. To estimate the contribution of continuous study moderators to the overall heterogeneity, we ran a meta-regression analysis with fixed-effects estimates. When pooled RRs were statistically significant, the presence of publication bias was investigated graphically by funnel plots of precision, and its implications for our results were assessed by the Duval and Tweedie trim-and-fill method35 and the classic fail-safe N method as introduced by Rosenthal.36 All analyses were performed with comprehensive meta-analysis version 2 (Biostat, Englewood, NJ).

Results

Literature Search

Literature search results are shown in Figure 1. We retrieved 8335 articles from our preliminary search. Of these, 23 were identified for full review. For details on the exclusion of studies10,2225,28,29 after full review, see the online-only Data Supplement.

Figure 1.

Figure 1. Flowchart of selection of studies for inclusion in meta-analysis. CVD indicates cardiovascular disease; and ED, erectile dysfunction.

Study Characteristics

Our meta-analysis included 16 original articles. In total, the included studies analyzed 92 757 subjects. Several populations such as patients with hypertension, diabetes mellitus, or coronary artery disease and subjects from the general population were included. Details of the individual studies are shown in Table 1. Of the 14 studies included (92 757 participants; mean follow-up, 6.1 years; 16 full-text articles),79,1121,26,27 13 (14 full-text articles)79,1115,1821,26,27 reported results on total CV events (91 831 individuals; 5005 incident cases during 471 636 person-years of follow-up), 4 reported on CV mortality (34 761 individuals),14,16,20,27 4 reported on MI (35 523 individuals),8,9,14,20 6 reported on cerebrovascular events (27 689 individuals),9,14,18,20,21,26 and 5 reported on all-cause mortality (17 869 individuals).9,16,17,20,21 All studies were published since 2003, and the mean/median follow-up ranged from 1 year7,8 to 15.2 years.16 The sample sizes ranged from 29114 to 31 29627 individuals. Age and other risk factors for CVD were controlled for in most of the studies.

Table 1. Overview of Studies on the Association Between ED and Clinical End Points

