Recurrent urinary tract infection (RUTI) can damage renal function and has impact on healthcare costs and patients’ quality of life. There were 2 stages for development of prediction models for RUTI. The first stage was a scenario in the clinical visit. The second stage was a scenario after hospitalization for urinary tract infection caused by Escherichia coli. Three machine learning models, logistic regression (LR), decision tree (DT), and random forest (RF) were built for the RUTI prediction. The RF model had higher prediction accuracy than LR and DT (0.700, 0.604, and 0.654 in stage 1, respectively; 0.709, 0.604, and 0.635 in stage 2, respectively). The decision rules constructed by the DT model could provide high classification accuracy (up to 0.92 in stage 1 and 0.94 in stage 2) in certain subgroup patients in different scenarios. In conclusion, this study provided validated machine learning models and RF could provide a better accuracy in predicting the development of single uropathogen (E. coli) RUTI. Both host and bacterial characteristics made important contribution to the development of RUTI in the prediction models in the 2 clinical scenarios, respectively. Based on the results, physicians could take action to prevent the development of RUTI.

Chronic kidney disease (CKD) is highly prevalent in Taiwan. More than two-thirds of end-stage renal disease is associated with diabetes mellitus (DM) or hypertension (HTN). Therefore, the formulation of a special preventative policy of CKD in these patients is essential. This study surveyed 14 traditional risk factors and identified their effects on CKD in patients with HTN/DM and compared these with their effects in the general population. This study included 5328 cases and 5135 controls in the CKD/HTN/DM outpatient and health centres of 10 hospitals from 2008 to 2010. Fourteen common effect factors were surveyed (four demographic, five disease and five lifestyle), and their effects on CKD were tested. Significance tests were adjusted by the Bonferroni method. Results of the stratified analyses in the variables were presented with significant heterogeneity between patients with different comorbidities. Male, ageing, low income, hyperuricemia and lack of exercise habits were risk factors for CKD, and their effects in people with different comorbidities were identical. Anaemia was a risk factor, and there was an additive effect between anaemia and HTN on CKD. Patients with anaemia had a higher risk when associated with HTN [odds ratio (OR) = 6.75, 95 % confidence limit (95 % CI) 4.76–9.68] but had a smaller effect in people without HTN (OR 2.83, 95 % CI 2.16–3.67). The association between hyperlipidaemia-related factors and CKD was also moderated by HTN. It was a significant risk factor in people without HTN (OR = 1.67, 95 % CI 1.38–2.01) but not in patients with HTN (OR =1.03, 95 % CI 0.89–1.19). Hepatitis B, hepatitis C, betel nut chewing, smoking, alcohol intake and groundwater use were not associated with CKD in multivariate analysis. We considered that patients with HTN and anaemia were a high CKD risk population. Physicians with anaemic patients in outpatient clinics need to recognise that patients who also have HTN might be latent CKD cases.

Rationale and Objective: Gut microbiota have been targeted by alternative therapies for non-communicable diseases. We examined the gut microbiota of a healthy Taiwanese population, identified various bacterial drivers in different demographics, and compared them with dialysis patients to associate kidney disease progression with changes in gut microbiota.Study Design: This was a cross-sectional cohort study.Settings and Participants: Fecal samples were obtained from 119 healthy Taiwanese volunteers, and 16S rRNA sequencing was done on the V3-V4 regions to identify the bacterial enterotypes. Twenty-six samples from the above cohort were compared with fecal samples from 22 peritoneal dialysis and 16 hemodialysis patients to identify species-level bacterial biomarkers in the dysbiotic gut of chronic kidney disease (CKD) patients.Results: Specific bacterial species were identified pertaining to different demographics such as gender, age, BMI, physical activity, and sleeping habits. Dialysis patients had a significant difference in gut microbiome composition compared to healthy controls. The most abundant genus identified in CKD patients was Bacteroides, and at the species level hemodialysis patients showed significant abundance in B. ovatus, B. caccae, B. uniformis, and peritoneal dialysis patients showed higher abundance in Blautia producta (p ≤ 0.05) than the control group. Pathways pertaining to the production of uremic toxins were enriched in CKD patients. The abundance of the bacterial species depended on the type of dialysis treatment.Conclusion: This study characterizes the healthy gut microbiome of a Taiwanese population in terms of various demographics. In a case-control examination, the results showed the alteration in gut microbiota in CKD patients corresponding to different dialysis treatments. Also, this study identified the bacterial species abundant in CKD patients and their possible role in complicating the patients’ condition.

Sclerostin, an antagonist of the Wingless-type mouse mammary tumor virus integration site (Wnt) pathway that regulates bone metabolism, is a potential contributor of chronic kidney disease (CKD)–mineral and bone disorder (MBD), which has various forms of presentation, from osteoporosis to vascular calcification. The positive association of sclerostin with bone mineral density (BMD) has been demonstrated in CKD and hemodialysis (HD) patients but not in peritoneal dialysis (PD) patients. This study assessed the association between sclerostin and BMD in PD patients. Eighty-nine PD patients were enrolled; their sera were collected for measurement of sclerostin and other CKD–MBD-related markers. BMD was also assessed simultaneously. We examined the relationship between sclerostin and each parameter through Spearman correlation analysis and by comparing group data between patients with above- and below-median sclerostin levels. Univariate and multiple logistic regression models were employed to define the most predictive of sclerostin levels in the above-median category. Bivariate analysis revealed that sclerostin was correlated with spine BMD (r = 0.271, P = 0.011), spine BMD T-score (r = 0.274, P = 0.010), spine BMD Z-score (r = 0.237, P = 0.027), and intact parathyroid hormone (PTH; r = − 0.357, P < 0.001) after adjustments for age and sex. High BMD, old age, male sex, increased weight and height, diabetes, and high osteocalcin and uric acid levels were observed in patients with high serum sclerostin levels and an inverse relation was noticed between PTH and sclerostin. Univariate logistic regression analysis demonstrated that BMD is positively correlated with above-median sclerostin levels (odds ratio [OR] = 65.61, P = 0.002); the correlation was retained even after multivariate adjustment (OR = 121.5, P = 0.007). For the first time, this study demonstrated a positive association between serum sclerostin levels and BMD in the PD population.