1. Design and study participants

We conducted a secondary analysis of the data from the Ansan and Ansung cohort of the Korean Genome Epidemiology Study (KoGES) conducted by the Korea Disease Control and Prevention Agency (KDCA). The KoGES was a community-based prospective cohort study, and details of KoGES have previously been described [28]. The Ansan and Ansung cohort, representing small/medium-sized cities and rural areas in Korea, comprised 10,030 adult men and women aged between 40 and 69 years selected through two-stage cluster sampling based on information from the governing district in the telephone directory and information on demographic characteristics from the 2000 Census. The baseline survey of the cohort was completed in 2001 and 2002, and follow-up surveys are ongoing every other year. The cohort consists of a health examination and health interview of demographic information, medical history, family history of diseases, and health behaviors (including smoking, alcohol drinking, physical activity, and dietary intake).

Considering that obesity status changes over time and UTI can recur, we investigated the relationship between obesity and UTI occurrence during a possible short-term time course. As the most recent data available among the Ansan and Ansung cohort of KoGES was the 6th follow-up data, we used the 5th follow-up data as the baseline and identified the UTI incidence until 6th follow-up during a 2-year follow-up period. The 5th follow-up survey was conducted in 2011 and 2012 with 6,238 people, and the 6th was completed in 2013 and 2014 with 5,906 people [29].

The inclusion criteria were participants who had no UTIs and no missing values in height or weight in the baseline (the 5th follow-up survey), no follow-up loss, no missing values in UTI incidence data during the follow-up period, and BMI ≥ 18.5 kg/m². People with BMI < 18.5 kg/m² were excluded due to small sample size (n = 88). Among the 6,238 people participated in the 5th follow-up survey, 1,312 who did not meet the inclusion criteria were excluded and the remaining 4,926 (response rate, 79.0%) were included in the analysis. Finally, UTI occurred in 251 (5.1%) during the two-year follow-up period (Figure 1).

Figure 1
Flowchart of study subjects' selection.
KoGES = Korean Genome Epidemiology Study; UTI = Urinary tract infection.

• Download Figure

2. Study variables

The variables included in this study were obesity (independent variable), UTI (outcome variable), and general characteristics (covariates).

3. Data analysis

Data were analyzed using the IBM SPSS Statistics for Windows ver. 23.0 (IBM Corp., Armonk, NY, USA). The significance level of the two-tailed statistical test was .05. The participants' baseline characteristics were analyzed with frequency and percentage for categorical data and mean and standard deviation for continuous data. The differences in the BMI groups were compared using the chi-squared test, analysis of variance, and Tukey's post hoc test. The incidence of UTI was calculated using incidence proportion and incidence density, which was calculated by dividing the number of UTIs by the total number of participants or per 1,000 person-years, respectively. We calculated individual person-years by obtaining elapsed days from the 5th to the 6th follow-up dates, and by converting the days into years. Individual person-years was added together to obtain total person-years. The probability of remaining UTI free was determined by the Kaplan-Meier method, and the BMI groups were compared using the log-rank test. The incidence density of UTI among BMI groups was compared by simple Cox proportional hazards regression analysis. A multivariate Cox proportional hazards regression analysis was performed to identify the association of obesity and UTI incidence after controlling general characteristics in the baseline (the 5th follow-up survey). Subgroup and interaction analyses were performed to identify the relationship between BMI groups and UTI incidence according to gender and age groups (aged less than 60 years vs. aged older than and equal to 60 years).

4. Ethical considerations

This study was conducted after receiving approval of exempt review from the Institutional Review Board (IRB) of Pusan National University Yangsan Hospital (IRB No. 05-2018-072). Our data came from the department in charge of KoGES in KDCA after submitting the IRB approval letter and study proposal to the Department, and all data were provided in anonymized form.

1. Baseline characteristics of the participants according to the BMI group

Among the 4,926 participants, for obese and obese class II+ people were 42.9% (n = 2,115) and 5.0% (n = 245), respectively. For total participants, the mean age was 60.9 years old, ranging from 48.0 to 80.0 years old, 13.2% were current smokers, and 12.9% had a history of diabetes. For obese and obese class II+ people, the mean age was 60.5 and 60.6 years old, current smokers were 11.5% and 6.9%, and with a history of diabetes were 15.1% and 16.3%, respectively. Obese people were younger, less likely to currently be smokers and had higher history of diabetes than the normal weight group. People in obese class II+ was higher in women and those living in a rural area, and lower in those living with a spouse, currently smokers, and currently drinkers than in the normal weight group (Table 1).

Table 1
Baseline Characteristics of Subjects according to the Body Mass Index Group

• Download Table

2. Incidence of urinary tract infection according to the BMI group

UTI incidence proportion was 4.8%, 5.4%, and 10.6%, and the incidence per 1,000 person-years was 24.0, 27.3, and 53.3 in the normal weight group, obese and obese class II+ people, respectively. UTI incidence in obese people was not significantly different from the normal weight group. People aged ≥ 60 years, obese people were 1.72 times more likely to have UTI than those with normal weight, but did not show an increase in risk after controlling general characteristics.