First Author, Year (Ref. No.)Population (Sample Size)Mean Age (SD) or Age Range, yFollow-up Duration, yED PopulationEventsDiagnosis of EDAdjusted forQuality Analysis*
Blumentals et al,7 200325 650 men40.1112 825 men (50%)76 cases of PVDED was diagnosed with ICD-9 codesAge at ED diagnosis, smoking, obesity and use of ACE inhibitors, β-blockers, and statinsY Y N
Blumentals et al,8 200425 650 men40.1112 825 men (50%)72 cases of MIED was diagnosed with ICD-9 codesAge at ED diagnosis, smoking, obesity, and use of ACE inhibitors, β-blockers, and statinsY Y N
Thompson et al,9 20058063 men with no CVD62±67 y ±90 d3816 men (47.3%)CVD as any of the following events: MI or surgical treatment of CAD, including CABG or angioplasty, angina, cerebrovascular accident, TIA, HF graded at a minimum of mild, fatal cardiac arrest, or nonfatal cardiac arrhythmia (955: first CV event, 382: death by any cause)ED was graded as follows: grade 0, absent; grade 1, decrease in normal function but ability to achieve vaginal penetration with difficulty; or grade 3, no erections. ED was defined as the first report of ED of any gradeAge, BMI, SBP, DBP, total cholesterol, HDL, history of diabetes mellitus, parent or sibling with a history of MI, race (white vs other), current smoking, current use of antihypertensive medication, physical activity (moderate or very active vs sedentary or light), and global, self-reported health status (excellent or very good vs fair or poor)N N N
Frantzen et al,11 20061183 men before the introduction of sildenafil35–74Up to 2278 men38 incident CVD casesED reported by general practitionersNo adjustmentY Y N
Schouten et al,12 20081248 men in a community-based study60.676.33392 men (31.4%, 284 with reduced rigidity and 108 with severely reduced rigidity)58 cardiovascular events (39 MIs, 5 sudden deaths, or 14 strokes)α single question on erectile rigidity included in the International Continence Society male sex questionnaireAge, total cholesterol, HDL, SBP, diabetes mellitus, and smokingN N N
Ma et al,13 20082306 diabetic men54.2±12.74.0616 men (26.7%)123 CHD events (MI or death resulting from coronary cause; or other nonfatal CHD)A single question on the inability to attain and maintain penile erection sufficient for satisfactory sexual performanceAge, duration of diabetes mellitus, SBP, DBP, albuminuria, retinopathy at baseline, eGFR, use of lipid-lowering agents, use of antihypertensive medications, use of ACE inhibitors/ARBsN N N
Gazzaruso et al,14 2008291 type 2 diabetic men with silent CAD documented with angiography54.8±7.33.9±1.8118 men (40.5%)49 MACE: CAD death (3), sudden death (2), nonfatal MI (14), death resulting from HF (1), unstable angina (8), need for repeat revascularization (3, aside from restenosis), stroke or TIA (16), and symptomatic PVD (2) documented by angiographyIIEF-5 questionnaireAge, diabetes mellitus duration, hypertension, family history of CAD, smoking, microalbuminuria, glycohemoglobin, BMI, cholesterol, triglycerides, LDL, HDL, and autonomic dysfunctionN N Y
Inman et al,15 20091402 community-dwelling men55.410158 men (11.3%)156 (23 MI, 123 angiographic anomalies, and 10 sudden cardiac death)Brief male sexual function inventoryDiabetes mellitus, hypertension, history of smoking, and BMIY N N
Araujo et al,16 2009§1655 men in population-based study55.215.2338 men (20.4%)CVD death (140, includes CHD, HF, PVD, cerebrovascular disease, and other vascular diseases), death resulting from malignant neoplasms (124), and death resulting from other causes (107)23-item questionnaire on sexual activityAge, BMI as continuous variables, and the following as categorical variables: alcohol consumption (<1, 1, and ≥2 drinks/d), calories expended in physical activity (none, <200 kcal/d, and ≥200 kcal/d), current smoking, self-assessed health (excellent, very good, good, fair/poor), and self-reported chronic disease (heart disease, hypertension, and diabetes mellitus)N N N
Hebert et al,17 2009328 male HF patients55.48.4293 men (89.3%)All-cause mortality (96 deaths)SHIM...Y N N
Ponholzer et al,18 20102506 men without a history of CHD or cerebral vascular disease45±126.5870 men (34.7%, 634 with mild ED, 236 with moderate to severe ED)58 CVD eventsIIEF-5 questionnaireAgeY N N
Araujo et al,19 2010§1057 men free of CVD and diabetes mellitus5411.7178 men (16.8%)261 new cases of CVD23-item questionnaire on sexual activityBMI (continuous) and the variables that are part of the Framingham risk score: age, HDL, and total cholesterol (all as continuous variables), as well as current smoking (yes/no), and hypertension categorized according to BP readings (optimal, normal, high normal, stage I, and stage II to IV)N N N
Bőhm et al,20 20101519 high-risk CVD patients64.84.4 (ONTARGET) and 4.5 (TRANSCEND)842 men (55.4%)A composite (primary outcome) of death resulting from CV causes, MI, stroke, or hospitalization for HF (206). Other outcomes were death resulting from any cause (133)The IIEF-5 and the Kőlner (Cologne) evaluation of ED scoresAge, SBP, DBP, smoking, history of hypertension, diabetes mellitus, MI, stroke/TIA, PVD, alcohol consumption, use of β-blockers and calcium channel blockers, ankle-brachial index, and lower urinary tract surgeryN N N
Batty et al,21 20106304 type 2 diabetic men65.953158 men (50.1%)Deaths resulting from any cause (695), CV events (1549), CHD events (773), cerebrovascular events (411),dementia events (58), cognitive decline events (1013)Nurses asked subjects whether they had ED (categorized as yes or no)Treatment, age, use of metformin or β-blockers, history of macrovascular or microvascular disease, or those requiring assistance with daily activities, plus diabetes mellitus duration, cigarette smoking, alcohol intake, and vigorous physical activity in previous week, glycosylated hemoglobin, creatinine, BMI, total cholesterol, HDL, resting heart rate, SBP, DBP, quality of life (EQ-5D score) and Mini-Mental State Examination score, age at completion of highest level of education and height, treatment allocation, and ethnicityN N N
Chung et al,26 2011 9006 men in a nationwide population-based study58.5±11.451501 men (16.7%)Strokes (918)ICD-9 codes for EDPatient’s monthly income, geographical location, hypertension, PVD, diabetes mellitus, CHD, atrial fibrillation, and hyperlipidemiaY Y N
Hotaling et al,27 201231 296 men in western Washington627.87762 men (24.8%)486 CV deathsAnswering yes to the question “Have you experienced impotence in the last year?”Age, ED status, marital status, race, education, self-rating of health, antihypertensive drug use, lipid-lowering drug use, family history of CAD, current smoking status, current/former pack-years of smoking, BMI at 45 y of age, past and current exercise, diagnosis of chronic kidney disease, insulin use, oral hypoglycemic use, aspirin use, fruit/vegetable intake, and percentage of calories from saturated fatN N N

ACE indicates angiotensin-converting enzyme; ARB, angiotensin II receptor blockers; BMI, body mass index; CABG, coronary artery bypass graft; CAD, coronary artery disease; CHD, coronary heart disease; CVD, cardiovascular disease; DBP, diastolic blood pressure; ED, erectile dysfunction; eGFR, estimated glomerular filtration rate; HDL, high-density lipoprotein; HF, heart failure; ICD, International Classification of Diseases; IIEF, international index erectile function; LDL, low-density lipoprotein; MACE, major adverse cardiac events; MI, myocardial infarction; PVD, peripheral vascular disease; SBP, systolic blood pressure; SHIM, sexual health inventory for men; and TIA, transient ischemic attack.

Studies have the same population but different clinical end points. The Blumentals et al 2003 study was used for estimation of total CV events and myocardial infarction, and the Blumentals et al 2004 study was used for estimation of total CV events.

Retrospective studies.

§Studies have a part of their population in common. The Araujo et al 2009 study (which is larger) was used for estimation of CV and all-cause mortality, and the Araujo et al 2010 study was used for estimation of total CV events.

Meta-analysis

ED and Total CV Events

The magnitude of risk in individuals who had ED was significantly higher compared with the risk of individuals without ED. The pooled RRs for ED were 1.44 (95% CI, 1.27–1.63) for total CV events (Figure 2A). By applying a sensitivity analysis, we excluded the 3 retrospective studies without significant changes in our final results for total CV events (RR, 1.42; 95% CI, 1.22–1.65; P<0.001).

Figure 2.

