Depression and lower urinary tract symptoms: Two important correlates of erectile dysfunction in middle-aged men in Hong Kong, China


Samuel Y. S. Wong, md, ccfpc, fracgp, Department of Community and Family Medicine, The Chinese University of Hong Kong, 4/F, School of Public Health, Prince of Wales Hospital, Shatin, N.T., Hong Kong.


Aim: To evaluate the correlates of erectile dysfunction (ED) in Hong Kong middle-aged Chinese men aged 45–64 years.

Methods: A community-based cross-sectional household survey was performed in Hong Kong. The Chinese abridged version of the International Index of Erectile Function (IIEF-5) was used to measure erectile function. The International Prostate Symptom Score (IPSS) was used to measure lower urinary tract symptoms (LUTS) and depressive symptoms were mesured by the Center for Epidemiological Studies Depression Scale (CES-D). Demographic and lifestyle data were also collected. The association between ED and its correlates was analyzed using bivariate and multivariate analyses.

Results: Of the 545 subjects who agreed to participate in the survey, 75 refused to answer questions about their sexual activities and function. Out of those who responded, 118 (22%) subjects were not sexually active (not sexually active over the past 4 weeks). Out of 352 subjects, 60.3% suffered from some degree of ED. Age, presence of depression defined by CES-D and moderate LUTS were associated with increased odds of having ED. In multivariate analysis, depressive symptoms identified by CES-D (OR = 2.3, CI: 1.2–4.6) and moderate LUTS (OR = 3.7, CI: 1.6–8.3) were independently associated with increased odds of having ED.

Conclusion: ED is an important public health problem in Chinese middle-aged men, with more than half suffering from some degree of ED. Depression and LUTS were significant and important risk factors associated with ED.


Erectile dysfunction (ED) is a significant public health problem with a large impact on psychosocial well-being in men with a high prevalence.1–3 Apart from age, ED has been shown to be related to a number of medical factors. These inlcude hypertension,4–6 diabetes mellitus,4,6 stroke,6 high cholesterol,6 vascular diseases,6 intake of sleeping pills and antihypertensives,7 and depressive symptoms.8

For lifestyle risk factors, smoking was found to be related to ED in several studies.4,9 Conversely, the relationship between ED and alcohol consumption was not as clear because both negative10 and positive associations were found.4 Physical activity was associated with a lower risk for ED in two large studies6,11 while obesity was associated with a higher risk.6

Most studies of ED have been conducted in Western countries. Only a few studies were reported in Asia.10,12,13 As the Asian population continues to grow and occupies more than one-third of the world’s populaiton, it is important to study this significant public health problem in Asian countries. We have hereby undertaken the first population-based study of the prevalence and correlates of ED in Chinese Hong Kong, China.

Materials and methods

Study population and sampling frame

The study was a cross-sectional, population-based, household survey conducted from September 2003 to July 2004 in Hong Kong. The total population in Hong Kong is 6.7 million. Ninety-nine percent of families live in housing blocks, which are either owned by the families (private housing) or by the Hong Kong SAR Governement (public housing). There are neither electorate registers nor general practitioner registers in Hong Kong for population sampling. Cluster sampling is the most frequently used sampling method, where subjects are interviewed face-to-face in their homes.14

Geographically, Hong Kong is made up of two parts; the Kowloon peninsula and the Hong Kong island. Appoximately 80% of the total population resided in Kowloon peninsula, in which about half of the residents lived in public housing. In this study, the community sample was drawn from the Kowloon peninsula. A list of housing blocks in the peninsular was obtained and were stratified into private and public housing to avoid over-sampling of public housing blocks. Cluster samples of households were randomly selected. A total of 11 752 households from three public estates and two private estates were included. The study was approved by the Joint Chinese University of Hong Kong and the New Territories East Cluster Clinical Research Ethics Committee.

Trained interviewers visited all households identified in the cluster sample. Informed consent was obtained from eligibile subjects (aged 45–64 years) who agreed to participate in the study. This was followed by the administration of a standardized questoinnaire which took approximately 15–20 min to complete. Appointments were made to interview eligible subjects who were not available at the time of visit. Among the 753 men in the appropriate age groups identified, 208 refused to participate. Thus the total population was 545 men, and the overall response rate was 72.4% (Appendix I).


