The risk of urinary tract infection in vegetarians and non-vegetarians: a prospective study

Urinary tract infection (UTI) is caused principally by ascending Escherichia coli infection via an intestine-stool-urethra route. Recent studies found that the strains of E. coli causing UTIs, called extra-intestinal pathogenic E. coli (ExPEC), were distinct from the intestinal pathogenic strains and normal commensal strains. Further analysis found the meat including poultry and pork is the major reservoir for ExPECs. Vegetarians avoid meat and should theoretically have less exposure to ExPEC. However, no study thus far has examined whether vegetarian diets reduce the risk of UTI. Our aim was to examine the association between vegetarian diet and UTI risk in a Taiwanese Buddhist population. We prospectively followed 9724 Buddhists free of UTI from 2005 to 2014. During the 10-year follow-up, 661 incident UTI cases were confirmed. Diet was assessed through a food frequency questionnaire. Cox regression was used to evaluate the prospective association between a vegetarian diet on risk of UTI while adjusting for age, sex, educational level, alcohol-drinking, smoking, hypertension, diabetes mellitus, hyperlipidemia, and disease conditions predisposing to UTIs. Overall, vegetarian diet was associated with 16% lower hazards (hazard ratio [HR]: 0.84, 95% confidence interval [CI]: 0.71–0.99). In subgroup analysis, the protective association between vegetarian diet and UTI is observed mainly in the female (HR: 0.82, 95% CI: 0.69–0.99), never smokers (HR: 0.80, 95% CI: 0.67–0.95), and for uncomplicated UTI (HR: 0.81, 95% CI: 0.68–0.98).

Assessment of disease status. We followed the TCVS participants from 2005 to 2014 by linking the baseline data to the National Health Insurance Research Database (NHIRD) and the death records at the National Health and Welfare Data Science Center (HWDC), Ministry of Health and Welfare.
In Taiwan, the National Health Insurance Program is a governmental universal health insurance program covering nearly 100% of the population. Individuals' medical claim data including diagnosis from both inpatient and outpatient are available. In order to protect the privacy of individuals as required by local law, personal identification number is masked and all analyses must be performed within the HWDC. Only summarized research results (not individual data), could be released.
Dietary assessment. The diet section in the questionnaire was adopted from the food frequency questionnaire (FFQ) used in Nutrition and Health Survey in Taiwan, and included 57 items on food and food groups, and a section on cooking methods and cooking oil. A similar questionnaire had been validated later and showed good reliability and validity among Tzu Chi volunteers 9 .
Participants were categorized as vegetarians if they (1) self-identified as vegetarians in a question asking vegetarian status, and (2) reported "no eating" in frequency questions for all individual meat and fish items in the FFQ. Those who self-identified as vegetarians but reported eating meat or fish in the FFQ were categorized as non-vegetarians. exclusion criteria. Individuals with age < 20 years old (n = 85), no available data on NHIRD (n = 677), incomplete data on the questionnaire (n = 950), and prior history of UTI (n = 626) were excluded from this study (Fig. 1). case ascertainment. Incident cases of UTI were identified by the International Classification of Diseases, Ninth Revision (ICD-9) for UTIs and the codes for the execution of bacterial culture and prescription of antibiotics against UTI within the HWDSC database during follow-up.
The ICD-9 codes used include 590 for infections of the kidney, 595 for cystitis, and 599.0 for UTI, site not specified. The code for the execution of bacterial culture is 13007C. We identified UTI cases by the above ICD-9 diagnosis in either inpatient or outpatient, accompanied by the execution of bacterial culture and prescription of antibiotic against UTI 10,11 (Fig. 1).

Statistical analysis. Baseline characteristics of vegetarians and non-vegetarians were compared using
Chi-square (χ 2 ) test for categorical variables and independent sample t test for continuous variables.
Cox regression was used to analyze the association between different factors and risks of UTI, presented as hazard ratio (HR) with 95% confidence interval (CI). Model 1 adjusted for age and/or sex. Model 2 additionally adjusted for educational level and lifestyle factors (smoking and alcohol-drinking). Model 3 additionally adjusted for hypertension, diabetes mellitus, and hyperlipidemia. Model 4 additionally adjusted for disease conditions that predispose to complicated UTIs, including urinary tract obstruction caused by hyperplasia or tumor, urolithiasis (calculus), urine retention, renal failure, immune dysfunction, pregnancy, and urinary catheter indwelling. Kaplan-Meier curves of crude data were plotted to show the disease-free survival throughout follow-up informed consent. Informed consent was obtained from all individual participants included in the study.

