Erectile Dysfunction as a Predictor for Subsequent Atherosclerotic Cardiovascular Events: Findings from a Linked-Data Study

Authors

  • Kew-Kim Chew MBBS, FRCPEdin, FRCPGlasg,

    Corresponding author
    1. Keogh Institute for Medical Research, Nedlands, Perth, WA, Australia;
    2. University of Western Australia, School of Population Health, Nedlands, Perth, WA, Australia;
      Kew-Kim Chew, MBBS, FRCPEdin, FRCPGlasg, Queen Elizabeth II Medical Centre—Keogh Institute for Medical Research, Nedlands, Perth, WA 6009, Australia. Tel: +618 93462008; Fax: +618 93898384; E-mail: kewkimchew@hotmail.com
    Search for more papers by this author
  • Judith Finn PhD, MEdStudies, GradDipPH, BSc, DipAppSc(Nsg), RN, RM, ICCert,

    1. University of Western Australia, School of Population Health, Nedlands, Perth, WA, Australia;
    2. Sir Charles Gairdner Hospital, Centre for Nursing Research, Nedlands, Perth, WA, Australia;
    Search for more papers by this author
  • Bronwyn Stuckey MBBS, FRACP,

    1. Keogh Institute for Medical Research, Nedlands, Perth, WA, Australia;
    2. Sir Charles Gairdner Hospital, Centre for Nursing Research, Nedlands, Perth, WA, Australia;
    3. University of Western Australia, School of Medicine and Pharmacology, Nedlands, Perth, WA, Australia;
    Search for more papers by this author
  • Nicholas Gibson RN, BAppSci, PGradDipHlthAdmin, PhD, FRCNA,

    1. University of Western Australia, School of Primary, Aboriginal and Rural Health, Nedlands, Perth, WA, Australia;
    Search for more papers by this author
  • Frank Sanfilippo BSc, BPharm, PGradDipPharm, FPS, PhD,

    1. University of Western Australia, School of Population Health, Nedlands, Perth, WA, Australia;
    Search for more papers by this author
  • Alexandra Bremner BSc(Hons), GradDipAppStats, PhD,

    1. University of Western Australia, School of Population Health, Nedlands, Perth, WA, Australia;
    Search for more papers by this author
  • Peter Thompson AM, MD, MBBS, FRACP, FACP, FACC,

    1. University of Western Australia, School of Population Health, Nedlands, Perth, WA, Australia;
    2. Sir Charles Gairdner Hospital, Centre for Nursing Research, Nedlands, Perth, WA, Australia;
    3. University of Western Australia, School of Medicine and Pharmacology, Nedlands, Perth, WA, Australia;
    Search for more papers by this author
  • Michael Hobbs MBBS, DPhil, FRACP, FAFPHM,

    1. University of Western Australia, School of Population Health, Nedlands, Perth, WA, Australia;
    Search for more papers by this author
  • Konrad Jamrozik MBBS, D Phil

    1. University of Adelaide, School of Population Health and Clinical Practice, Adelaide, SA, Australia
    Search for more papers by this author

  • Funding: The linked data for the study was acquired partly with the assistance of an unrestricted research grant from the Asia Pacific Society for Sexual Medicine awarded to the leading investigator of the study.

Kew-Kim Chew, MBBS, FRCPEdin, FRCPGlasg, Queen Elizabeth II Medical Centre—Keogh Institute for Medical Research, Nedlands, Perth, WA 6009, Australia. Tel: +618 93462008; Fax: +618 93898384; E-mail: kewkimchew@hotmail.com

ABSTRACT

Introduction.  In spite of the mounting interest in the nexus between erectile dysfunction (ED) and cardiovascular (CV) diseases, there is little published information on the role of ED as a predictor for subsequent CV events.

Aim.  This study aimed to investigate the role of ED as a predictor for atherosclerotic CV events subsequent to the manifestation of ED.

Method.  The investigation involved the retrospective study of data on a cohort of men with ED linked to hospital morbidity data and death registrations. By using the linked data, the incidence rates of atherosclerotic CV events subsequent to the manifestation of ED were estimated in men with ED and no atherosclerotic CV disease reported prior to the manifestation of ED. The risk of subsequent atherosclerotic CV events in men with ED was assessed by comparing these incidence rates with those in the general male population.

