https://orcid.org/0000-0002-5824-2040
Figures
Abstract
Purpose
Prostatitis has been reported in patients with type 2 diabetes mellitus (T2DM) receiving antihyperglycemic agents (AHAs). This study was conducted to evaluate the risk of prostatitis with canagliflozin in response to a specific Health Authority query.
Methods
This retrospective cohort study used data from adult male patients with T2DM who were new users of canagliflozin (target) or comparators (empagliflozin, dapagliflozin, sitagliptin, and liraglutide). Data were obtained from 8 global administrative claims databases, transformed to a Common Data Model for consistent analysis across databases. Pairwise comparisons were conducted using propensity scores to match canagliflozin users to users of each comparator at a 1:n ratio (maximum n = 100). Hazard ratios were estimated using a Cox proportional hazards model, conditioned on the matched set.
Results
A total of 388,893 adult male patients with T2DM received canagliflozin across databases (mean age, 51.2–71.7 years) and were matched to 657,134 patients receiving empagliflozin, 340,539 receiving dapagliflozin, 819,047 receiving sitagliptin, and 278,684 receiving liraglutide. On-treatment incidence rates showed that prostatitis was uncommon in the canagliflozin cohort in nearly all databases (4.2–7.7 per 1000 person-years) and were similar to those of the comparator treatments. The exception was a higher crude incidence rate in the Merative MarketScan® Medicare Supplemental Database (10.1–12.1 per 1000 person-years). Propensity score matching achieved good balance in all available covariates, and effect estimates were relatively close to a hazard ratio of 1.0, varying on both sides of the null effect. Minimum detectable relative risks were low in most databases, and meta-analytic estimates were near 1.0, with all upper bounds <1.50. No association (either increased or decreased risk) was found with canagliflozin versus other AHAs.
Citation: Yuan Z, Jeffcoat CH, Ali SR, Sena AG, Schuemie MJ, Ryan PB, et al. (2026) Risk of prostatitis in patients with type 2 diabetes mellitus: An observational retrospective cohort study of canagliflozin versus other antihyperglycemic agents using propensity score matching. PLoS One 21(2): e0341745. https://doi.org/10.1371/journal.pone.0341745
Editor: Marwan Salih Al-Nimer, University of Diyala College of Medicine, IRAQ
Received: January 31, 2025; Accepted: January 9, 2026; Published: February 2, 2026
Copyright: © 2026 Yuan et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: The datasets generated and/or analyzed during the current study are not publicly available as they were made available by each company and used under license for the current study. To obtain access to the datasets, researchers can contact the data vendors directly: IQVIA: Anthony Macahilig (Associate Director, Product Management; commercial owner of IQVIA Ambulatory EMR-US and IQVIA PharMetrics® Plus; anthony.macahilig@iqvia.com); Optum: Bree Tremain (Director, Optum Life Sciences Data Support; point of contact for licensed data inquiries; bree.tremain@optum.com); Merative: Luke Boulanger (Director, Technology Product Management; Merative Product Management contact for research; lboulang@merative.com); and JMDC: Marimo Mori (Consultant, Account Service Group; main contact for Johnson & Johnson and initial point of contact for data inquiries; marimo.mori.zuw@jmdc.co.jp).
Funding: This study was funded by Janssen Research & Development, LLC, a Johnson & Johnson Company. The study sponsor/funder was involved in the design of the study; the collection, analysis, and interpretation of data; writing the report; and decisions regarding the publication of the report. Article processing charges were funded by Janssen Research & Development, LLC, a Johnson & Johnson Company.
Competing interests: This study was funded by Janssen Research & Development, LLC, a Johnson & Johnson Company, who is the Marketing Authorization Holder of INVOKANA® (canagliflozin), an SGLT2i evaluated in the current analysis. The study sponsor/funder was involved in the design of the study; the collection, analysis, and interpretation of data; writing the report; and decisions regarding the publication of the report. No other individuals or parties had any influence on the reporting of the study findings of this manuscript. Article processing charges were funded by Janssen Research & Development, LLC, a Johnson & Johnson Company. Zhong Yuan, Carolyn H. Jeffcoat, Saberi Rana Ali, Anthony G. Sena, Martijn J. Schuemie, Patrick B. Ryan, and Sergio A. Fonseca are employees of Johnson & Johnson, and may own stock or stock options in Johnson & Johnson. This does not alter our adherence to PLOS ONE policies on sharing data and materials.
Introduction
Infections, particularly genitourinary infections, such as urinary tract infections (UTIs; e.g., cystitis, pyelonephritis) and genital tract infections (e.g., prostatitis), are common in patients with type 2 diabetes mellitus [1–4]. The predisposition may be the result of several factors, including glucosuria (i.e., an increased presence of glucose in urine), which can promote bacterial growth, an increased adherence of bacteria to uroepithelial cells among patients with suboptimal glycemic control, and immune dysfunction associated with hyperglycemia [2–5]. Incomplete bladder emptying may also increase the risk of genitourinary tract infections in men; in the setting of diabetes, autonomic neuropathy can be present and asymptomatic in a substantial proportion of patients [6]. The UTIs associated with sodium-glucose cotransporter 2 inhibitors (SGLT2is) are typically mild to moderate in intensity and respond to standard treatment. Severe events or events associated with hospital admission were reported in 0.1% to 0.4% of patients in clinical trials; kidney infections and sepsis were “rare” [1]. Men with type 2 diabetes mellitus are at increased risk of developing prostatitis compared with nondiabetic men, which can result in pain, fever, dysuria, urinary retention, and incontinence [7,8].
