The prevalence of lower urinary tract symptoms in men and women in four centres. The UrEpik study

Authors


P. Boyle, Division of Epidemiology and Biostatistics, European Institute of Oncology, via Ripamonti 435, 20141 Milan, Italy.
e-mail: director.epi@ieo.it

Abstract

The topics covered in this section relate to areas of considerable interest for urologists everywhere. The UrEpik study is introduced in the ‘Comments’ section and the results of the study will appear in a series of papers published over the next few months. They will give a new insight into many aspects of LUTS.

Authors from London investigate the place of urodynamics in the evaluation of patients with symptoms of overactive bladder. They found that patients with such symptoms and with apparently normal urodynamic findings often respond equally well to antimuscarinic therapy as do those with positive urodynamic findings. They indicate that these findings cast doubt on the value of urodynamics in such patients before treatment.

Authors from North America address the problem of mixed urinary incontinence. They examined the efficacy of tolterodine in patients with this condition and compared it with patients having urge incontinence alone. They found that it was equally effective in reducing leakage and other symptoms of overactive bladder in patients with mixed incontinence as in urge incontinence alone.

OBJECTIVE

To evaluate the epidemiology of lower urinary tract symptoms (LUTS) among men and women, as there are significant unanswered questions about the prevalence and impact of LUTS in different populations.

SUBJECTS AND METHODS

A population-based, cross-sectional survey was completed in Boxmeer (the Netherlands), Auxerre (France), Birmingham (UK) and Seoul (Republic of Korea), using culturally and linguistically validated versions of the International Prostate Symptom Score (IPSS). The aim was to estimate the distribution of symptoms of LUTS in men and women. Stratified random samples of men aged 40–79 years in each community were collected. Postal questionnaires were used in Europe and direct interviews in Korea.

RESULTS

In all, 4979 index men and 3790 women were recruited, with age-adjusted response rates among men of 72% in Boxmeer, 28% in Auxerre, 60% in Birmingham and 68% in Seoul. The percentages of men and women with an IPSS of 8–35, indicating moderate to severe symptoms, were, respectively, 20.7 and 18.0 (Boxmeer); 19.2 and 12.6 (Auxerre); 25.1 and 23.7 (Birmingham); 16.2 and 19.9 (Seoul). Among women the relationship between symptoms and age was not as strong as in men. The percentages of men and women with moderate to severe symptoms were by age group, respectively, 10.6, 15.5 (40–49); 19.0, 18.2 (50–59); 30.5, 23.8 (60–69); 40.4, 28.7 (70–79). Among those aged 40–49 the main differences between men and women were in the questions about frequency of urination during the day and holding back urine. Among the older groups men reported more symptoms on all questions apart from urination at night and difficulty in holding back urine, both of which were equally prevalent among men and women.

CONCLUSIONS

The overall prevalence of LUTS was high and showed no marked cultural variation. Prevalence increased with age, with severe LUTS commoner in older men. Women reported similar levels of the symptoms traditionally associated with LUTS in men. In each age group there were no major cultural differences in the frequency of LUTS. There were differences with age between men and women; younger men had a lower prevalence of LUTS than younger women but older men a much higher prevalence than older women. These findings emphasize that the IPSS should be confined to within-patient comparisons and not used as a diagnostic tool. The IPSS performs very similarly regardless of gender.

INTRODUCTION

Before the age of 40 years LUTS are relatively uncommon but the prevalence increases with age such that a large proportion of men and women aged > 70 years may have them. This condition can be a major contributor to a reduced health-related quality of life (QoL) and the consequent psychological sequelae in many ageing people [1]. BPH has been the focus of the greatest body of epidemiological research. Although major advances have been made in the last 10 years in improving epidemiological knowledge, there are several important issues which remain poorly understood [2]. The impact of lifestyle factors as determinants of the risk of developing BPH is unknown and prospects for prevention are poor.

LUTS are not specific and can be increased by factors unrelated to prostatic conditions, including diet, fluid intake, alcohol intake, and anticholinergic effects of commonly used medications available without prescription [3]. It is difficult to draw conclusions about causes from cross-cultural comparisons of symptoms. Nonetheless, such comparisons are useful in describing the patient-perceived problems, and within-country analyses may help to refine methods for the study of urological conditions.