Figure 2. Relative risk (RR) and 95% confidence interval (CI) for erectile dysfunction (ED) and clinical events. RR and 95% CI for ED and total cardiovascular (CV) events (A), CV mortality (B), myocardial infarction (C), cerebrovascular events (D), and all-cause mortality (E). Studies are listed alphabetically. Boxes represent the RR and lines represent the 95% CI for individual studies. The diamonds and their width represent the pooled RRs and the 95% CI, respectively. CVD indicates cardiovascular disease; DM, diabetes mellitus; GEN, general population; and HF, heart failure.

Because we observed significant heterogeneity (I2=66.4%, P<0.001) among the included studies, we conducted between-study subgroup analyses to investigate its sources. The RR was significantly different across studies with different baseline CV risk populations. (P=0.003; Figure 3). Specifically, the RR for ED was significantly higher in intermediate-risk9,12,15,19 populations compared with high-risk13,14,20,21 and low-risk18,27 populations (RR, 1.51; 95% CI, 1.35–1.70 versus RR, 1.30; 95% CI, 1.20–1.42; P=0.048; and RR, 0.93; 95% CI, 0.72–1.19; P=0.001, respectively). Moreover, the RR for ED was significantly higher in high-risk populations compared with low-risk populations (P=0.011).

Figure 3.

Figure 3. Pooled relative risk (RR) and 95% confidence interval for erectile dysfunction and total cardiovascular (CV) events, according to CV risk assessed by Framingham Risk Score. The center circle of the bar denotes the pooled RR; the extremes of the bars, the upper and lower limits of 95% of the data. Value by test for interaction.

In terms of the method of ED diagnosis, the RR for ED was higher in studies in which ED was diagnosed with a questionnaire14,15,1820 compared with a single question9,12,13,21,27 (RR, 1.61; 95% CI, 1.38–1.86 versus RR, 1.27; 95% CI, 1.18–1.37; P=0.006).

To further investigate the incremental predictive role of ED above and beyond conventional risk factors, we performed a sensitivity analysis in which we included studies9,12,14,19,21,27 that had adjusted for age, smoking, diabetes mellitus, cholesterol, and hypertension/blood pressure (as opposed to the remainder of the studies that had adjusted for only some of those parameters). The RR in these studies (RR, 1.41; 95% CI, 1.16–1.71; P<0.001) was similar to the overall combined estimated risk. Moreover, we applied a sensitivity analysis in which we included only high-quality studies as assessed by our quality analysis9,12,13,1921,27 with similar results (RR, 1.34; 95% CI, 1.17–1.54; P<0.001).

ED and CV Mortality, MI, and Cerebrovascular Events

The magnitude of risk for most clinical end points in individuals who had ED was significantly higher compared with the risk of individuals without ED, except for CV mortality, which showed a trend toward significance (P=0.089). The pooled RRs for ED were 1.19 (95% CI, 0.97–1.46), 1.62 (95% CI, 1.34–1.96), and 1.39 (95% CI, 1.23–1.57) for CV mortality, MI, and cerebrovascular events, respectively (Figure 2B–2D).

ED and All-Cause Mortality

The magnitude of risk in individuals who had ED was significantly higher compared with the risk in individuals without ED. The pooled RR for ED was 1.25 (95% CI, 1.12–1.39) for all-cause mortality (Figure 2E).

Because we observed heterogeneity (I2=31.9%; P=0.204) among the included studies, we conducted between-study subgroup analyses. The RR of ED patients for all-cause mortality was significantly higher in patients with known CVD17,20 compared with patients without (predominantly) known CVD9,16,21 (RR, 1.90; 95% CI, 1.31–2.77 versus RR, 1.20; 95% CI, 1.07–1.34; P=0.021).

Publication Bias

The funnel plots were asymmetrical at the bottom (Figure I in the online-only Data Supplement) for all end points, suggesting an absence of small studies with small or negative risk estimates in our meta-analysis. The trim-and-fill method imputed missing studies and recalculated our pooled risk estimate (Figure IA–ID in the online-only Data Supplement). The imputed RR was 1.32 (95% CI, 1.16–1.51), 1.52 (95% CI, 1.28–1.80), 1.37 (95% CI, 1.21–1.54), and 1.20 (95% CI, 1.08–1.33) for total CV events, MI, cerebrovascular events, and all-cause mortality, respectively, which are lower than our original risk estimates but are still significant. Importantly, the results of the fail-safe N test of our pooled analysis are 300, 21, 34, and 22 respectively, which for all end points are reassuring. The fail-safe N test computes the number of missing studies (with a mean effect of zero) that would need to be added to the analysis to yield a statistically nonsignificant overall effect, and it is unlikely that there are >23 (300/13=23.1), >5 (21/4=5.3), >5 (34/6=5.7), and >4 (22/5=4.4) unpublished or undiscovered studies for every 1 study that we found for total CV events, MI, cerebrovascular events, and all-cause mortality, respectively. These findings suggest that the apparent publication bias is insufficient to affect our results or interpretations in a meaningful way.

Meta-regression Analysis

Age at enrollment was the strongest predictor of the magnitude of the log RR in ED patients and inversely related to the predictive role of ED for total CV events (P<0.001; Figure 4A). Duration of follow-up was not a predictor (P=0.52). The percentage of smokers and total cholesterol in each study showed positive associations with the predictive role of ED (P=0.003 and P=0.001, respectively), and pulse pressure, HDL, and BMI showed inverse associations with the predictive role of ED (P=0.03, P=0.01, and P=0.02, respectively), whereas inverse associations of percentage of diabetics and systolic blood pressure with the predictive role of ED were not statistically significant or showed a trend (P=0.29 and P=0.07, respectively; Figure 4A–4H). Exclusion of outliers had minimal effects on the above-mentioned results.

