The association between lower urinary tract symptoms and erectile dysfunction in four centres: the UrEpik study


P. Boyle, Division of Epidemiology and Biostatistics, European Institute of Oncology, via Ripamonti 435, 20141 Milan, Italy.



To report a large-scale multinational investigation of lower urinary tract symptoms (LUTS) and sexual function, designed to investigate the independent association between them, as recent small-scale epidemiological studies suggest an association between benign prostatic hyperplasia and sexual dysfunction; both conditions are strongly associated with age and no study has been able to exclude age as a confounding factor in this relationship.


Culturally and linguistically validated versions of standard questionnaires were used to estimate the prevalence of LUTS (using the International Prostate Symptom Score, IPSS) and erectile dysfunction (ED) score (using O’Leary's Sexual Function Inventory) in regions of the UK (Birmingham), the Netherlands (Boxmeer), France (Auxerre) and Korea (Seoul). In each centre, stratified random samples were selected from population registers to provide representative samples of the population of men aged 40–79 years (and their partners) in each community. Direct interviews were held in Seoul and postal questionnaires used in the three European centres. The samples were selected randomly, providing representative samples in each community.


In all, 4800 men and 3674 women responded; the response rates among men were 77% in Boxmeer, 21% in Auxerre, 42% in Birmingham and 65% in Seoul. The overall prevalence of ED for men aged 40–79, estimated as an ED score of 0–4, was 21.1%. There was evidence of a linear increase with age (P < 0.001) and the pattern was very similar in the four centres. From the weighted logistic model, there was evidence of an association between sexual dysfunction, other self-reported diseases and lifestyle. From the ED score, after adjusting for age and country, men with diabetes were more likely to score of 0–4 (odds ratio 1.57, 95% confidence interval 1.09–2.25), as were those with high blood pressure (1.38, 1.09–1.75) and with an IPSS of 8–35 (1.39, 1.10–1.74). For lifestyle, smokers were more likely to score 0–4 (1.54, 1.23–1.92), while physical activity during leisure time was slightly associated with a reduction in the chance of scoring 0–4 (0.87, 0.77–0.99). The analysis gave the same results when repeated using self-reported ED instead of the dichotomised score.


ED is clearly age-related and a problem for a large proportion of men in the community. It can have a profound impact on the quality of life of the man and on his partner. Were all men with this problem to seek medical help there would be a large burden on healthcare systems. There are cultural and age effects on the assessment of this problem.


erectile dysfunction


sexual problem assessment.


The impact of erectile dysfunction (ED) on quality of life is not yet fully recognized and men tend to under-report this problem, both because it is not life-threatening and because it can be difficult to confront sexual dysfunction. However, ED is difficult to evaluate because it has many potential physical and psychological causes.

There is a clear need for more population-based studies to better assess the prevalence of ED, which is likely to increase with the ageing of the population [1], and to study the association with potential risk factors. From the Massachusetts Male Aging Study, Feldman et al.[2] reported, in the area around Boston, a prevalence of 52% for ED of any grade among men aged 40–70 years. According to Laumann et al.[3], in the USA the prevalence is 31% for men aged 18–59 years. In Europe, Virag and Beck-Ardilly [4] (France) estimated a prevalence of 39% for men aged 18–70 years; Dunn et al.[5] (England) reported 34% of men with at least one sexual problem; Parazzini et al.[6] (Italy) estimated an increasing linear trend with age from 2% (18–39) to 48% (≥ 70 years). Pinnock et al.[7] (South Australia) estimated the prevalence as 3% (40–49 years) to 64% (70–79), while Koskimoki et al.[8] (Finland) estimated a 26% prevalence of moderate/complete ED for men aged 50–70 years. Most studies conducted in previous years were based on selected patients, while those cited here are population-based. However, they used different questionnaires to assess ED, so restricting the possibility to compare the values reported or to use them as a reference.

The association between ED and other conditions, e.g. cardiovascular disease, diabetes, renal failure and LUTS, has been confirmed by many studies [9]. There was also evidence of a positive association with cigarette smoking and this was measured, together with alcohol, diet and physical activity, to assess how much lifestyle might affect ED [10]. Treatment-induced ED is often investigated, as the dysfunction can be a consequence of surgery or the treatment of BPH or prostate cancer [11,12], or caused by drugs prescribed to treat other conditions, e.g. hypertension, atherosclerosis and depression [13]. Hence the previous medical history and present conditions of the subjects are important in the analyses.