A structured questoinnaire, containing approximately 83 items, was administered in person to each subject by the means of self-report. This data collection method was widely used in similar studies on ED.15,16

ED was measured by the International Index of Erectile Function (IIEF). This is a multidimensional scale for assessment of erectile function with excellent reliability, validity and sensitivity. The abridged validated five-item version, IIEF-5,17 was used in this study. This was first translated into Chinese and then back-translated into English by a team of experienced and bilingual ‘field workers’ who had translated and participated in previous questionnaire surveys. Five items were selected from the IIEF-5 and were based on ability to identify the presence or absence of ED and on adherence to the National Institute of Health’s definition of ED over the previous month. ED was classified as normal (22–25), mild (12–21), moderate (8–11) and severe (5–7) based on the answers to the five questions.

The questionnaire also included potential medical, sociodemographic and health related covariates of ED. Current depressive symptoms were assessed by the validated Chinese shortened (10-item) version of the Center for Epidemiologic Studies Depression Scale (CES-D 10),18 with scores ranging from 0 to 10. A cut-off of 4 or less was defined as being depressed. This screening tool has been well-validated and highly recognized in worldwide community studies on depression.19–21 Lower urinary tract symptoms (LUTS) were measured by the International Prostate Symptom Score (IPSS),22 with prostate symptoms classified as absent or mild (IPSS <7), moderate (8–19), or severe (20–35). Smokers were classified as former or current smokers that included men who smoked cigarettes, cigars or pipes. Alcohol drinkers included men who reported drinking wine, beer or spirits for more than 12 times over the past 12 months. Physical activity was measured by asking subjects the frequency of exercise performed, and of walking outdoors, upstairs, uphill and with a load during the previous week. These questions were used in previous studies23 and were found to be valid and reliable measures of physical activity. Subjects were considered sexually active if they reported at least one episode of sexual intercourse during the previous month.

Analysis and statistical methods

Data were analyzed using SAS (Cary, NC, USA) statistical software. The association between ED and its correlates was analyzed by bivariate and multivariate analyses. ED was dichotomized as ‘none’ or ‘mild’ versus ‘moderate’ or ‘severe’ in all bivariate and multivariate analyses. For each independent variable, crude and age-adjusted bivariate odds ratios (ORs) and 95% confidence intervals (CIs) were obtained. A two-sided P-valued of 0.05 or less was considered statistically significant and the χ2 test and Student’s t-test and analysis of variance (anova) were used for categorical and continuous variables, respectively. All significant variables in the primary bivariate analyses were entered in the logistic regression.


Demographic, medical and lifestyle characteristics of the study population are shown in Table 1. The mean age was 54.3 years, 92% were married, 3% were living alone, 66.5% were currently employed, 52.1% had primary or fewer years of education, 31.1% currently smoked tobacco and 40.7% currently consumed alcohol. A history of comorbid medical conditions included 6.4% men with diabetes, 19.5% with hypertension, 2.6% with heart disease, 0.6% with depression and 2.2% with any prostate disease. Only three men used antidepressants and four used sleeping pills in our sample.

Table 1.  Demographic variables
Demographic characteristics(N = 545) n (%)
  1. CESD, Center for Epidemiological Studies Depression Scale; ED, erectile dysfunction.

Age (years)
45–49135 (24.77)
50–54153 (28.07)
55–59137 (25.14)
60–64120 (22.02)
Marital status
 Married or living in a married-like relationship501 (91.93)
 Divorced, widowed, separated 23 (4.22)
 Single, never married 21 (3.85)
 Live alone 16 (2.94)
 Currently employed361 (66.48)
 Primary school or less284 (52.11)
 Secondary school144 (26.42)
 Graduated from college or university117 (21.47)
Tobacco use
 Former101 (18.60)
 Current169 (31.12)
 Alcohol consumption221 (40.70)
Self-report medical diseases
 Diabetes 35 (6.43)
 Hypertension106 (19.49)
 Heart disease 14 (2.57)
 Depression  3 (0.55)
 Antidepressants use  3 (0.55)
 Benzodiapines/zoplicone use  4 (0.73)
 Prostate disease 12 (2.21)
Severity of ED dysfunctionN = 352
 Severe  0 (0)
 Moderate  9 (2.56)
 Mild203 (57.67)
 Normal140 (39.77)
 Depression 99 (18.17)

Of the 545 subjects who agreed to participate in the survey, 75 refused to answer questions about their sexual activities and function. A comparison in demographics and medical variables in those who answered questions on ED and those who refused are shown in Table 2. Those who refused to answer questions regarding their sexual activity were older (P < 0.05 by χ2) and were less likely to be regular drinkers of alcohol (P < 0.05 by χ2).