Epidemiological surveys of LUTS [2,4,5] give estimates of prevalence of up to 50% in some populations, with large variations evident even within the same age strata. Many earlier surveys were not population-based, using a random sampling mechanism [6], although many more recent ones have been [7–12]. All studies suffer, to some extent, from the lack of a common definition of the condition, although recently there has been a convergence towards the use of the AUA Symptom Index (AUA-SI) to measure LUTS [13].

SUBJECTS AND METHODS

A population-based, cross-sectional survey was completed in four centres: Boxmeer (the Netherlands), Auxerre (France), Birmingham (UK) and Seoul (Korea) [4,14]. A random sample of men aged 40–79 years was recruited in each centre. If the index man was living with a female partner then the woman was also invited to participate in the study. Culturally and linguistically validated versions of standard questionnaires [15], supplemented with extra questions, were administered. Self-completed postal questionnaires were used in the UK, Netherlands and France, while direct interviews were used in Korea.

A sample size of 1300 was required to estimate the overall prevalence of moderate to severe LUTS to within 3% in each centre, calculated using a normal approximation, specifying that the width of the 95% CI must be 3% to either side of the true value, and assuming the true value range was 40–60%. Recent surveys suggest that this was an overestimate and the precision of a prevalence of 20% is ± 2%. Different response rates were anticipated in each centre and the number of men contacted was varied to ensure a target sample of 1300 [14].

In each centre random samples of index men were selected. Local authority administrative registers were used in the Netherlands and Korea, GP registers in the UK and electoral registers in France. Stratification by age was used in the Netherlands, France and Korea, while stratification was by Townsend score [16], an index of social deprivation, in the GP practices in the UK. In Seoul, only couples were included in the study.

Prevalence was estimated using sampling weights based on the ratio of the number of men in each age group in the samples to the number of men in that age group in the population [14]. The sampling weights for women were calculated in the same way. All data presented are based on the weighted data, necessary because a stratified random sample was used and the number of men selected in each age group was not proportional to the number of men in the population. The Auxerre sample had the poorest response rate and results are presented separately by centre, or in pooled analyses adjusted for centre.

The relationship between moderate to severe LUTS or severe LUTS and age, and other factors, e.g. centre and physical activity, was estimated using weighted logistic regression. The results are reported in the text as odds ratios (OR) and 95% CI. All statistical calculations were carried out using SAS (1996, SAS Institute Inc., Cary, NC). Physical activity was recorded in two ways, as the response to ‘How would you describe your level of physical activity during work hours?’ and ‘How would you describe your level of physical activity during leisure time?’ Smoking status was classified as never, current or ex-smoker.

RESULTS

In all, 4979 index men and 3790 partners responded, including 3474 couples. Responses were received from 326 women whose partner did not reply, from 513 single men, and from 992 men whose partner did not reply. For the index man, the response rates were 72% in Boxmeer, 28% in Auxerre, 49% in Birmingham and 68% in Seoul. Adjusting the Birmingham response rate to the age distribution in the other centres gives an age-adjusted response rate of 60%.

There was a remarkable similarity in the IPSS distributions for men and women over the four centres (Table 1). Among men, the prevalence of severe LUTS, assessed by an IPSS of 20–35, ranged from 3.0% of men aged 40–79 in Auxerre to 5.1% in Birmingham. The rates for women were only slightly lower in each centre, ranging from 1.6% in Auxerre to 4.2% in Birmingham. The percentages of men and women reporting moderate to severe symptoms was higher in Birmingham (Table 1).

Table 1.  The reporting of the severity of LUTS by men and women in the four centres
GroupCentre
AuxerreBoxmeerBirminghamSeoul
  • *

    The Birmingham sample was a simple random sample, whereas the samples in the others centres were stratified by age. Adjusting the response rate in Birmingham to mimic the same age stratification gives an estimated response rate of 60%. For men the sems of the percentages are 0.6% for those near 5%, and 1.2% for those near 20%; thus the CIs are ± 1.2% and ± 2.4%, respectively. The corresponding sems for women in Auxerre are 0.9% and 1.6%, giving CIs of ± 1.8% and ± 3.2%, respectively. From these values it is clear that the CIs for prevalence will overlap considerably.