Results
Baseline characteristics. Table 2 shows the baseline characteristics of both vegetarian and non-vegetarian groups. The ratio of non-vegetarians to vegetarians was about 2:1. The vegetarian group was older, had a higher proportion of female, received less education, was less likely to have a history of smoking or alcohol-drinking, and had a significantly lower proportion of individuals with hypertension, diabetes, hyperlipidemia, and disease conditions that predispose to UTI.

Risk factors associated with Uti.
In the 10 years of follow-up, 661 individuals were diagnosed with UTIs. Table 3 shows the Cox regression analysis. Overall, UTI risk in vegetarians was significantly lower than non-vegetarians (HR = 0.84, 95% CI: 0.71-0.99, p = 0.038). Age, female sex, diabetes, urine retention, and renal failure were associated with significant increases in UTI risks. Hypertension, urinary obstruction (hyperplasia/ tumor of the urinary tract), urolithiasis, and immune dysfunction were associated with non-significant higher UTI risks.
Subgroup analysis. Subgroup analysis of the association between vegetarian diet and UTI by sex in various models were shown in Table 4. Vegetarian diet was significantly associated with a reduced risk of UTI mainly in female, particularly after adjustment for education and lifestyle (smoking and alcohol-drinking) in Model 2.
Additional adjustments for hypertension, diabetes, hyperlipidemia, and conditions predisposing to UTI (as in Model 3 and 4) yielded similar results. Kaplan-Meier curves of UTI-free survival over the ten-year period for vegetarians and non-vegetarians ( Fig. 2) highlight the protection of vegetarian diet in female, while no distinct difference was found in male. Additional subgroup analyses and the test for interactions are presented in Table 5. The test of interaction between dietary patterns and smoking was significant (P-interaction = 0.029), with the protection mainly among www.nature.com/scientificreports www.nature.com/scientificreports/ never smokers. No significant test of interaction was found between diet and other factors (sex, educational level, alcohol-drinking, hypertension, diabetes, hyperlipidemia, and disease conditions predisposing to UTI).
The association between vegetarian diet and subtypes of UTI is shown in Table 6. Vegetarian diet was associated with significantly reduced risk of mainly the uncomplicated UTI but not complicated UTI.