Table 2.  Comparison of selected demographic variables in those who answered questions on the IIEF-5 and those who refused questions on the IIEF-5 (N = 545)
N = 470 (%)
N = 75 (%)
P-value (χ2)
  • *

    P < 0.05. IIEF, International Index of Erectile Function.

Age (years)
 45–49113 (24.04)22 (29.33)0.0261*
 50–54142 (30.21)11 (14.67) 
 55–59118 (25.11)19 (25.33) 
 60–64 97 (20.64)23 (30.67) 
Marital status
 Married or living in a married-like relationship431 (91.70)70 (93.33)0.4259
 Divorced, widowed, separated 19 (4.04) 4 (5.33) 
 Single, never married 20 (4.26) 1 (1.33) 
Live alone 13 (2.77) 3 (4.00)0.5566
Currently employed319 (68.02)42 (56.76)0.0818
 Primary school or less239 (50.85)45 (60.00)0.3314
 Secondary school128 (27.23)16 (21.33) 
 Graduated from college or university103 (21.91)14 (18.67) 
 Tobacco use
  Former 89 (18.98)12 (16.22)0.6873
  Current143 (30.49)26 (35.13) 
 Alcohol consumption199 (42.43)22 (29.73)0.0387*
Self-report medical diseases
 Diabetes 32 (6.82) 3 (4.00)0.4546
 Hypertension 94 (20.04)12 (16.00)0.5298
 Heart disease 11 (2.35) 3 (4.00)0.4236
 Depression  3 (0.64) 0 (0)1.0000
 Prostate disease 10 (2.13) 2 (2.67)0.6751

Out of those who responded, 118 (22%) subjects were not sexually active (not sexually active over the past 4 weeks) and could not answer questions on IIEF-5. As a result, 352 subjects completed all parts of the IIEF-5.

Comparing the demographic and medical variables of sexually active and inactive men in married or cohabitated relationships, those who were sexually inactive over the previous 4 weeks were more likely to be older (P < 0.001 by χ2), less likely to be employed (P < 0.0001 by χ2), less likely to consume alcohol (P < 0.05 by χ2) and more likely to have diabetes (P < 0.05 by χ2) (Table 3).

Table 3.  Comparison of selected demographic variables of subjects in married or cohabitated relationships who were sexually active and those who were not sexually active (N = 431)
CharacteristicSexually active
N = 352 (%)
Not sexually active
N = 99 (%)
P-value (χ2)
  • *

    P < 0.05.

Age (years)
 45–49 94 (28.31) 9 (9.09)<0.0001*
 50–54106 (31.93)16 (16.16) 
 55–59 86 (25.90)26 (26.26) 
 60–64 46 (13.86)48 (48.48) 
Currently employed248 (74.70)49 (50)<0.0001*
 Primary school or less163 (49.10)50 (50.51) 0.8999
 Secondary school 95 (28.61)26 (26.26) 
 Graduated from college or university 74 (22.29)23 (23.23) 
 Tobacco use
  Former 57 (17.17)25 (25.25) 0.1834
  Current 97 (29.22)28 (28.28) 
 Alcohol consumption153 (46.08)29 (29.29) 0.0030*
Clinical diagnosis
 Diabetes 14 (4.22)15 (15.31) 0.0001*
 Hypertension 61 (18.37)25 (25.51) 0.1207
 Heart disease  6 (1.81) 5 (5.10) 0.0695
 Depression  1 (0.30) 2 (2.04) 0.0691
 Prostate disease  6 (1.81) 4 (4.08) 0.1893


The prevalence of any degree of ED in our sample was 60.3% (mild 57.7%, moderate 2.6%, severe 0%). The prevalence of any ED increased with age from 49.5% in men aged 45–49 years to 69.4% in men aged 60–64 years. The prevalence of moderate ED increased with age, from 1% in men aged 45–49 years to 8.2% in men aged 60–64 years. There was also an increased prevalence of mild ED from 48.5% in men aged 45–49 years to 61.2% in men aged 60–64 years, although the increase was non-linear and plateaued at age 50–54 years. For those who were taking antidepressants, one was not sexaully active and two suffered from mild ED.