Response rate, %  28  72  49*  68
Men
N1217123311691360
IPSS group, %
0–7  80.8  79.3  74.9  83.8
8–19  16.1  17.3  20.0  12.5
20–35    3.0    3.4    5.1    3.7
Women
N  6121066  7221360
IPSS group, %
0–7  87.4  82.0  76.3  80.1
8–19  11.0  15.0  19.5  16.6
20–35    1.6    3.0    4.2    3.3

Among men the prevalence of LUTS increased with age, with 5.4% of men aged 60–69 and 7.5% of men aged 70–79 reporting symptom scores of 20–35. There were similar trends of increasing severity with increasing age across all centres (P = 0.34). There were no significant differences in age-specific rates of severe symptoms over the four centres (P = 0.07). In the three European centres 6.6% of men aged 70–79 reported an IPSS of 20–35, compared with 1.8% of men aged 40–49. In Seoul, the corresponding values were 13.0% and 2.5%, respectively.

The odds of a man having moderate to severe LUTS (an IPSS of 8–35) increased with age at the same rate per year as for severe LUTS in all centres (P = 0.08). There were differences among centres in the prevalence of moderate to severe symptoms (P < 0.001). A higher prevalence of moderate to severe LUTS was reported in Birmingham (Fig. 1).

Figure 1.

The percentage of men (a) and women (b) with moderate to severe symptoms, by age and centre (Boxmeer, green open circles; Auxerre, light green closed circles; Birmingham, red open squares; Seoul, light red closed squares).

The prevalence of an IPSS of 20–35 among women increased with age but there was no evidence of any systematic differences among centres (P = 0.08). The relationship between age and the prevalence of mild to moderate symptoms was not the same in all centres for women (P = 0.01); in Birmingham there was only a very weak increase in symptoms with increasing age.

The relationship between the prevalence of the symptoms and age was stronger among men (Fig. 1). Among men the mean (se) prevalence of moderate to severe symptoms increased by 9.8 (0.57)% per decade from 40 to 79 years while among women the increase was only 4.6 (0.62)%. In Birmingham the increase was 1.6 (1.59)% per decade and 5.3 (0.69)% per decade in the other three centres. The increase in the prevalence of severe symptoms was also steeper in men than in women, at 1.8 (0.28)% and 1.0 (0.28)% per decade, respectively.

Among men, greater physical activity at work, involving lifting and heavy manual labour, was associated with more LUTS than in men whose work was mainly sedentary or involved no lifting. The OR of moderate to severe LUTS was 1.30 (1.11–1.53), while that of severe LUTS was 1.79 (1.31–2.45). This association between physical activity at work and LUTS explained most of the differences in prevalence among centres. The highest prevalence was reported in Birmingham, where 39% of men were involved in heavy manual labour, compared with 21% of men in Auxerre, 29% in Boxmeer and 33% in Seoul. Independently of physical activity at work, men who were currently involved in regular walking or at least moderate exercise had a lower prevalence than those who were mostly sedentary or only engaged in light exercise or some walking. The OR was 0.80 (0.66–0.98) for severe and 0.92 (0.83–1.01) for moderate to severe LUTS. Among women there was no association between LUTS and physical activity at work or during leisure. There was no association between smoking status and LUTS in men or women.

The mean percentage contributions of each question to the total IPSS were similar in men and women (Fig. 2). The two frequency questions contributed slightly more than the other questions. Frequency of urination was the most common symptom reported by young women and the prevalence of this symptom did not increase with age among women (Table 2). This symptom was one of the two most common among men aged 40–49 years, along with frequency of night-time urination, the prevalence of both increasing with age among men.

Figure 2.

The contribution of the individual questions to the total IPSS in men (a) and women (b) in each centre.

Table 2.  The percentage of men and women with scores of ≥ 2 on the individual questions in the IPSS
Gender/
age group
IPSS question, % with scores ≥ 2
Frequency
of urinating
Stopping
and starting
Holding
back
Lower strength
+ force of stream
Straining to
begin urination
Frequency
of nocturia
Sensation
of voiding
Men
40–49  6.412.7  6.1  6.0  8.4  4.814.4
50–59  9.720.210.910.117.3  6.716.7
60–6916.124.515.317.227.710.228.3
70–7920.636.120.224.036.011.247.9
Women
< 49  9.823.1  7.511.9  8.5  5.215.9
50–5910.523.4  9.615.111.4  4.421.4
60–6910.923.811.019.513.9  5.728.7
70–7914.926.915.324.822.6  6.943.8

The percentage of men scoring of ≥ 2 from 5 on each question in the IPSS increased by 5–9% per age decade for all questions except for ‘straining to begin urination’ (2% per decade) and nocturia (11% per decade; Table 2). Straining was the least reported symptom in all age groups in both genders, among whom there was no increase with age. The rate of increase in the prevalence of nocturia was 9% per decade among women while in the other five questions the rate of increase was 1–4%. Among those aged 40–49 years the differences between men and women related to the IPSS questions about holding back urine, and frequency of urination during the day. Among the older groups, men reported more symptoms than women for all questions except those describing nocturia and difficulty in holding back urine.