Figure 4.

Figure 4. Relative risk (RR) of total cardiovascular events in patients with erectile dysfunction as a function of (A) age (data from 13 studies79,1115,1821,26,27), (B) smoking percentage in study population (data from 10 studies79,1214,1821,27), (C) systolic blood pressure (data from 7 studies9,12,13,1821), (D) pulse pressure (data from 6 studies9,12,13,18,20,21), (E) cholesterol (data from 8 studies9,1214,1821), (F) high-density lipoprotein (data from 7 studies9,1214,18,19,21), (G) body mass index (data from 7 studies9,1214.18,19,21), and (H) diabetes mellitus percentage in study population (data from 10 studies9,1215,1821,27). Each study is represented by a circle that shows the actual coordinates (observed effect size by each of the above-mentioned variables) for that study. The size of each circle is proportional to the weight of the respective study in the analysis, that is, the inverse of the within-study variance for each study. The center line shows the predicted values by fixed-effects meta-regression. The vertical axis is on a log scale.

Discussion

In this systematic review and meta-analysis, we pooled the data for 92 757 subjects investigated for ED from 16 available published articles who were followed up for a mean of 6.1 years. Our study is the first to investigate in a thorough manner whether the presence of ED increases the risk for future events and to assess factors influencing such a predictive ability. Our principal finding is that patients with ED compared with subjects without ED have a significantly increased risk by 44% for total CV events, 62% for MI, 39% for cerebrovascular events, and 25% for all-cause mortality. The risk for CV mortality was also increased by 19%, but this increase only showed a trend toward significance. Importantly, the RR is higher in younger patients and in patients with intermediate CV risk. Furthermore, diagnosis of ED by a questionnaire is associated with a higher RR compared with diagnosis of ED with a single question. Finally, the RR increases with increased percentage of smokers, decreased pulse pressure, and deteriorated dyslipidemic profile.

Clinical Implications

Our findings are potentially applicable to clinical practice. First, they support inclusion of ED by the European guidelines for CVD prevention.37 Our analysis showed that the risk conferred by ED on events is of a magnitude similar to that of the risk conferred on events by established risk predictors such as hypertension and dyslipidemia commonly encountered in clinical practice. Thus, our results stress the importance of early diagnosis of ED and the meticulous CV investigation that is required in specific groups of ED patients.38 Furthermore, an important finding of our analysis is that ED is a significant predictor of all-cause mortality in addition to CV outcomes. Interestingly, ED is a stronger predictor of all-cause mortality in patients with known CVD compared with patients without CVD. Although pathophysiological explanations are not readily identifiable, this predictive ability for all-cause mortality could reflect the existence of common pathogenetic mechanisms such as aging, inflammation, and oxidative stress over a wide range of conditions. Moreover, rates of depression are increased in patients with CVD and ED. Depression worsens the outcome of comorbid physical conditions, and depressed patients show decreased adherence to treatment.

Improvement of ED by lifestyle interventions per se might be beneficial in terms of prognosis,14,39,40 and our results highlight the role of ED as a potential low-cost biomarker that would call for more aggressive CV risk factor modification.41 Of special interest is the effect of pharmacological treatment of ED, because it seems that this may also have a beneficial impact on risk. Indeed, Frantzen et al11 showed that 2 years after the introduction of sildenafil, the RR of the incidence of CVD among men with ED compared with healthy men significantly decreased from 1.7 to 1.1. Furthermore, Gazzaruso et al14 showed that type 5 phosphodiesterase inhibitors offer marginal protection against the development of major adverse cardiac events in diabetic patients with coronary artery disease and ED. Undoubtedly, more data are needed, and future follow-up studies should ideally collect information on ED treatment and investigate such an effect.

Further dissection of our principal finding provided interesting information. The RR was higher in patients with intermediate baseline CV risk compared with ED patients with high or low CV risk. This is particularly important because this intermediate-risk group is in need of further risk reclassification with a predictor such as ED. Furthermore, the RR was higher in younger ED patients despite the fact that the probability of ED increases with age. In fact, ED as a predictor may be particularly useful in these young patients in whom the Framingham Risk Score may underestimate risk by examining forward only 10 years.42 In addition to younger age, the RR was higher in the presence of smoking and dyslipidemic profile. Although risk factors become more frequent with advancing age, we interpret these findings as being complementary rather than contradictory. Overall, this combination of younger age and multiple risk factors may imply an aggressive pathophysiological background that is exacerbated by environmental factors such as smoking and unhealthy diet and stresses the necessity for their identification and treatment in patients with ED.

Our study provides interesting caveats for the heterogeneity of methods used across studies to diagnose ED. Although the medical and sexual history is essential and frequently the most revealing aspect of the ED assessment process, the diagnostic potential of assessment questionnaires has certain limits. The use of a validated questionnaire such as the International Index of Erectile Function questionnaire, which has been widely used for assessment of presence and severity ED and has been shown to correlate with Gensini score in patients undergoing coronary angiography,43 improves the diagnosis of ED and is the preferred strategy; however, this is not universally adopted. The importance of the use of a validated questionnaire is supported by our findings that in patients in whom ED was diagnosed with a questionnaire, the RR for total CV events was higher compared with that in patients in whom ED was diagnosed with a single question. Thus, it seems reasonable that ED when better substantiated and carefully investigated could provide more useful information about the future CV risk of the ED patient.