A more complicated task is to assess the causal relationship between general and specific quality of life. The mental health status must be analysed both as a cause and as a consequence of the ED, as their influence is reciprocal. It remains controversial whether the androgen decline in the ageing man, with manifestations like diminished sexual desire and erections, changes in mood, decreased intellectual activity or depression [14–16], is to be considered as an independent factor in ED.

Thus the aims of this study were, within a large epidemiological survey of urological conditions, to estimate the prevalence of ED, to investigate potentially associated comorbidity, to evaluate the impact on quality of life and the attitude towards specific consultation with a doctor.


The full details of the survey methods, population sampling, statistical analysis and response rates were reported previously [17]. ED was assessed using the Sexual Function Inventory, developed by O’Leary et al.[18], addressing four topics: sexual drive, erections (ED), sexual problem assessment (SPA) and overall satisfaction. ED and SPA scores (measured by three questions) were 0–12, the sexual drive score (two questions) 0–8, and for overall satisfaction there were five options (Appendix). For each of these measures, a higher score corresponded to a better condition. Subjects were also asked directly whether they had ED.

These subscores were analysed separately and then compared to measure their rank correlation using Spearman's coefficient. The agreement between scores and self-reported ED was measured using the weighted κ coefficient. Weighted logistic regression was applied, to relate ED, sexual drive and consultation with a doctor, to age, centre, physical status and lifestyle. A weighted linear regression was used to estimate the influence of sexual dysfunction on the physical and the mental components of the SF-12 [19].


In all, 4800 men participated to the study and, for the index man, the overall response rates were 77% in Boxmeer, 21% in Auxerre, 49% in Birmingham and 65% in Seoul. These rates are acceptably high in the Netherlands and Seoul but low in Auxerre and, consequently, a major part of the interpretation of these findings is from comparing findings among the centres.

In Korea none of the sexual questions were unanswered because the questionnaires were administered by direct interview, making ‘no response’ less likely. In the three European centres, 5–7% did not answer the questions relating to the four sexual topics; this increased with age, doubling with each decade from 40 to 60 years, and was higher in France.

The overall prevalence of ED for men aged 40–79 years, estimated from an ED score of 0–4, was 21.1%. There was evidence for a linear increase with age (P < 0.001) and the pattern was very similar in the four centres (Fig. 1a). The prevalence was lower when self-reported; overall 16.6% reported that they currently had ED (Table 1).

Figure 1.

Weighted percentages and 95% CI, for the various age groups in each country, of men with: a, ED , 0–4, b, SDRV 0–5, c, directly reporting ED, and d, SPA 0–5. (Boxmeer, green open circles; Auxerre, light green closed circles; Birmingham, red open squares; Seoul, light red closed squares).

Table 1.  Sexual dysfunction in the four centres, expressed as the ED score, self-reported ED, SPA and sexual drive score
 Weighted percentage
ED score
0      4.1      7.0    12.8      5.0      6.9
1–4    12.8    17.7    14.5    12.3    14.2
5–8    35.6    51.9    26.7    43.3    40.1
9–12    47.5    23.4    46.0    39.5    38.8
Self reported ED
Yes    12.8    23.9    20.5      11.3    16.6
No    87.2    76.1    79.5    88.7    83.4
SPA score
0–2      1.7      4.9      9.3      6.3      5.4
3–5      4.4      6.2      8.3    14.6      8.7
6–8      11.9      9.8    12.1    20.2    13.9
9–12    82.0    79.1    70.3    58.9    72.0
Sexual drive score
0–1      5.2      6.5    12.7      6.4      7.42
2–3    19.4    15.0    22.6    23.6    20.2
4–5    54.1    54.1    41.3    52.2    50.9
6–8    21.3    24.4    23.4    17.8    21.5

There was moderate agreement between the ED score (dichotomised into 0–4 vs 5–12) and self-reported ED (yes/no). Overall, adjusting for country, the agreement between the measures gave a κ of 0.52 (95% CI 0.49–0.55). There was no significant difference among the three European centres (Boxmeer 0.59, Auxerre 0.54, Birmingham 0.58; P = 0.50) while in Seoul it was significantly lower (k 0.33; P < 0.001).

The correlation between the ED and sexual drive score (Spearman's rank coefficient) in the four centres was Boxmeer 0.55, Auxerre 0.61, Birmingham 0.65 and Seoul 0.64 (all P < 0.001). Overall, adjusted for age and centre, the correlation was 0.60 (P < 0.001).