DISCUSSION

The prevalence of moderate to severe LUTS was notably high in men and women and showed no marked cultural variation. We estimated the same prevalence among Korean men as European men [17,18]. Furthermore, the prevalence of both moderate and severe LUTS increased with age such that 40% of men and 28% of married women aged 70–79 years reported moderate or severe symptoms, while 7.4% of men and 5.7% of women reported severe symptoms. The prevalence rates from this study are comparable with those reported for men elsewhere [12,18].

The very low response rate in Auxerre is a potential problem for this survey. Data were presented separately for each centre to check on the consistency of the responses; this was confirmed and suggests that there was no serious bias in Auxerre. In the other centres the response rates were as least as high as those reported in other studies [10,11]. Furthermore, our investigation of non-response bias [14] suggested that no response was related to age but not to lower values of the IPSS [11], suggesting that our estimates of prevalence are likely to be valid.

The prevalence of LUTS was associated with high physical activity at work and possibly with low current physical activity in men, but not in women. Current physical activity has also been indicated as protective for LUTS [19]. Alcohol consumption and smoking status [20,21] have also been cited as associated with BPH, although no thorough epidemiological assessment has yet been made. We did not measure alcohol consumption in the present study and found no evidence of an increased risk of more symptoms associated with current smoking status [12,22].

It is important to be aware of confounding effects which were not controlled for; one of these may be the consumption of large amounts of liquid, e.g. beer, which in the UK and the Netherlands is more common among manual labourers. Also, the negative association between prevalence and current physical activity during leisure time may be a function of some confounding by indication. Men with severe symptoms may not take part in some physical activities because the symptoms prevent them from doing so.

A few recent studies have reported the use of the IPSS among women [23–25]. Although none of these studies was as large as the present, and nor were they population-based, the present study only refers to currently married women. One of the reasons for using the IPSS in women was not to prove that it is not specific for BPH, but because there are few validated outcome measures for both men and women [23,25]. The need to have a uniform and reliable method to evaluate symptoms in women was also noted in the development of a symptom index and a problem index for interstitial cystitis [26].

Although the IPSS is used to estimate LUTS, high values are often taken as an indication of BPH. However, the diagnosis of BPH requires a urological examination and should not be based solely on the IPSS. This means that the IPSS cannot be used to infer BPH in epidemiological studies. In connection with the lack of specificity of the AUA-SI for BPH, Roberts et al.[27] argued that the similarity in the distribution of the index in men and women is partly a result of the potential confounding effect of continence status. They suggested that urinary incontinence may contribute to a high AUA-SI (IPSS) in both sexes. This is certainly borne out in the present study, as the question concerned with frequency identified it as the most common symptom reported by young women.

Although the present study shows similar distributions of IPSS in men and women, and similar contributions to the total score from each of the seven questions, the underlying mechanisms for the urological symptoms may differ. There was no urological examination for all subjects and so we cannot provided detailed information on sensitivity and specificity of the IPSS for BPH in men.

The close agreement of the severity of LUTS in men and women is notable. Moreover, women reported the symptoms traditionally associated with LUTS in men, showing that the IPSS is not specific for gender. The similarity of the contribution of the same individual symptom questions to the total IPSS in men and women was also notable and surprising. There are two important messages which emerge. First, the lack of specificity of the IPSS for BPH is obvious. Clearly there is no role for the IPSS in the diagnosis of BPH, as was always the case, although it remains of value in assessing the response of symptoms to treatments. Second, and most clearly, the IPSS can be used in women and men to give an indication of their urological well-being. We do not imply that the IPSS has clinical significance for women and it is possible that there are more important urological questions for women than those in the IPSS. However, it measures symptom severity and can be of value, as in men, for assessing the response of symptoms to treatment.

ACKNOWLEDGEMENTS

This contribution of P. Boyle, C. Robertson and C. 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.

Abbreviations
QoL

quality of life

AUA-SI

AUA Symptom Index

OR

odds ratio.

Ancillary