Main Outcome Measure.  Standardized incidence rate ratio (SIRR), comparing the incidence of atherosclerotic CV events subsequent to the manifestation of ED in a cohort of 1,660 men with ED to the incidence in the general male population.

Results.  On the basis of hospital admissions and death registrations, men with ED had a statistically significantly higher incidence of atherosclerotic CV events (SIRR 2.2; 95% confidence interval 1.9, 2.4). There were significantly increased incidence rate ratios in all age groups younger than 70 years, with a statistically highly significant downward trend with increase of age (P < 0.0001) across these age groups. Younger age at first manifestation of ED, cigarette smoking, presence of comorbidities and socioeconomic disadvantage were all associated with higher hazard ratios for subsequent atherosclerotic CV events.

Conclusions.  The findings show that ED is not only significantly associated with but is also strongly predictive of subsequent atherosclerotic CV events. This is even more striking when ED presents at a younger age. Chew K-K, Finn J, Stuckey B, Gibson N, Sanfilippo F, Bremner A, Thompson P, Hobbs M, and Jamrozik K. Erectile dysfunction as a predictor for subsequent atherosclerotic cardiovascular vents: Findings from a linked-data study. J Sex Med 2010;7:192–202.

Introduction

Erectile dysfunction (ED), the consistent or recurrent inability to achieve and/or maintain a penile erection sufficient for satisfactory sexual performance [1,2], is a prevalent and mostly vasculogenic condition [3,4]. It is age related, and, as age advances, erectile function is adversely affected further by various age-related comorbidities, particularly cardiovascular (CV) risk factors and diseases, either independently or as clusters of interrelated conditions.

Cross-sectional studies have shown that CV risk factors and diseases are significantly more prevalent in men with ED, especially those with severe ED [5,6]. Men with CV risk factors, such as central obesity, diabetes mellitus, hypertension and hyper- or dyslipidemia, clinical components of the metabolic syndrome, have a significantly higher risk of ED [7–19]. Indeed, the greater the number of these CV risk factors, the higher the risk [12]. Such findings suggest that ED could be a manifestation of generalized atherosclerotic CV disease [20–23].

In spite of the interest in the nexus between ED and CV diseases, there is little published information from longitudinal studies concerning the role of ED as a predictor for subsequent atherosclerotic CV events. To investigate this role, we estimated the incidence of first-ever atherosclerotic CV events subsequent to the manifestation of ED in a retrospective linked-data study in Western Australia (WA).

Method

Study Design

The study comprised the acquisition of a study cohort of men with ED in WA who did not have a history of atherosclerotic CV disease prior to the manifestation of ED, the estimation of the incidence rates of atherosclerotic CV events subsequent to the manifestation of ED in the study cohort, and the comparison of these incidence rates for the same calendar period with those in a reference population representative of the WA general male population.

Study Cohort

The study cohort was derived from the Western Australia Erectile Dysfunction Research Dataset (WAEDRD), a linked data set comprising the Keogh Institute for Medical Research Erectile Dysfunction Dataset (KIMREDD) and two population-based administrative data sets: the Western Australian Hospital Morbidity Data (WAHMD) and the Western Australian Death Data (WADD). As illustrated in Figure 1, the acquisition of the WAEDRD was achieved through the Western Australian Data Linkage System (WADLS) [24,25].

Figure 1.

The acquisition of the Western Australia Erectile Dysfunction Research Dataset (WAEDRD) through the Western Australia Data Linkage System (WADLS). CV = cardiovascular; ED = erectile dysfunction; WA = Western Australia.

The KIMREDD is a data set of men who attended the Keogh Institute for Medical Research (KIMR) for ED over a 10-year period, from 1 January 1995 to 31 December 2004. It contains the sociodemographic and clinical details of the attendees, including each man's surname, given name, date of birth, residential address, postcode, available Medical Record Number (a personal identification number allotted to a WA resident attending a public hospital in WA), and self-reports and/or the referring doctors' summaries of existing atherosclerotic CV disease. The WAHMD contains data related to all public and private hospital admissions in WA since 1980, with up to 21 International Classification of Diseases (ICD-9-CM or ICD-10) [26] codes for discharge diagnoses and 11 codes for procedures, and the WADD holds details of the date, cause(s), and place of all deaths registered in WA.

The study cohort of 1,660 men was derived by excluding from the WAEDRD all men with any atherosclerotic CV disease over a 10-year look-back period from the dates of their first attendances at the KIMR.