Of the studies included in the analysis, 5 used a validated questionnaire,14,15,1820 and of these, only 3 reported data according to the severity of ED.1820 Two more studies,12,23 despite using a single question to assess ED, categorized the answers according to ED severity on a 3-point scale. Taken as a whole, this evidence points toward a grading effect of the severity of ED in the predictive ability of this condition for CV events.12,20,23 However, more studies are needed to substantiate this notion.

Methodological Considerations

An important strength of our study is the exhaustive search strategy that likely enabled us to capture most, if not all, relevant studies. Moreover, our study for the first time uses data from the published studies for meta-regression analyses, thus enhancing the identification of the predictive role of ED in CVD. Furthermore, as a meta-analysis, the present study overcomes the potentially biased inclusion and weighing of results that may appear in reviews when interpreting the available evidence. Finally, although we would ideally want to have a larger number of high-quality studies, we dealt effectively with potential publication bias.

We intentionally included both prospective and retrospective longitudinal studies to allow a multifaceted approach to the estimation of risk. However, we took under consideration the more confounding and biased nature of retrospective studies. By applying a sensitivity analysis, we excluded these 3 retrospective studies without significant changes in our final results for total CV events.

In the majority of studies, ED patients were in most cases older, had higher blood pressure, and were more often diabetic or dyslipidemic. Thus, it is reasonable to assume that ED patients were a priori at higher baseline risk than non-ED patients. However, this inherent limitation of most prospective studies was dealt with in most of them by adjustment for the potential confounders between ED and non-ED patients. Furthermore, as it was shown in our sensitivity analysis for adjustment of risk factors, the RR in studies that adjusted for all conventional risk factors was lower but not substantially different from the overall combined risk. Thus, it seems that the increased risk for ED patients is unlikely to be a consequence of the higher baseline CV risk of ED patients.

Our results are in accordance with the results of 2 previous meta-analyses30,31 that estimated the RR for CVD at 47% and 48%, respectively. However, these meta-analyses did not include landmark studies such as the VITAL (Vitamins and Lifestyle),30,31 ONTARGET/TRANSCEND (Ongoing Telmisartan Alone and in Combination With Ramipril Global Endpoint Trial/Telmisartan Randomized Assessment Study in ACE Intolerant Subjects With Cardiovascular Disease),30,31 and ADVANCE (Action in Diabetes and Vascular Disease: Preterax and Diamicron Modified-Release Controlled Evaluation)30 studies. Furthermore, in 1 study31 results may have been influenced by a duplicate study inclusion (similar populations in the 2 studies)19,24 for the same end point, as well as by inclusion of a study23 with a population made up exclusively of ED patients.

Links Between ED and Events

Because CVD and ED overlap in risk factors, prevalence, and manifestation, they are thought to share both pathophysiological basis of pathogenesis and progression.4 The artery-size hypothesis6,43 proposes that the differential timing of the onset of signs (2–5 years) and symptoms of ED and CVD is caused by the difference between larger vessels and smaller ones (eg, pudendal or cavernous arteries) with regard to their ability to tolerate encroachment on the lumen. Moreover, we6,44 have shown that ED is associated with an incremental inflammatory and endothelial-prothrombotic activation on top of that found in patients with coronary artery disease. Accelerated arterial aging may be the background responsible for adverse outcome in ED, as pointed by the independent predictive ability of aortic stiffness (Vlachopoulos et al, unpublished observations) and pulse pressure (a crude index of aortic stiffness)45 for events in ED patients. Moreover, low testosterone may be implicated in the common pathogenetic pathways of ED and CVD; however, this warrants further substantiation.4648

Study Limitations

We used aggregate data as reported in published articles (or calculated from other data provided in these) rather than individual data, and thus we could not deal with potential methodological problems of the original studies. Only 1 included study19 provided robust estimates of the discriminatory and reclassification power of ED beyond classic risk factor or the Framingham Risk Score. The ability of ED to discriminate, calibrate, and reclassify risk can be assessed only on a patient-level data analysis, which seems to be the next desirable step. Finally, although CV mortality and all-cause mortality were uniformly defined, the definition of total CV events differed among the studies included in the analysis.

Conclusions

The presence of ED increases the risk for future CV events, MI, cerebrovascular events, and all-cause mortality, whereas it shows a trend to increase risk for CV mortality. The RR is higher in young subjects, in subjects with intermediate baseline CV risk, and when ED is diagnosed by a questionnaire rather than a single question. These findings support implementation of ED into clinical practice and stress the need to establish standardized methods to diagnose ED and to investigate the potential effect of treatment of ED on CV events and all-cause mortality.

Footnotes

The online-only Data Supplement is available at http://circoutcomes.ahajournals.org/lookup/suppl/doi:10.1161/CIRCOUTCOMES.112.966903/-/DC1.