The impact on quality of life associated with problems with erections, sexual drive and ejaculation was more difficult to assess, because this aspect was more strongly influenced by the varying cultures. The same level of the symptoms might affect subjects in a different way, depending on their perception of the problem and their will to report such a state of mind (Fig. 1d). In particular, when measuring the Spearman's coefficient between ED score and SPA score, the overall partial correlation, adjusting for age and country, was only 0.25 (P < 0.001) and among the four centres it was Boxmeer 0.30, Auxerre 0.31, Birmingham 0.45 (all P < 0.001) and Seoul 0.05 (P = 0.05). When assessing the overall satisfaction with sexual life, the percentage of men who declared being ‘very dissatisfied’ or ‘mostly dissatisfied’ in the four centres were Boxmeer 11.1%, Auxerre 19.7%, Birmingham 22.4% and Seoul 9.9% (Fig. 1c,d).

From the weighted logistic model, there was evidence of an association between sexual dysfunction, other self-reported diseases and lifestyle. From the ED score, after adjusting for age and country, men with diabetes were more likely to score 0–4, as were those with high blood pressure and with an IPSS of 8–35 (Table 2). For lifestyle, smokers were more likely to score 0–4, while physical activity during leisure time was slightly associated with a reduction in the odds of scoring 0–4. The results were the same when the analysis was repeated using self-reported ED instead of the dichotomised score.

Table 2.  Different impact of comorbidity and lifestyle on ED and sexual drive. Multivariate estimates (95% CIs) adjusted for age and country, from a weighted logistic regression
Self-reportedOdds ratio (95% CI) for
ED score 0–4
vs 5–12
Sexual drive score
0–3 vs 4–8
  • *

    Moderate or regular vs none or light physical activity, during leisure time.

Diabetes  1.57 (1.09–2.25)  0.97 (0.68–1.37)
High blood pressure  1.38 (1.09–1.75)  0.96 (0.77–1.19)
Stroke  1.07 (0.58–1.99)  1.11 (0.61–2.02)
Heart attack  1.41 (0.88–2.24)  0.78 (0.50–1.22)
High cholesterol  0.97 (0.76–1.25)  1.02 (0.81–1.27)
Liver problems  1.55 (1.03–2.33)  0.95 (0.65–1.39)
Prostatitis  0.75 (0.49–1.14)  1.40 (0.94–2.08)
Kidney stones  1.02 (0.72–1.44)  0.96 (0.70–1.33)
UTIs  1.16 (0.72–1.86)  0.91 (0.58–1.41)
BPH  1.21 (0.83–1.78)  0.89 (0.61–1.29)
IPSS = 8–35  1.39 (1.10–1.74)  1.37 (1.11–1.70)
Smoker  1.54 (1.23–1.92)  1.01 (0.84–1.23)
Physical activity*  0.87 (0.77–0.99)  0.91 (0.82–1.02)
Sexual drive15.49 (12.51–19.17)
ED15.59 (12.60–19.30)

A similar model was fitted to assess whether these factors had the same influence on sexual drive and the results differed. The probability of having a low sexual drive score (<4), adjusted for age and country, increased with an IPSS of > 7, decreased with physical activity, but was unaffected by cigarette smoking. After adjusting for ED, which obviously was strongly associated, only an IPSS of > 7 was an important factor (Table 2), together with age and country, and none of the self-reported comorbidity was relevant to the sexual drive score.

On the other hand, the SPA seemed to be more complex, as many factors contributed to the extent to which subjects considered ED as a problem (Table 2). The effect of age was no longer significant after adjusting for ED score, and the percentage of men with a low SPA score increased progressively in Boxmeer, Auxerre, Birmingham and Seoul, indicating a cultural influence. While a low ED score was strongly associated with an increase in the odds of having a low SPA score, the sexual drive score was not. From the self-reported comorbidity, there was an association with heart attack and kidney stones, while prostatitis was of borderline importance. For major lifestyle factors there was an association with cigarette smoking but not with physical activity.

The sexual scores were also analysed as covariates influencing the general quality of life, as measured by the physical and mental components. This analysis was separate for the Korean centre, as it had a very different trend of the SF-12. A weighted linear regression was fitted, considering the sexual scores as continuous variables, instead of their categorizations. Table 3 shows that in the three European centres, after adjusting for age and IPSS, ED significantly influenced both mental and physical components, while SPA seemed to be an important predictor only for the mental. In Seoul, age was a very poor predictor for both the components; the effect of ED was halved and SPA did not affect the mental component.