Data Linkage

We were able to achieve data linkage by probabilistic matching [27] because the sociodemographic details used as identification codes for the KIMREDD are also among the principal matching fields in the research infrastructure of the WADLS. These provided the linkage keys to enable links to be identified, stored, and retrieved [25,27,28].

There is, thus, for each of the 1,660 men with ED in the study cohort one single chain of records representing the continuum of the men's health care from 1980 to 2005, including details of any existing atherosclerotic CV disease. In the same single chain of records, details of the men's hospital admissions for any CV event, procedure or diagnosis indicative of atherosclerotic CV disease, prior and/or subsequent to the manifestation of ED, or of his death, if applicable, were also recorded.

In our study, manifestation of ED refers to the time at which the affected man first became aware of ED as a problem regardless of when he attended the KIMR or sought treatment for ED.

Atherosclerotic CV Disease

For our study, atherosclerotic CV disease was identified by the use of the ICD-9-CM or ICD-10 [26] diagnostic codes for ischemic heart disease (410–414), cardiac arrest (427.5), cerebrovascular disease (431–434), peripheral vascular disease (440–443), or CV procedures codes for percutaneous transluminal coronary angioplasty (PTCA) (36.01–36.05), PTCA with stenting (36.06 and 36.07), and coronary artery bypass grafting (36.10–36.19).

The same codes were used for identifying atherosclerotic CV disease in the reference population selected for comparison of incidence rates with the study cohort.

Reference Population

To provide a reference population representative of the general male population in WA for external comparison with our study cohort, we selected a linked, population-based administrative health data set that had been extracted from the WADLS. This data set contained for the period of 1980–2004 a person-linked file of records for all males who had been admitted to hospitals in WA for atherosclerotic CV disease or had died from a CV cause [29,30].

First-ever hospital admissions for atherosclerotic CV disease as well as registrations of deaths from CV causes in the period of 1995–2004 were identified in the reference population by using a fixed look-back period of 15 years. Incidence rates were calculated by using, as numerator, the incident cases identified and, as denominator, the estimated average male population in WA over the period of 1995–2004, based on the Australian Bureau of Statistics (ABS) data for calendar years from 1995 to 2004 [31]. These incidence rates constituted the reference population rates to provide comparison with the incidence rates within our study cohort of men with ED.

Internal Comparisons within the Study Cohort

Internal comparisons within the study cohort were conducted to examine the factors associated with the hazard or risk of atherosclerotic CV events subsequent to the manifestation of ED.

We included the Charlson Comorbidity Index (CCI) and Socioeconomic Indexes for Areas (SEIFA) in the calculation of the hazard ratios (HRs) and 95% confidence intervals (CIs) of atherosclerotic CV events subsequent to manifestation of ED based on hospital admission and death registration data.

CCI

From the diagnostic and procedural codes recorded for each patient in the WAHMD, we computed a comorbidity index for the individual using the model developed by Charlson and associates [32] and adapted to the Dartmouth–Manitoba algorithm [33–35], which has been used extensively for risk adjustment in the analysis of vertically linked longitudinal data [34]. The weights assigned to 17 diseases, including diabetes mellitus, with appropriately adjusted relative risks, were added to form a single ordinal score, the CCI [32], where a score of 0 reflects no mention of a prior hospital admission for 1 of the 17 conditions within 5 years of the first presentation to the KIMR for ED, and the higher the score, the greater the comorbid burden.

SEIFA

The SEIFA were derived from population census data by the ABS as a measure of socioeconomic status [36]. We used the SEIFA score to provide an estimate of socioeconomic disadvantage for each man in the study cohort, based on the postcode of his residential address when he first attended the Keogh Institute for ED.

SEIFA 1996 scores were used for records up to 1996, and SEIFA 2001 scores were used for records from 1997 onward, corresponding to the 1996 and 2001 census years, respectively.

Data Analyses

We calculated the age-specific rates (ASRs) and the incidence rates of atherosclerotic CV events for each age group in the study cohort of 1,660 men with ED. For the different periods of follow-up in the study cohort, ASRs were corrected with the total person-time of follow-up for each man as the denominator.

Rates in our study were calculated over the whole period of 1995–2004 rather than for individual calendar years. For the number of death registrations in the study cohort, in addition to men who had died from CV causes and had records of hospital admissions for CV disease, we also identified and included men in the study cohort who had died from CV causes but had no records of any hospital admission for CV diseases.