Bivariate age–adjusted associations

The age-adjusted bivariate associations between ED and potential covariates are shown in Table 4, comparing non-ED (control) to any ED (cases). As there were very few men who complained of moderate to severe ED, we could only compare those with normal erectile function to those with any ED, including mild ED. Age, presence of depression defined by CES-D and moderate LUTS were associated with increased odds of having ED.

Table 4.  Multivariate logistic regression model, prevalence odds ratios for mild/moderate/severe erectile dysfunction (ED) versus non-ED
CharacteristicOR (95% CI)
  1. * P < 0.05. †Odds ratio (95% confidence interval). CESD, Center for Epidemiological Studies Depression Scale; LUTS, lower urinary tract symptoms.

Age (years)
 50–541.732 (0.988–3.036)
 55–591.639 (0.889–3.021)
 60–642.056 (0.951–4.448)
 Depressed2.301 (1.162–4.560)*
 Mild1.483 (0.814–2.700)
 Moderate3.668 (1.605–8.382)*
 Severe2.844 (0.513–15.766)


This is the first household survey conducted in Hong Kong to investigate the prevalence and correlates of ED in middle-aged men. The prevalence of ED in our sample (men aged 45–64 years) was 60%. In the Massachusetts Male Aging Study, 39% of men between the ages of 40-50 years, 46% of men between 50 and 60 years and 70% of men aged over 70 years suffered from ED. Studies in other countries showed that the prevalence of ED ranged from 38% (Thai men aged 40–70 years) to 51.3% (men aged 30 and above in Singapore).

One significant finding in our study is the low prevalence of moderate and severe ED. This could be explained in several ways. First, we have restricted the age of our sample to those aged 45–64 years, while most other studies have included men in their 70s. Second, only those men who have been sexually active in the previous 4 weeks were asked the IIEF-5 questions. In comparing the demographics and medical history of those sexually active and inactive men in marriage or cohabitation relationships, sexually inactive men resemble the ED patients, in that they tended to be older and have diabetes mellitus. As many men who suffer from ED may not be sexually active, this could have significantly underestimated the prevalence of ED. Third, as the interviews were conducted at household visits, men could have been reluctant to answer these sensitive questions when their families might have been around at the time of the interview, resulting in underestimation of prevalence.

As was demonstrated in previous studies in Asia10,12,13 and the West,2–4 increasing age is correlated with prevalence of ED after controlling for all other significant correlates of the condition. However, among all potential risk factors associated with ED, only depression and moderate LUTS were found to be independently associated with ED in this study. Previous studies in other Asian countries showed that hypertension, diabetes, heart disease and lifestyle factors such as smoking were independently associated with ED. The lack of significant association between these other risk factors and ED in our study could be due to the comparatively small sample size, and the small number of men who suffered from moderate to severe ED. Conversely, the strong association between depressive symptoms (adjusted OR = 2.3) and moderate LUTS (adjusted OR = 3.7) and ED in our study, despite the small sample size, showed that depressive symptoms and LUTS are indeed important risk factors for ED in middle-aged Chinese men.

This is the first study to demonstrate an independent association between depressive symptoms and ED in Chinese men. Previous cross-sectional studies conducted in the USA and other countries8,24 showed similar significant associations between depressive symptoms and ED, although the instruments for the assessment of depression varied in other studies. Although many of the antidepressant medications could cause ED, only three men in our sample used antidepressants and thus antidepressant use was unlikely to account for the association between ED and depression in our study.

It was shown in a large prospective study5 that depression observed in those with ED was likely to be caused by ED itself rather than vice versa. In this longitudinal study, depressive symptoms at baseline were shown not to be a predictor of the development ED over an 8-year period.