Correspondence to Charalambos Vlachopoulos, MD, Cardiovascular Diseases and Sexual Health Unit, First Department of Cardiology, Athens Medical School, Profiti Elia 24, Kerassoundos 17, Athens 14575, Greece. E-mail

References

  • 1. Feldman HA, Goldstein I, Hatzichristou DG, Krane RJ, McKinlay JB. Impotence and its medical and psychosocial correlates: results of the Massachusetts male aging study.J Urol. 1994; 151:54–61.CrossrefMedlineGoogle Scholar
  • 2. Ayta IA, McKinlay JB, Krane RJ. The likely worldwide increase in erectile dysfunction between 1995 and 2025 and some possible policy consequences.BJU Int. 1999; 84:50–56.CrossrefMedlineGoogle Scholar
  • 3. Feldman HA, Johannes CB, Derby CA, Kleinman KP, Mohr BA, Araujo AB, McKinlay JB. Erectile dysfunction and coronary risk factors: prospective results from the Massachusetts Male Aging Study.Prev Med. 2000; 30:328–338.CrossrefMedlineGoogle Scholar
  • 4. El-Sakka AI, Morsy AM, Fagih BI, Nassar AH. Coronary artery risk factors in patients with erectile dysfunction.J Urol. 2004; 172:251–254.CrossrefMedlineGoogle Scholar
  • 5. Vlachopoulos C, Rokkas K, Ioakeimidis N, Aggeli C, Michaelides A, Roussakis G, Fassoulakis C, Askitis A, Stefanadis C. Prevalence of asymptomatic coronary artery disease in men with vasculogenic erectile dysfunction: a prospective angiographic study.Eur Urol. 2005; 48:996–1002; discussion 1002.CrossrefMedlineGoogle Scholar
  • 6. Vlachopoulos C, Ioakeimidis N, Terentes-Printzios D, Stefanadis C. The triad: erectile dysfunction–endothelial dysfunction–cardiovascular disease.Curr Pharm Des. 2008; 14:3700–3714.CrossrefMedlineGoogle Scholar
  • 7. Blumentals WA, Gomez-Caminero A, Joo S, Vannappagari V. Is erectile dysfunction predictive of peripheral vascular disease?Aging Male. 2003; 6:217–221.CrossrefMedlineGoogle Scholar
  • 8. Blumentals WA, Gomez-Caminero A, Joo S, Vannappagari V. Should erectile dysfunction be considered as a marker for acute myocardial infarction? Results from a retrospective cohort study.Int J Impot Res. 2004; 16:350–353.CrossrefMedlineGoogle Scholar
  • 9. Thompson IM, Tangen CM, Goodman PJ, Probstfield JL, Moinpour CM, Coltman CA. Erectile dysfunction and subsequent cardiovascular disease.JAMA. 2005; 294:2996–3002.CrossrefMedlineGoogle Scholar
  • 10. Cameron A, Sun P, Lage M. Comorbid conditions in men with ED before and after ED diagnosis: a retrospective database study.Int J Impot Res. 2006; 18:375–381.CrossrefMedlineGoogle Scholar
  • 11. Frantzen J, Speel TG, Kiemeney LA, Meuleman EJ. Cardiovascular risk among men seeking help for erectile dysfunction.Ann Epidemiol. 2006; 16:85–90.CrossrefMedlineGoogle Scholar
  • 12. Schouten BW, Bohnen AM, Bosch JL, Bernsen RM, Deckers JW, Dohle GR, Thomas S. Erectile dysfunction prospectively associated with cardiovascular disease in the Dutch general population: results from the Krimpen Study.Int J Impot Res. 2008; 20:92–99.CrossrefMedlineGoogle Scholar
  • 13. Ma RC, So WY, Yang X, Yu LW, Kong AP, Ko GT, Chow CC, Cockram CS, Chan JC, Tong PC. Erectile dysfunction predicts coronary heart disease in type 2 diabetes.J Am Coll Cardiol. 2008; 51:2045–2050.CrossrefMedlineGoogle Scholar
  • 14. Gazzaruso C, Solerte SB, Pujia A, Coppola A, Vezzoli M, Salvucci F, Valenti C, Giustina A, Garzaniti A. Erectile dysfunction as a predictor of cardiovascular events and death in diabetic patients with angiographically proven asymptomatic coronary artery disease: a potential protective role for statins and 5-phosphodiesterase inhibitors.J Am Coll Cardiol. 2008; 51:2040–2044.CrossrefMedlineGoogle Scholar
  • 15. Inman BA, Sauver JL, Jacobson DJ, McGree ME, Nehra A, Lieber MM, Roger VL, Jacobsen SJ. A population-based, longitudinal study of erectile dysfunction and future coronary artery disease.Mayo Clin Proc. 2009; 84:108–113.CrossrefMedlineGoogle Scholar
  • 16. Araujo AB, Travison TG, Ganz P, Chiu GR, Kupelian V, Rosen RC, Hall SA, McKinlay JB. Erectile dysfunction and mortality.J Sex Med. 2009; 6:2445–2454.CrossrefMedlineGoogle Scholar
  • 17. Hebert K, Lopez B, Macedo FY, Gomes CR, Urena J, Arcement LM. Peripheral vascular disease and erectile dysfunction as predictors of mortality in heart failure patients.J Sex Med. 2009; 6:1999–2007.CrossrefMedlineGoogle Scholar
  • 18. Ponholzer A, Gutjahr G, Temml C, Madersbacher S. Is erectile dysfunction a predictor of cardiovascular events or stroke? A prospective study using a validated questionnaire.Int J Impot Res. 2010; 22:25–29.CrossrefMedlineGoogle Scholar