The standardization of rates to the average WA male population in 1995–2004 was achieved by the indirect method. Standardized incidence rates for the various age groups in the study cohort were compared with those in the reference population by the corresponding incidence rate ratios (IRRs). The standardized incidence rate ratio (SIRR) was calculated by dividing the observed number of atherosclerotic CV events in the entire study cohort by the expected number of events.

Time intervals from manifestations of ED in the study cohort to the first hospital admissions for atherosclerotic CV events or to death from CV causes were calculated and tabulated as univariate frequencies, with corresponding summary statistics.

Multivariate Cox proportional hazard regression was used to examine the association of age at first manifestation of ED, CCI, SEIFA, and cigarette smoking with atherosclerotic CV events subsequent to the manifestation of ED. Poisson regression was used to assess the trend of IRRs across the different age groups.

For the calculation of HRs and 95% CIs, CCIs were categorized into CCI Category 0 (CCI = 0), CCI Category 1 (CCI = 1–2), and CCI Category 2 (CCI ≥ 3). The SEIFA were categorized in quartiles (SEIFA 1 to SEIFA 4), with the lowest quartile (SEIFA 1) indicating the greatest socioeconomic disadvantage.

Analyses were performed with the Statistical Package of Social Science (SPSS) for Windows 15.0 (SPSS Inc., Chicago, IL, USA).

Ethics Approval

The Human Research Ethics Committee of Sir Charles Gairdner Hospital and the WA Confidentiality of Health Information Committee granted approval for the study.

Results

Of the 2,318 men with ED in the data set linking the KIMREDD to the WAHMD and the WADD, there were 1,660 men who, prior to manifestation of ED, did not have atherosclerotic CV disease documented in either their Keogh Institute records or the WAHMD. These 1,660 men constituted the study cohort of men with ED and no prior CV diseases.

By the censor date of December 31, 2005, hospital admissions for atherosclerotic CV events on at least one occasion subsequent to their manifestation of ED were recorded in the WAHMD for 293 (17.7%) of the 1,660 men. The ages of the 293 men when first seen at the KIMR for ED ranged from 20 to 89 years (mean 54 years, median 55 years, standard deviation 11 years). The majority (75.1%) were married or had partners, 8.9% had never been married, 4.8% were separated or divorced, and 3.4% were widowers.

Of these 293 men who had hospital admissions for atherosclerotic CV events, 27 subsequently died from CV causes, and registrations of their death were recorded in the WADD. In addition, 15 registrations of death from CV causes involving men in the study cohort with no records of hospital admissions for atherosclerotic CV events subsequent to manifestation of ED were identified from the WADD. There were, therefore, a total of 308 (18.6%) men with atherosclerotic CV events subsequent to the manifestation of ED within the study cohort of 1,660 men.

Table 1 shows the age distribution of the average WA male population in 1995–2004 and of the study cohort. There were 52,291 first-ever hospital admissions for atherosclerotic CV disease or registrations of death from CV causes in the reference population. The ASRs in both the reference population and the study cohort, as well as the IRRs in the study cohort and their respective 95% Cis, are also shown in Table 1.

Table 1.  Atherosclerotic cardiovascular (CV) events subsequent to manifestation of erectile dysfunction (ED)
Age group (years)Western Australian (WA) male population 1995–2004*Study cohortHospital admissions for atherosclerotic CV events subsequent to the manifestation of ED and registrations of death from atherosclerotic CV causes
N%N%Reference populationStudy cohort
NASRNASR§IRR95% CI**
ObservedExpected
  • *

    Average WA male population 1995–2004 = the estimated average of the WA male populations for calendar years from 1995 to 2004 [27].

  • Reference cohort = Cohort of men in WA who had been admitted to hospital for atherosclerotic CV disease for the first time in 1995–2004 (excluding cases that occurred during a fixed 15-year look-back period from 1995 to 2004) or who had died from atherosclerotic CV disease.

  • ASR = Age-specific rate (per 1,000 person-years) for the average WA male population in 1995–2004.

  • §

    ASR = Age-specific rate (per 1,000 person-years) for the study cohort of men with ED based on their total follow-up time.

  • IRR = Incidence rate ratio (ratio of ASR§ to ASR).

  • **

    95% CI = 95% confidence interval for IRR.

