Subjects
The present study used data from two large-population surveys: a community-based interview survey and a university hospital outpatient-based interview survey. Both were conducted with male subjects aged 40 years or older who gave and wrote the informed consents to the survey and voluntarily expressed their intention to participate. This study was approved by the Institutional Review Board of Korea University Ansan Hospital. In order to enhance the validity of International Prostate Symptom Score (IPSS), both surveys excluded the following patients: 1) those who had undergone urological surgery which might affect their IPSS score; 2) those who had received any treatment for BPH or prostate cancer; 3) those who had evidence of neurological condition, un-controlled diabetes mellitus, un-controlled hypertension, history of malignancy, urinary tract infection within 3 months, psychiatric illness with medications and alcohol or substance abuse; 4) those who were taking or had taken any drug for the same complaints.
Community-based interview survey
One investigator visited senior welfare centers in South Korea between May 2010 and April 2013 and carried out a survey of 1,030 males using the IPSS questionnaire. During this period, the survey was conducted over a total of 36 times in six metropolitan areas of the country: Seoul, Gyeonggido, Incheon, Daejeon, Daegu, and Busan. In total, the study had a sample size of 518 subjects after excluding those described above.
University hospital outpatients-based survey
Another IPSS questionnaire survey was performed with 2,493 male outpatients who visited university hospitals in South Korea between September 2010 and September 2011. The survey included 20 university hospitals in nine major areas of the country: Seoul, Gyeonggido, Incheon, Daejeon, Daegu, Busan, Gangwondo, Gwangju, and Ulsan. In total, the study had a sample size of 1,278 subjects after excluding those described above.
Study design and measuring tools
This was a cross-sectional study. To evaluate the factors that might affect the severity of LUTS, the study investigated age, S-PP, and IPSS.
IPSS questionnaire
The severity of LUTS was measured by IPSS based on the American Urological Association (AUA) symptom index, with one additional question on quality of life. IPSS questionnaire has been translated into many different world languages and adapted based on the circumstances of each country. IPPS questionnaire is now widely used for objective assessment of LUTS [6,18]. The Korean version of the IPSS verified by Choi et al. in terms of relevance and reliability is now the most typical diagnostic instrument for LUTS in Korea [19].
The IPSS questionnaire consisted of eight items, which included seven 6-point scale questions on symptoms (feeling of incomplete emptying, urinary frequency, interrupted stream, urinary urgency, weak urinary stream, urinary hesitancy, and nocturia) and one 7-point scale question on patient’s satisfaction with their urinary condition. Based on the criteria of Barry et al., symptom severity was divided into three groups: mild (a symptom score of 0–7), moderate (8–19), and severe (20–35) [20]. The quality of life or level of satisfaction of LUTS patients was represented by seven grades: “No problem” (0 point = very satisfied), “I’m all right” (1 point), “Somewhat satisfied” (2 points), “Half-satisfied, half-dissatisfied” (3 points), “Somewhat dissatisfied” (4 points), “Distressed” (5 points), and “I can’t stand it” (6 points = very dissatisfied).
Age questionnaire
Age as an important factor has impact on generation-specific prevalence, IPSS, and S-PP of LUTS. Therefore, this study queried each participant’s date of birth.
S-PP of LUTS
The S-PP of LUTS was defined as the period between the moment the participant perceived any inconvenience resulting from LUTS (feeling of incomplete emptying, interrupted stream, urgency, weak urinary stream, hesitancy, or nocturia) and the time the interview survey was conducted. The longest periods of any LUTS symptoms were regarded as S-PPs.
Reliability
Cronbach’s α was 0.652 for the seven 6-point scale questions about symptoms and the one 7-point scale question on satisfaction with urinary conditions, indicating acceptable internal consistency and reliability. Internal consistency with each item excluded did not substantially change the observed value. The reliability of the questionnaire used in this study was estimated to be similar to, or at least not lower than, that in previous studies [14]. This suggested that we used the same method as previous studies. The interview survey of university hospital outpatients was conducted by well-trained professional investigators. The present survey enrolled hospital patients through a formal procedure in compliance with the guidelines of the individual hospitals’ institutional ethics committees.
Propensity score matching
Propensity score matching (PSM) was used to organize the population in both surveys into randomized groups to reduce selection bias in sampled population. Since the population of this study included two different groups of people, those “who do not visit hospitals” (in the community-based interview survey) and those “who visit hospitals” (in the interview survey of university hospital outpatients), we allowed Berkson’s bias which may result from differences in characteristics between the two groups [21]. PSM typically involves the formation of pairs of treated and untreated subjects with similar propensity score (PS) values. Hence, a logistic regression model was used to calculate and save the predicted probability of the dependent variable and the PS for each observation in the data set. This single score (between 0 and 1) represented the relationship between multiple characteristics and the dependent variable as a single characteristic. Age and S-PP, whose significance was established in previous studies [12-14], were used as independent variables in the analysis.
In this study, in-caliper nearest-neighbor matching proposed by Rosenbaum and Rubin was taken into consideration as a PSM method. Rosenbaum and Rubin suggest use caliper value equal to 0.25 of the standard deviation of the logit of the PS [22]. Accordingly, this study used a caliper of 0.25 times the standard deviation of the PS. Furthermore, one-to-one matching was performed in order to optimize possible effects through several simulations. Excluding those with missing values, 483 subjects were selected from each group using PSM. In this study, the multivariate imbalance measure decreased from 0.50 before PSM to 0.28 after PSM. The implementation of PSM was accordingly evaluated to be appropriate.
Statistical analysis
In order to examine the association between BPH and related risk factors, we analyzed distribution patterns before and after performing PSM. We performed t-test on individual variables to determine whether the confounder was properly controlled between the non-visiting group (for the community-based interview survey) and the visiting group (for the interview survey of university hospital outpatients). For the prevalence of BPH, a frequency analysis was performed on IPSS scores. A one-way ANOVA was carried out to examine the relationships between different BPH severity groups. In an attempt to examine the correlation between the IPSS and the S-PP of LUTS, partial correlation coefficient was measured while controlling for age. A multiple linear regression analysis was conducted to assess how IPSS had changed over a year with respect to risk factors for BPH. In the analysis, independent variables included age and the S-PP of LUTS, both reported to be significant in previous studies [12-14]. The IPSS was used as a dependent variable. The significance of multicollinearity was assessed by comparing the variation inflation factors (VIFs) between independent variables.
All data were presented as mean and standard deviation (SD). Statistical analysis was performed using SPSS version 21.0 software (IBM, New York, NY, USA) with an R module available for PS analysis. All statistics were two-tailed and p-values <0.05 were considered to be significant.