Indeed, the association between ED and LUTS has received considerable attention recently.25 A large multinational survey,26 the MSAM-7, conducted in the USA and six European countries in 12 815 men aged 50–80 years showed that LUTS is a major risk factor for sexual dysfunction in older men, independent of other risk factors including age, hypertension and diabetes. Both erection and ejection problems were shown to be more than twice as common in men with moderate to severe LUTS than in those with none or mild LUTS. Similarly, in another multinational survey27 conducted in the UK, the Netherlands, France and Korea in men aged 40–79 years, it was shown that ED was associated with LUTS independent of other potential confounders including age and other comorbidities.

In a recent study conducted in urological patients,28 depression and obstructive IPSS (LUTS) were shown to be the only statistically significant predictors of ED, even after controlling for age and comorbidities. This finding is similar to results from our current study.

Although the pathogenetic relationship between LUTS and ED is not yet completely understood, some suggest that both psychosocial and pathophysiological processes29–31 could account for the observation. In a study by Frankel et al.,29 men who experienced LUTS were more likely to experience ED, as well as be bothered by ED. Moreover, they were more likely to attribute worsening of both ED and sexual bother to their LUTS as a causative agent. Others suggest that the association could be from common pathophysiological processes.29–31 For example, the alpha-adrenergic receptor may be the common link between the two diseases,30 as it affects the bladder neck and prostatic smooth muscle tone of benign prostate hyperplasia (BPH), as well as the vascular tone of the corpora cavernosa in ED. In a study conducted in male rabbits, Khan et al.29 showed that partial bladder outlet obstruction in male rabbits could result in loss of smooth muscle in the rat corpora cavernosa. This suggests that both nitric oxide and the endothelin-1 pathway could be disrupted in both the upper and lower urinary tracts and the corpus cavernosum.

Several limitations of the present study are worth noting. First of all, out of the 753 eligible men approached, only 72.4% (545 men) agreed to be interviewed. As these men refused to answer questions asked by the trained interviewers, differences between the responders and those who refused were unknown. As a result, a selection bias could not be prevented. Second, as this is only a cross-sectional survey, a temporal relationship between risk factors and ED could not be determined. As a result, the contribution to the etiology of ED is limited as only associations could be found from this study design. Third, the survey was conducted in middle-aged men. As a result, the findings could only be applied to middle-aged men. Fourth, except for depression and LUTS, only self report of medical diseases was used. As a result, the presence of common diseases such as hypertension and diabetes mellitus could have been underestimated and this might have affected the study findings. As this survey was conducted during household visits, the prevalence of ED and other disorders such as depressive symptoms could be underestimated, as these men might have felt embarrassed about responding truthfully to questions in the presence of their families. Finally, as 22% (118) of the men were not sexually active for 4 weeks prior to the household visits and were not eligible to answer the IIEF-5 questions, we could not determine if these men who have not been sexually active suffered from ED. This might have underestimated the extent of ED in our sample.


We showed that ED is an important public health problem in Chinese middle-aged men, with more than half suffering from some degree of ED. Moreover, we showed that depression and LUTS were significant and important risk factors associated with ED, although further longitudinal studies in Chinese men are needed to delineate the causality between LUTS, depression and ED. As ED is prevalent in middle-aged Chinese men, clinicians should increase their awareness of ED in middle-aged and older men, particularly in those who suffer from depression or LUTS at the same time.


We thank Dr Edith Lau for her earlier efforts in obtaining funding and her earlier contributions to the initial design of this study. This study was supported by a donation from Pfizer, Hong Kong.


Appendix I: Distribution of the households and response rate of subjects

 Total flatsFinished questionnaireEligible subjectResponse rate (%)
Public estate
 Estate A 357318528964.01
 Estate B 428410715270.39
 Estate C 592*152171.43
 Sub-total 844930746266.45
Private estate
 Estate D 13688511077.27
 Estate E 193515918785.03
 Subtotal 330324429782.15
Total11 75255175972.60
 Response rate (%)
  • Excluded six interviewed subjects who were ineligible (five from Estate A, one from Estate E).

  • *

    Parts of Estate C were visited until the intended sample size was reached; – Response rate, Finished Questionnaire/Eligible Subject.

 Estate A63.38
Public estate66.08
 Estate E84.95
Private estate82.09