<20273,80129.3392.31390.0500.030.00.00.0–90.4
20–29139,46914.91529.23270.2350.661.87.62.5–17.8
30–39145,52315.620612.41,4881.02202.707.67.44.5–11.5
40–49140,35515.046728.15,2883.777220.7313.13.52.7–4.4
50–59107,39811.546628.110,5169.7910647.1722.02.21.8–2.7
60–6967,6407.226816.112,91119.098351.8330.61.61.3–2.0
70–7942,8684.6583.513,35631.161817.7931.51.00.6–1.6
≥8016,6801.840.28,26649.5641.97100.42.00.6–5.2
Total933,7341001,66010052,291 308142.87   
  Standardized incidence rate ratio (SIRR) (Standardized to WA male population 1995–2004)2.21.9–2.4

Overall, the incidence of atherosclerotic CV events in the study cohort of men with ED was double the incidence in the reference population representative of the general male population in WA (SIRR = 2.2, 95% CI 1.9, 2.4). There were significantly increased IRRs in all age groups younger than 70 years, with a statistically highly significant downward trend with increase of age (P < 0.0001) across these age groups. The sevenfold increase in the age groups younger than 40 years was particularly striking.

Table 2 shows that, up to the censor date, there were 132 deaths including 42 (31.8%) deaths from CV causes among the 1,660 men with ED in the study cohort. Of the 42 registrations of death from CV causes, 15 did not have any record of hospital admissions for atherosclerotic CV disease.

Table 2.  Death from cardiovascular (CV) causes among men in study cohort
Cause of deathNumber of deaths in study cohort (N = 1,660)% of all deaths
Without records of hospital admissionWith records of hospital admissionTotalCumulative total%Cumulative %
Coronary artery disease81927 20.5 
Cardiac arrest426334.525.0
Peripheral arterial disease044373.028.0
Stroke325423.831.8
CV causes1527424231.8 
Other causes61299013268.2100.0
Total7656132 100.0 

The HRs for atherosclerotic CV event subsequent to the manifestation of ED are shown in Table 3. Increased age at first manifestation of ED, greater comorbid burden indicated by higher CCI score, and current cigarette smoking were significantly associated with higher HRs, suggesting increased hazard or risk. The risk in men with the greatest socioeconomic disadvantage (SEIFA category 1) was almost 50% higher (HR = 1.47; 95% CI 1.07, 2.01) than in men with the lowest socioeconomic disadvantage (SEIFA category 4).

Table 3.  Atherosclerotic cardiovascular (CV) events subsequent to manifestation of erectile dysfunction (ED)
Confounding variableHR*95% CI
  • *

    HR = adjusted hazard ratio from a multivariate Cox proportional regression model containing all the variables listed in the table.

  • CI = confidence interval; CCI Category 0 = CCI 0; Category 1 = CCI 1–2; Category 2 = CCI 3 and above.

Age at first manifestation of ED1.071.06–1.08
Charlson Comorbidity Index (CCI)  
 CCI Category 01.00 
 CCI Category 11.651.10–2.50
 CCI Category 23.221.51–6.89
Socioeconomic Index for Area (SEIFA)  
 SEIFA 11.471.07–2.01
 SEIFA 21.260.90–1.78
 SEIFA 31.290.92–1.82
 SEIFA 41.00 
Cigarette smoking  
 Never smokers1.00 
 Former smokers1.160.87–1.54
 Current smokers1.641.17–2.29

Table 4 shows that 76.0% of subsequent atherosclerotic CV events had occurred within 15 years of the manifestation of ED, and 4.2% had occurred within 2 years. The median time interval was 11.9 years (mean 11.2 years, standard deviation 6.9 years).

Table 4.  Time interval between manifestation of erectile dysfunction (ED) and first atherosclerotic cardiovascular (CV) events subsequent to manifestation of ED
Time interval from manifestation of ED (years)Atherosclerotic CV events subsequent to manifestation of ED
NCumulative NCumulative %
≤213134.2
2.1–5253812.3
5.1–107711537.3
10.1–1511923476.0
15.1–204527990.6
>2029308100.0
Median 11.9Mean 11.2Standard deviation 6.9

Discussion

Based on hospital admissions and death registrations, our findings show that the incidence of atherosclerotic CV events subsequent to ED is twice that in the general male population (SIRR 2.2; 95% CI 1.9, 2.4). ED is thus significantly associated with subsequent atherosclerotic CV events at the severe end of the spectrum requiring management in hospital and eventuating in death.