Table 3.  The value (sem) of the weighted linear regression coefficients estimating the effects of sexual dysfunction on the mental and physical components of the SF-12
 Mental component scorePhysical component score
  • *

    original score, not categorized, included as a continuous variable. For the IPSS a high score indicates urinary dysfunction, for ED, sexual drive and SPA a low score indicates sexual dysfunction, and for the mental and physical components the higher the score the better the health status.

Korea 43.77 (0.74) 50.75 (0.76)
France39.14 (0.90) 50.77 (0.82) 
Holland44.94 (0.90) 50.18 (0.82) 
UK42.51 (0.91) 49.83 (0.82) 
Age   0.21 (0.02)   0.02 (0.02) −  0.13 (0.02) −  0.01 (0.03)
IPSS* −  0.27 (0.03) −  0.14 (0.03) −  0.34 (0.03) −  0.32 (0.03)
ED*   0.32 (0.08)   0.14 (0.08)   0.33 (0.08)   0.16 (0.09)
Sexual drive* −  0.13 (0.13)   0.09 (0.14)   0.03 (0.12)   0.59 (0.14)
SPA*   0.42 (0.07)   0.07 (0.04)   0.08 (0.06)   0.09 (0.04)

Self-reported ED increased with age but, while 35.7% of men in the whole sample aged 70–79 years reported currently having ED, only 4.8% went to the doctor for this reason. When considering all men in the study, ED-specific consultations had different trends with age in the four centres (Fig. 2). In Boxmeer the trend was linear, while in Birmingham it was quadratic, with the proportion consulting increasing from 40% to 60% and then starting to decrease. In Auxerre there was a similar quadratic trend but shifted forwards by a decade, while in Seoul the consultation was almost non-existent and unaffected by age.

Figure 2.

Different attitudes in the four centres towards doctor consultation, showing the weighted percentages of men who reported currently having ED (green closed circles), vs the weighted percentages of men who reported having consulted a doctor for this reason (red open circles); ED = “Yes”: Do you have erectile dysfunction? (Yes/No); DOC = ”Yes”: Did you see a doctor for this symptom? (Yes/No).

Those men with a self-reported ED were then considered in an attempt to determine which factors influenced their consulting a doctor. After adjusting for age and country, including a quadratic age effect for Birmingham, self-reported consultation was apparently influenced by some of the factors considered before. In particular, an increase in the odds of consulting was associated with physical activity, for the lifestyle, with SPA among the sexual scores, and with diabetes, high blood pressure, heart attack, prostatitis and BPH, among the comorbidity. A decrease in such odds was instead associated with self-reported high cholesterol and kidney stones, while no there was no evident association with the other covariates considered in Table 4.

Table 4.  The SPA score; multivariate estimates of the OR (95% CI) from a weighted logistic regression
FactorOdds ratio (95% CI) for SPA 0–5 vs 6–12
  • *

    Age is included in the model as a continuous variable; there was no significant age by country interaction.

  • † Moderate or regular vs none or light physical activity, during leisure time.

Auxerre1.62 (1.15–2.29)
Birmingham2.98 (2.14–4.14)
Seoul4.82 (3.49–6.64)
Age*0.99 (0.98–1.00)
Self-reported (Yes/No):
Diabetes1.32 (0.95–1.83)
High blood pressure1.19 (0.95–1.50)
Stroke0.61 (0.32–1.15)
Heart attack1.73 (1.12–2.69)
High cholesterol1.11 (0.87–1.41)
Liver problems0.92 (0.63–1.36)
Prostatitis1.42 (0.96–2.12)
Kidney stones1.53 (1.09–2.14)
UTIs0.93 (0.58–1.47)
BPH0.97 (0.66–1.44)
Smoker1.39 (1.13–1.71)
Physical activity0.96 (0.86–1.07)
IPSS 8–351.67 (1.34–2.08)
ED 0–47.12 (5.43–9.34)
Sexual drive 0–31.03 (0.79–1.33)


The present results confirm the need for more investigation of ED, as it affects a substantial proportion of men and its prevalence is likely to increase. It is important to use standard validated methods to collect data, to better assess such prevalence in the population and to compare the results from as many studies as possible. The combination of psychological and physical causes of sexual dysfunction complicates the causal relationship between ED and a low quality of life; Goldstein [16] postulated a mutually reinforcing relation between depressive symptoms, cardiovascular disease and ED. Many comparisons between the attitude and the actual symptoms are needed for a better understanding.