Discussion
In this prospective cohort study, vegetarian diet (versus non-vegetarian diet) is associated with a lower risk of UTI particularly in female, non-smokers, and mainly for the uncomplicated-subtype of UTI. To the best of our knowledge, this is the first study that evaluated the prospective association between a vegetarian diet and UTI.
Previous research suggested that meat-related foodborne ExPECs may be the main cause of uncomplicated UTIs and a half of complicated UTIs (Table 1) [2][3][4][5][6][7][8] . Consistent with this finding, we found that the protective association between vegetarian diet and UTI risk was mainly in uncomplicated rather than complicated UTI, and this association is independent of diseases and risk factors predisposing conditions to complicated Non-vegetarians Vegetarians P-value mean (std a )/n (%) mean (std a )/n (%)   www.nature.com/scientificreports www.nature.com/scientificreports/ UTI including hypertension 12 , diabetes 13 , and hyperlipidemia 14 . This suggests that the effect of vegetarian diet is not mediated through these related diseases. The test of interaction showed that vegetarian diet (versus non-vegetarian diet) was associated with protection among the never smoking group, but not the ever smoking group (P-interaction = 0.029, Table 5). It is possible that smoking compensates the benefit of vegetarian diet by increasing blood pressures (which predisposing to renal failure) 12,15 , renal stone formation 16 , and urinary tract cancers 17 . All of which are also the disease conditions predisposing to complicated UTI 2-4 . However, the number of smokers in our study is small and the significant interaction could arise from chance that further confirmation from other studies is needed.
The principal pathogenic strains causing UTIs are ExPECs (accounting for 65~75% UTIs), a group of strains distinct from the intestinal pathogenic and normal commensal E. coli 3,4 . ExPECs express many characteristic virulent factors which are not present in other strains of non-ExPECs [2][3][4]18 . Many studies have illustrated that meat, including poultry and pork, as the major reservoirs for ExPECs [5][6][7][8]19 . Vincent et al. identified the E. coli strains by using O:H serotype and sequence type (ST) and discovered the O25:H4-ST131 and O114:H4-ST117 strains acquired from retail chicken were very similar to, if not indistinguishable from E. coli of UTI patients 5 . Jacobsen et al. performed the phylogenetic analysis by triplex polymerase chain reaction using three DNA markers showed the UTI-associated B2 and D phylogroups are found among isolates from broiler chicken meat, broiler chickens, Vegetarians (n = 3257) Non-vegetarians (n = 6467)  pork, and pigs. Further, antimicrobial resistance phenotypes of E. coli from meat were very similar to isolates from UTI patients 7 . In addition, sequencing analysis of ExPECs showed UTI-causing ST10, ST69, ST95, ST117, and ST131 are associated with poultry and retail poultry meat 8,19 . Vegetarians do not consume meat and therefore avoid uptake of animal-food-associated ExPECs.   www.nature.com/scientificreports www.nature.com/scientificreports/ Recent studies discovered the gut and fecal microbiotas were different between vegetarians and non-vegetarians [20][21][22][23] . The high fiber contents in vegetarian diet may modulate the intestinal microbiota in humans 20,21 . Fiber is metabolized by intestinal anaerobic microflora to produce short-chain fatty acids, which decrease intestinal pH 20,21 . Vegans or vegetarians were shown to have lower total counts of E. coli and Enterobacteriaceae spp. and lower stool pH than non-vegetarians, and the lower stool pH could prevent the growth of E. coli and Enterobacteriaceae 22 . The shifts in the gut and fecal microbiotas in vegetarians may potentially contribute to the protection from UTI.
ExPECs isolated from animal food products and UTI patients are very similar in antibiotic resistant patterns and virulence factor profiles 5,6,8,44,45 . Antibiotics use in animal agriculture may contribute to the increasing antibiotic resistance in humans 44 . The extended spectrum β-lactamase (ESBL)-producing and fluoroquinolone-resistant ExPECs, most of which are globally disseminated E. coli strain O25:H4-ST131 (accounting for 78%), are a major problem with resistance to penicillins, cephalosporins, and fluoroquinolones 2,8,46 . Moreover, the ESBLs encoded on plasmids typically carry other antibiotic resistance genes against aminoglycosides, sulfonamides, and quinolones, leading to multidrug resistant 2 . The high recurrence rates and increasing multidrug resistance in ExPECs 2 make antibiotic treatment of UTI challenging. Therefore, alternative non-antibiotic management for UTI is important. High phytochemical content in a healthy vegetarian diet may provide an alternative prophylaxis from and bactericidal effect against UTI 27,[30][31][32][33][34][35][36][37][38][39][40] , and a study had demonstrated the PAC from cranberry also exhibited the anti-adherence property in the multidrug resistant strains of uropathogenic P-fimbriated E. coli 47 .
Several limitations must be noted in our study.
(1) Our case definition is based on ICD-9 code rather than the gold standard, which should be clinical symptoms and laboratory examination including pyuria and urine culture with a more than 100,000 colony forming units per milliliter 48,49 . (2) Residual confounding may still remain as several known confounders, including water intake 50 , sexual intercourse 33,51 , intake of honeydew melon (which may also contain ExPEC 5 , though evidence is insufficient), were not measured. (3) Although ExPEC is found mainly from meat sources, one study reported the detection of ExPEC isolates from a honeydew melon sample of restaurant/ready-to-eat foods in North America 5 . To the best of our knowledge, E. Coli or ExPEC had never been reported in plant based foods in Taiwan. The sporadic case of ExPEC in non-meat product may be due to contamination and will require further research. However, if plant-based foods were a major source of ExPEC contributing to UTI, the protective association between vegetarian diet and UTI would have been attenuated, strengthening the null association. (4) Another limitation lies in our FFQ, which did not ask specifically about intake of honeydew melon, cranberries, or roselle. Thus, we were unable to conduct further analysis on the association between UTI and these plant foods. In addition, the FFQ was self-administered with some degree of missing values, that it is difficult to calculate the exact intake of specific phytochemicals known to contribute to UTI-protection.