These findings are broadly consistent with those from a prospective population-based longitudinal study of ED and future coronary artery disease [37] in confirming a new dimension to the intimate nexus between ED and CV diseases beyond the notion that these are just two common interrelated conditions. In addition to supporting the concept that ED and atherosclerotic CV diseases are manifestations of generalized vascular endothelial disease in different vascular beds, they foster ED as a possible marker and predictor of subsequent atherosclerotic CV diseases, a proposition promulgated in recent studies and reviews [37–41]. In contrast to the significantly higher rate of coronary artery disease for men with moderate to severe ED from model-based predictions [42], our finding of significantly increased risk was based on events that had occurred as hospital admissions or death registrations.

IRRs for subsequent atherosclerotic CV events were significantly higher in all age groups younger than 70 years in the study cohort and particularly so in age groups younger than 40 years. The downward trend with increasing age across these age groups was statistically significant (P < 0.0001). The findings are suggestive of a greater relative risk of atherosclerotic CV events in younger men with ED and decreased predictive value of ED for atherosclerotic CV event in the older persons. As the study cohort was derived from the database of the KIMR, which is the main but not the sole provider of diagnostic and therapeutic services for all the men with ED in WA, it was not possible to exclude the influence of selection bias or to estimate the level of such bias if it were present. However, findings with strikingly similar implications were also reported from a prospective population-based longitudinal study [37].

We have considered, as a limitation to our study, the possibility that some men with ED in our study cohort might have had subsequent atherosclerotic CV events or diseases for which treatment in hospital had not been needed. However, even if such information was available to us, its inclusion would only inflate the SIRR. At the same time, the possibility that the KIMR records might show CV morbidity in some men in the study cohort who subsequently developed atherosclerotic CV disease if the look-back period had been longer than 10 years. However, such men would be relatively young then and would be unlikely candidates for atherosclerotic CV diseases and, in the absence of any known need for medical attention over a 10-year period, any atherosclerotic CV disease, if reported, might have been either presumptive or mild, and any bias would be small.

To provide a reference or control population as a comparator for statistical analyses, we had also considered carefully the selection of a representative cohort of men without ED from general population samples. Investigators of the American Prostate Cancer Prevention Trial were able to use a sizable subset of men without ED for internal comparison [41], but the option was not available in our linked data set comprising exclusively men with ED. Just as it would not be possible to exclude unreported or undiagnosed ED from any randomly selected sample of men, whether directly from the general population or from another population-based data set, there might also be men with ED in the study cohort who had been treated for ED at treatment centers other than the KIMR. Nonetheless, any bias from these possible inclusions, if anything, would have either inflated the incidence rates of atherosclerotic CV diseases among the controls or moderated those of atherosclerotic CV events among men with ED, thus abating and not undermining the comparisons.

The conundrum and propriety of control group selection have been discussed in a comprehensive review on study designs [43]. In our selection of a linked population-based administrative health data set extracted from the WADLS as the reference population, we were aware of the possibility that it might include men with unreported or undiagnosed ED. Such a population sample, however, represents in reality the population at risk, the actual goal of control group selection. As the reference population we selected was sizable and there was considerable disparity in size between our study cohort and the reference population, we have estimated the incidence rates of atherosclerotic CV events and standardized IRRs for our study cohort by indirect standardization.

Endothelial dysfunction has been established as the common denominator in the pathophysiology of both ED and CV disease [21,44–47], and changes in the penile cavernosal vascular beds have often become clinically evident when a functional impairment of the vascular endothelium in CV disease is still in its early asymptomatic stages [48–50]. Not surprisingly, age, cigarette smoking, and diabetes mellitus were found to be significantly associated with subsequent atherosclerotic CV events. The adverse relationships between these risk factors and ED and between ED and CV disease have been consistently reported from various studies including the WAMHS [3–19,51].