The positive association of ED with diabetes, cardiovascular diseases, cigarette smoking and LUTS is consistent with results from other surveys. Such an association implies the need to use information from the medical history within the analyses, and not only to prevent selection bias; it also implies investigating treatment-induced ED. Information on surgery- or drug-induced ED is fundamental to help the patient to evaluate the outcome of these treatments, but ED is assessed within the context of the concurrent disease(s) with even more complex implications. Although such a context cannot be ignored, it is more difficult to justify the choice of a treatment in terms of ED as a consequence. It is very important, not only for the prevalence estimate but also for the quality of life assessment, to consider a sample representative of the whole population.

In the future physicians are likely to see more patients reporting urological problems, simply because the world population is ageing. At the start of this century only 25% of the USA population lived to 65 years old. Currently, 70% are living to this age, with 30% reaching their eighth decade. Before the end of this century, life-expectancy is likely to exceed 80 years in several countries [21]. Furthermore, those born in the ‘baby boom’ after World War II are approaching the age of 60 years and early this century will reach an age when both sexual dysfunction and LUTS become important health concerns. Sadly there has been little work to quantify the impact of this ageing process in the future, but the effect of the ‘greying’ of the population that has already occurred is evident in the increasing burden of urological conditions in populations around the world.

The present study provides an unequivocal demonstration of the association between ED and LUTS that cannot be explained simply by considering age or other comorbidity. This has several important implications, including the potential to open new research areas that might explain the joint pathophysiology. In this context, evidence has recently emerged linking abnormally high activity within the sympathetic nervous system and LUTS [22]. Likewise, a close relationship between the sympathetic nervous system and penile erection and the pathophysiology of ED is well-documented [23], as is the differential impact on sexual function of therapies commonly used for treating BPH [24–26].

In terms of clinical practice, these findings have two important implications. First, among men who reported having ED a very low percentage also reported having consulted a doctor for this reason. Together with a gradual physical reduction of erectile function, the psychological component is also important when self-reporting such a dysfunction; older men might accept the condition as an ageing feature and consider it less of a problem. From this study there was evidence of a strong difference, among the four centres, in the attitude towards consultation for ED, especially in the Korean centre. Such cultural influence should be accounted for within international comparisons of this kind. An encouragement to consult for these problems and a wider diffusion of the information on prevalence and treatments should be pursued. When men consult for other problems, the opportunity to discuss sexual dysfunction could prove useful.

Second, the information generated by this survey is of fundamental significance for the doctor when confronted with a patient complaining of LUTS or sexual dysfunction. It is essential that both primary-care physicians and specialists in this situation are aware of these associations and consider them when assessing patient treatment choices and discussing them with the patient. At all stages of the follow-up of patients with LUTS, doctors must probe for changes in the various aspects of sexual function and activity.


This contribution of Peter Boyle, Chris Robertson and Chiara Mazzetta to this research was within the framework of support from the Associazione Italiana per la Ricerca sul Cancro (AIRC) (Italian Association for Cancer Research). The study was funded by an unrestricted research grant from GlaxoWellcome.


Table 5. Questions used to assess sexual dysfuction.
SEXUAL DRIVENo daysOnly a few
Some daysMost daysAlmost
every day
1. During the past month, on how many days have you felt sexual drive?     
 None at allLowMediumMedium highHigh
2. During the past month, how would you rate your level of sexual drive?     
ERECTIONSNot at allA few timesFairly oftenUsuallyAlways
3. Over the past month, how often have you had partial or full sexual
erections when you were sexually stimulated in any way?
4. Over the past month, when you had erections, how often were they firm
enough to have sexual intercourse?
 Did not get
erections at
A lot of
No difficulty
5. How much difficulty did you have getting an erection during the past
Very small
No problem
6. In the past month, to what extent have you considered a lack of sex drive
to be a problem?
7. In the past month, to what extent have you considered your ability to get
and keep erections to be a problem?
8. In the past month, to what extent have you considered your ejaculation
to be a problem?
Neutral or mixed
(about equally)
satisfied and
9. Overall, during the past month, how satisfied have you been with
your sex life?