  • 19. Araujo AB, Hall SA, Ganz P, Chiu GR, Rosen RC, Kupelian V, Travison TG, McKinlay JB. Does erectile dysfunction contribute to cardiovascular disease risk prediction beyond the Framingham risk score?J Am Coll Cardiol. 2010; 55:350–356.CrossrefMedlineGoogle Scholar
  • 20. Böhm M, Baumhäkel M, Teo K, Sleight P, Probstfield J, Gao P, Mann JF, Diaz R, Dagenais GR, Jennings GL, Liu L, Jansky P, Yusuf S; ONTARGET/TRANSCEND Erectile Dysfunction Substudy Investigators. Erectile dysfunction predicts cardiovascular events in high-risk patients receiving telmisartan, ramipril, or both: The ONgoing Telmisartan Alone and in combination with Ramipril Global End point Trial/Telmisartan Randomized AssessmeNt Study in ACE iNtolerant subjects with cardiovascular Disease (ONTARGET/TRANSCEND) Trials.Circulation. 2010; 121:1439–1446.LinkGoogle Scholar
  • 21. Batty GD, Li Q, Czernichow S, Neal B, Zoungas S, Huxley R, Patel A, de Galan BE, Woodward M, Hamet P, Harrap SB, Poulter N, Chalmers J; ADVANCE Collaborative Group. Erectile dysfunction and later cardiovascular disease in men with type 2 diabetes: prospective cohort study based on the ADVANCE (Action in Diabetes and Vascular Disease: Preterax and Diamicron Modified-Release Controlled Evaluation) trial.J Am Coll Cardiol. 2010; 56:1908–1913.CrossrefMedlineGoogle Scholar
  • 22. Han M, Trock BJ, Partin AW, Humphreys EB, Bivalacqua TJ, Guzzo TJ, Walsh PC. The impact of preoperative erectile dysfunction on survival after radical prostatectomy.BJU Int. 2010; 106:1612–1617.CrossrefMedlineGoogle Scholar
  • 23. Corona G, Monami M, Boddi V, Cameron-Smith M, Lotti F, de Vita G, Melani C, Balzi D, Sforza A, Forti G, Mannucci E, Maggi M. Male sexuality and cardiovascular risk. A cohort study in patients with erectile dysfunction.J Sex Med. 2010; 7:1918–1927.CrossrefMedlineGoogle Scholar
  • 24. Hall SA, Shackelton R, Rosen RC, Araujo AB. Sexual activity, erectile dysfunction, and incident cardiovascular events.Am J Cardiol. 2010; 105:192–197.CrossrefMedlineGoogle Scholar
  • 25. Chew KK, Finn J, Stuckey B, Gibson N, Sanfilippo F, Bremner A, Thompson P, Hobbs M, Jamrozik K. Erectile dysfunction as a predictor for subsequent atherosclerotic cardiovascular events: findings from a linked-data study.J Sex Med. 2010; 7(1 pt 1):192–202.CrossrefMedlineGoogle Scholar
  • 26. Chung SD, Chen YK, Lin HC, Lin HC. Increased risk of stroke among men with erectile dysfunction: a nationwide population-based study.J Sex Med. 2011; 8:240–246.CrossrefMedlineGoogle Scholar
  • 27. Hotaling JM, Walsh TJ, Macleod LC, Heckbert S, Pocobelli G, Wessells H, White E. Erectile dysfunction is not independently associated with cardiovascular death: data from the Vitamins and Lifestyle (VITAL) Study.J Sex Med. 2012. (in press)CrossrefMedlineGoogle Scholar
  • 28. Bittner N, Merrick GS, Galbreath RW, Butler WM, Lief JH, Allen ZA, Wallner KE. Erectile dysfunction is predictive of all-cause mortality in patients with prostate cancer treated with permanent interstitial brachytherapy.BJU Int. 2012; 109:220–225.CrossrefMedlineGoogle Scholar
  • 29. Aktoz M, Aktoz T, Tatli E, Kaplan M, Turan FN, Barutçu A, Atakan IH, Demir M, Altun A. Asymmetrical dimethylarginine and severity of erectile dysfunction and their impact on cardiovascular events in patients with acute coronary syndrome.Arch Med Sci. 2010; 6:168–175.CrossrefMedlineGoogle Scholar
  • 30. Guo W, Liao C, Zou Y, Li F, Li T, Zhou Q, Cao Y, Mao X. Erectile dysfunction and risk of clinical cardiovascular events: a meta-analysis of seven cohort studies.J Sex Med. 2010; 7:2805–2816.CrossrefMedlineGoogle Scholar
  • 31. Dong JY, Zhang YH, Qin LQ. Erectile dysfunction and risk of cardiovascular disease: meta-analysis of prospective cohort studies.J Am Coll Cardiol. 2011; 58:1378–1385.CrossrefMedlineGoogle Scholar
  • 32. Stroup DF, Berlin JA, Morton SC, Olkin I, Williamson GD, Rennie D, Moher D, Becker BJ, Sipe TA, Thacker SB. Meta-analysis of observational studies in epidemiology: a proposal for reporting: Meta-analysis Of Observational Studies in Epidemiology (MOOSE) group.JAMA. 2000; 283:2008–2012.CrossrefMedlineGoogle Scholar
  • 33. Higgins JP, Thompson SG, Deeks JJ, Altman DG. Measuring inconsistency in meta-analyses.BMJ. 2003; 327:557–560.CrossrefMedlineGoogle Scholar