ED was found to be associated with a significant HR of 1.45 (95% CI 1.25–1.69) for subsequent CV disease in the American Prostate Cancer Prevention Trial [41]. Among 25,650 men with ED, a 75% increased risk was reported for peripheral vascular diseases [52], which in a recent report was shown to predict mortality from heart failure in patients with ED [53]. In a study involving coronary artery calcium (CAC) levels by multidetector computed tomography in men with ED, CAC scores were found to increase with the prevalence of ED and its severity [54]. A significant proportion of men with ED were shown to have abnormal stress echocardiographic changes [55]. ED has also been shown to predict subsequent metabolic syndrome and subsequent CV diseases in men with normal body mass index at baseline [56]. Circulating CD34+KDR+ endothelial progenitor cells were found to be reduced in overweight men with ED, and the reduction was correlated with the severity of ED [57]. A significantly greater number of major CV events were shown to occur in diabetic men with ED than in those without ED during a follow-up of up to 82 months [58]. In another study involving a 4-year follow-up of diabetic men without baseline clinical CV disease, the rate of new CV events in those with ED, based on hospital admission codes or mortality, was more than double of that in those without ED [59]. In our study, age at diagnosis of ED, cigarette smoking, the presence of comorbidities, and socioeconomic disadvantage were significantly associated with higher HRs for subsequent CV disease.

Compared with 11% reported from the American Prostate Cancer Prevention Trial [41], 12.3% of men in our study had their first-ever atherosclerotic CV events within 5 years of the manifestation of ED. The striking similarity of findings from these two studies, despite their different designs, suggests that ED probably precedes a subsequent atherosclerotic CV event by a much longer time interval than that reported from a study in which most men with ED who had angiographic evidence of coronary artery disease reported having had sexual symptoms for 2–3 years prior to anginal symptoms [60,61]. The mystique of the contrasting time-to-event intervals is compounded by the finding that almost 30% of the patients with angiographically documented coronary artery disease in the same study did not have ED [60,61]. In a study involving a retrospective assessment of erectile function in patients undergoing angiographic investigation for various diagnostic and therapeutic reasons, it would be difficult to either exclude the biases of recall and suggestion of association or assess the levels of such biases, if there were any.

Of interest, too, are our findings that one-third of deaths from CV causes and 30% of deaths from coronary artery disease were of men without records of hospital admission for CV disease. Given that almost 50% of sudden cardiac deaths are known to occur in patients with silent coronary artery disease [62], these CV events subsequent to ED might have occurred without warning from existing undiagnosed atherosclerotic CV disease.

The epidemiological information from our analyses is thus consistent with the observations from various biophysiological and clinical studies [19–21,50–64]. Our findings, therefore, provide further evidence that ED is a marker of generalized vascular dysfunction and a possible predictor of subsequent CV morbidity. In keeping with the Second Princeton Consensus [65], these data provide compelling evidence that men presenting with ED should undergo a thorough assessment of their CV risk and appropriate management of modifiable factors contributing to that risk. The time-to-event interval of several years between the manifestation of ED and a subsequent atherosclerotic CV event makes a beneficial outcome achievable if CV risk prevention or minimization programs are initiated early.

Conclusions

Our longitudinal analyses of a cohort of men with ED show significantly higher HRs and IRRs of atherosclerotic CV events subsequent to the manifestation of ED. The findings are consistent with ED as a marker of generalized vascular dysfunction and a predictor of subsequent CV morbidity.

These findings provide raison d'être for a paradigm shift in the perception and management of ED and offer opportunities for CV risk prevention and minimization when CV disease may still be silent or asymptomatic.

Acknowledgments

We gratefully acknowledge the assistance of Messrs. Helena Ching, Joy Foyle, Emma Fuller, and Carolyn Bond in the acquisition of our linked data set.

Conflict of Interest: None.

Statement of Authorship

Category 1

  • (a)Conception and DesignKew-Kim Chew; Judith Finn; Bronwyn Stuckey; Frank Sanfilippo; Michael Hobbs; Peter Thompson; Konrad Jamrozik
  • (b)Acquisition of DataKew-Kim Chew; Judith Finn; Nicholas Gibson
  • (c)Analysis and Interpretation of DataKew-Kim Chew; Judith Finn; Bronwyn Stuckey; Nicholas Gibson; Alexandra Bremner; Frank Sanfilippo; Konrad Jamrozik

Category 2

  • (a)Drafting the ArticleKew-Kim Chew
  • (b)Revising It for Intellectual ContentKew-Kim Chew; Judith Finn; Bronwyn Stuckey; Nicholas Gibson; Frank Sanfilippo; Alexandra Bremner; Peter Thompson; Michael Hobbs; Konrad Jamrozik

Category 3

  • (a)Final Approval of the Completed ArticleKew-Kim Chew; Judith Finn; Bronwyn Stuckey; Nicholas Gibson; Frank Sanfilippo; Alexandra Bremner; Peter Thompson; Michael Hobbs; Konrad Jamrozik

Ancillary