Contrary to obese people, UTI incidence was significantly higher in obese class II+ people than those with normal weight (p < .01). For the unadjusted hazard ratio (HR) of UTI incidence using normal weight as the reference group, people in obese class II+ were 2.16 times more likely to have UTI (p < .01). Kaplan-Meier survival analysis also revealed a significant relationship between BMI groups and UTI incidence, where people in obese class II+ had the lowest UTI free survival rate (log-rank p = .002). After controlling general characteristics, people in obese class II+ remained 1.66 times (HR = 1.66, 95% confidence interval [CI] = 1.06~2.61; p < .05) more likely to have UTI than those with normal weight (Figure 2). This trend was similar in men and people aged equal to and older than 60 years (aged ≥ 60 years). Men in obese class II+ were 4.37 times (HR = 4.37, 95% CI = 1.11~17.15; p < .05) more likely to have UTI than those with normal weight. For people aged ≥ 60 years or in obese class II+ increased the risk of developing UTI by 2.00 times (HR = 2.00, 95% CI = 1.06~3.78; p < .05) more than those with normal weight (Table 2).

Figure 2
Survival curves for urinary tract infection according to the body mass index (BMI) group after controlling the general characteristics. (A) Four groups of BMI; (B) three groups of BMI. ^†The reference of the hazard ratio was BMI of 18.5~22.9 kg/m².
CI = Confidence interval.

• Download Figure

Table 2
Incidence and Hazard Ratio of Urinary Tract Infection according to the Body Mass Index Group

• Download Table

Table 3 showed the results of UTI risk according to BMI group by a combination of specific gender and age groups. UTI risk was 10.08 times and 10.30 times higher among women aged < 60 years and aged ≥ 60 years than men aged < 60 years, respectively, and the results were similar after controlling the general characteristics. Meanwhile, obese people did not show significant differences in UTI risk compared to those with normal weight in any combination of specific gender and age groups. However, women aged ≥ 60 years and in obese class II+ had 1.98 times (HR = 1.98, 95% CI = 1.01~3.86; p < 0.05) higher likelihood of having a UTI than those with normal weight, and the trend was the same after controlling the general characteristics.

Table 3
Hazard Ratio of Urinary Tract Infection according to Body Mass Index Group by Combination of Specific Gender and Age Groups

• Download Table

1. Strengths and limitations of the study

This study has several strengths. First, few studies have attempted to identify the association between obesity and UTI incidence among adults in the community, and to the best of our knowledge, this is the first study in South Korea. Second, the KoGES, the data source for this study, adopted a prospective design to reduce the chance of recall bias and provided good evidence between obesity and the development of UTI. Moreover, it comprised a large sample size, which increased its statistical power. Third, the follow-up period was relatively short, reducing information bias due to changes in baseline characteristics, including BMI, and a large number were lost to follow-up over time. Fourth, as the weight and height data used in BMI calculations were not self-reported, but directly measured using a standardized method by trained surveyors, the accuracy of BMI can be guaranteed.

However, careful interpretation of the results is required due to the following limitations. First, as previously mentioned, UTI was defined by self-report and dipstick test results because urine culture data were not available in the KoGES, which may lead to underestimation of the UTI incidence with some errors. The dipstick decision rule has shown good interrater reliability [37] and better diagnostic precision than the clinical decision rule [30], but the limited negative predictive value still needs to be improved. Second, we could not use some variables associated with UTI development, such as the history of UTI and urinary system diseases, for adjustment. Sexual life is one of the known risk factors of UTI. Subjects were considered sexually active if they have a spouse, which may be different from their actual sexual life. Third, since we used the 5th follow-up data as a baseline, most participants were aged between 50s and 70s, reducing the generalizability of the findings to those aged less than 50 years. Moreover, this age mix may lead to bias in UTI incidence. For example, women in post-menopause are more vulnerable to UTI because the decrease in estrogen leads to change in urogenital epithelium and distribution of Lactobacillus, which facilitates the increase of uropathogens and infection [38]. Therefore, further studies involving people in their 30s and 40s are suggested. Lastly, since the Ansan and Ansung cohort of KoGES is not a nationally representative sample, generalizability of the findings to the entire population in South Korea may be limited.

2. Implications for clinical practice

This study provides evidence that obesity with BMI ≥ 30.0 kg/m² is associated with development of UTI. Particularly, obesity in men over 40 years or women aged ≥ 60 years was found to be more strongly related to the development of UTIs in this study. Men may be less interested in UTI because of lower UTI incidence than women; however, UTI risk is rapidly increasing among men with obese class II+. By understanding the association between obesity and UTI, nurses in community settings should identify these high-risk groups for UTI and utilize the findings in health education for the public. In addition, policymakers should understand the additional benefits of weight loss and make efforts to manage obesity.