  • 34. Altman DG, Bland JM. Interaction revisited: the difference between two estimates.BMJ. 2003; 326:219.CrossrefMedlineGoogle Scholar
  • 35. Duval S, Tweedie R. Trim and fill: A simple funnel-plot-based method of testing and adjusting for publication bias in meta-analysis.Biometrics. 2000; 56:455–463.CrossrefMedlineGoogle Scholar
  • 36. Rosenthal R. The ‘file drawer problem’ and tolerance for null results.Psychol Bull.1979; 86:638–641.CrossrefGoogle Scholar
  • 37. Perk J, De Backer G, Gohlke H, Graham I, Reiner Z, Verschuren M, Albus C, Benlian P, Boysen G, Cifkova R, Deaton C, Ebrahim S, Fisher M, Germano G, Hobbs R, Hoes A, Karadeniz S, Mezzani A, Prescott E, Ryden L, Scherer M, Syvänne M, Scholte op Reimer WJ, Vrints C, Wood D, Zamorano JL, Zannad F. European guidelines on cardiovascular disease prevention in clinical practice (version 2012): the fifth joint task force of the European Society of Cardiology and other societies on cardiovascular disease prevention in clinical practice (constituted by representatives of nine societies and by invited experts): developed with the special contribution of the European Association for Cardiovascular Prevention & Rehabilitation (EACPR).Eur Heart J. 2012; 33:1635–1701.CrossrefMedlineGoogle Scholar
  • 38. Nehra A, Jackson G, Miner M, Billups KL, Burnett AL, Buvat J, Carson CC, Cunningham GR, Ganz P, Goldstein I, Guay AT, Hackett G, Kloner RA, Kostis J, Montorsi P, Ramsey M, Rosen R, Sadovsky R, Seftel AD, Shabsigh R, Vlachopoulos C, Wu FC. The Princeton III Consensus recommendations for the management of erectile dysfunction and cardiovascular disease.Mayo Clin Proc. 2012; 87:766–778.CrossrefMedlineGoogle Scholar
  • 39. Gupta BP, Murad MH, Clifton MM, Prokop L, Nehra A, Kopecky SL. The effect of lifestyle modification and cardiovascular risk factor reduction on erectile dysfunction: a systematic review and meta-analysis.Arch Intern Med. 2011; 171:1797–1803.CrossrefMedlineGoogle Scholar
  • 40. Vlachopoulos C, Terentes-Printzios D, Ioakeimidis N, Rokkas K, Stefanadis C. PDE5 inhibitors in non-urological conditions.Curr Pharm Des. 2009; 15:3521–3539.CrossrefMedlineGoogle Scholar
  • 41. US Preventive Services Task Force. Using nontraditional risk factors in coronary heart disease risk assessment: US Preventive Services Task Force recommendation statement.Ann Intern Med. 2009; 151:474–82.CrossrefMedlineGoogle Scholar
  • 42. Marma AK, Berry JD, Ning H, Persell SD, Lloyd-Jones DM. Distribution of 10-year and lifetime predicted risks for cardiovascular disease in US adults: findings from the National Health and Nutrition Examination Survey 2003 to 2006.Circ Cardiovasc Qual Outcomes. 2010; 3:8–14.LinkGoogle Scholar
  • 43. Montorsi P, Ravagnani PM, Galli S, Rotatori F, Veglia F, Briganti A, Salonia A, Dehò F, Rigatti P, Montorsi F, Fiorentini C. Association between erectile dysfunction and coronary artery disease: role of coronary clinical presentation and extent of coronary vessels involvement: the COBRA trial.Eur Heart J. 2006; 27:2632–2639.CrossrefMedlineGoogle Scholar
  • 44. Vlachopoulos C, Aznaouridis K, Ioakeimidis N, Rokkas K, Vasiliadou C, Alexopoulos N, Stefanadi E, Askitis A, Stefanadis C. Unfavourable endothelial and inflammatory state in erectile dysfunction patients with or without coronary artery disease.Eur Heart J. 2006; 27:2640–2648.CrossrefMedlineGoogle Scholar
  • 45. Corona G, Monami M, Boddi V, Rastrelli G, Melani C, Balzi D, Sforza A, Forti G, Mannucci E, Maggi M. Pulse pressure independently predicts major cardiovascular events in younger but not in older subjects with erectile dysfunction.J Sex Med. 2011; 8:247–254.CrossrefMedlineGoogle Scholar
  • 46. Basaria S, Dobs AS. Testosterone making an entry into the cardiometabolic world.Circulation. 2007; 116:2658–2661.LinkGoogle Scholar
  • 47. Corona G, Monami M, Boddi V, Cameron-Smith M, Fisher AD, de Vita G, Melani C, Balzi D, Sforza A, Forti G, Mannucci E, Maggi M. Low testosterone is associated with an increased risk of MACE lethality in subjects with erectile dysfunction.J Sex Med. 2010; 7(pt 1):1557–1564.CrossrefMedlineGoogle Scholar
  • 48. Jackson G, Boon N, Eardley I, Kirby M, Dean J, Hackett G, Montorsi P, Montorsi F, Vlachopoulos C, Kloner R, Sharlip I, Miner M. Erectile dysfunction and coronary artery disease prediction: evidence-based guidance and consensus.Int J Clin Pract. 2010; 64:848–857.CrossrefMedlineGoogle Scholar