The SGLT2is (e.g., canagliflozin, empagliflozin, dapagliflozin) are a class of oral antihyperglycemic agents (AHAs) that are commonly prescribed to improve glycemic control in patients with type 2 diabetes mellitus [9,10]. As a result of the mechanism of action of SGLT2is, there is increased urinary glucose excretion which may increase the risk of genitourinary infections [2,3]. Although genitourinary infections were commonly reported as adverse events in SGLT2i clinical trials, several case studies have reported instances of more serious genitourinary infections, such as prostatitis, urosepsis, and pyelonephritis, in patients with type 2 diabetes mellitus who were treated with SGLT2is; however, causality has not been comprehensively evaluated and this association is hypothesized to be secondary in nature [1,11–13]. There is mixed support from meta-analyses regarding the association between SGLT2is and the risk of genitourinary infections [13–15], and data are lacking on the risk of prostatitis (separate from UTIs or genitourinary infections as a whole), as well as the specific etiologies, particularly in relation to the use of SGLT2is.
The current study was conducted to evaluate the risk of prostatitis with canagliflozin in response to a Health Authority query. The study aimed to evaluate the incidence of prostatitis in adult male patients with type 2 diabetes mellitus receiving canagliflozin versus other AHAs using real-world clinical data.
Methods
Study design and patient population
This retrospective, observational, comparative, new-user cohort study evaluated data from adult men (aged ≥18 years) who were newly exposed to the target drug (canagliflozin) or comparator drugs (empagliflozin, dapagliflozin, sitagliptin, and liraglutide) during the study period lasting from 2013 to 2022 (with slight variation in dates of data availability across databases [ranges detailed in S1 to S5 Tables]). The comparator drugs were selected for several reasons: 1) to include a broad representation of different classes of AHAs (SGLT2is [empagliflozin and dapagliflozin], dipeptidyl peptidase-4 inhibitors [sitagliptin], and glucagon-like peptide 1 receptor agonists [liraglutide]); 2) they are highly effective agents commonly used in clinical practice for the management of type 2 diabetes mellitus; and 3) there is no well-established evidence linking these medications to the risk of prostatitis. High-level summaries of the overall study design are shown in Figs 1a and 1b, respectively.
CCAE, commercial claims and encounters; CDM, Clinformatics® Data Mart; EHR, electronic health record; EMR, electronic medical record; LASSO, least absolute shrinkage and selection operator; MDC, medical data center; MDCD, multi-state Medicaid database; MDCR, Medicare supplemental; PH, proportional hazards; T2DM, type 2 diabetes mellitus.
Data were included for patients who had ≥ 1 diagnosis of type 2 diabetes mellitus, measurement of elevated hemoglobin A1C, or drug treatment for type 2 diabetes mellitus, with ≥1 type 2 diabetes mellitus diagnosis at any time prior to or within 365 days following the first treatment exposure (index date), and no diagnosis of type 1 diabetes mellitus or secondary diabetes. Diagnosis of type 2 diabetes mellitus was based on International Classification of Diseases (ICD)-9 and ICD-10 codes. Additionally, eligible patients had ≥ 365 days of observational data prior to the index date within the database to capture patient characteristics and establish a clean period. There were no other exclusion criteria.
For each treatment cohort, the at-risk period included all time exposed to the index drug, calculated as index date + 1 day until event occurrence, drug discontinuation, or the end of the observation period within the database. Drug discontinuation was determined by a 30-day period between the end of a given drug’s supply and any subsequent exposure (i.e., ≥ 60 days pass without refilling a 30-day prescription). Only the first at-risk period was considered for each patient within a given treatment cohort, though patients may have belonged to ≥1 cohort at different times.
Data sources
Patient-level data from 8 global administrative claims databases were used for this study, including 7 United States (US) databases and 1 Japanese database. Data were accessed on December 14, 2022. The claims databases included: (1) IQVIA™ Ambulatory electronic medical records (IQVIA™ Ambulatory EMR), (2) IQVIA™ PharMetrics adjudicated health plan claims database (IQVIA™ PharMetrics Plus), (3) Optum’s de-identified electronic health record (Optum® EHR), (4) Optum® Clinformatics® Data Mart – Extended Data Mart – Socioeconomic Status (Optum® CDM), (5) Merative MarketScan® Commercial Claims and Encounters Database (Merative MarketScan® CCAE), (6) Merative MarketScan® Multi-State Medicaid Database (Merative MarketScan® MDCD), (7) Merative MarketScan® Medicare Supplemental Database (Merative MarketScan® MDCR), and (8) the Japan Medical Data Center claims database (JMDC). These databases are described in detail in S6 Table. All databases were transformed to the Observational Medical Outcomes Partnership Common Data Model, which provides a standardized representation of database structure and clinical content to enable consistent analysis across disparate health care databases [16,17]. The analyses used commercially licensed de-identified databases and no direct involvement with patients, so no one could identify the individual participants. All authors had access to information, including study reports and manuscript materials, which were aggregated information. The study analysts could have access to patient-level data.
Outcome assessment
Cases were defined as having a prostatitis diagnosis (based on ICD-9 and ICD-10 codes, listed in S7 Table) following the index date, without the presence of testicular lesions, bladder neoplasm, or abdominal or inguinal hernia between 365 days before and 7 days after the index date. Events were identified as the earliest observed diagnosis of prostatitis and new events were defined after no evidence of history of or chronic prostatitis was observed in the previous 365 days. The case ascertainment method followed standard practices [18].
Negative control outcomes were used to estimate systematic error in each analytic design and enable empirical calibration [19]. Negative controls are outcomes believed not to be caused by any of the medications being studied. Any effect size estimates for negative control ideally should be close to the null. Negative control outcomes are defined as the first occurrence of the negative control concept or any of its descendants.
Statistical analysis
The crude on-treatment incidence of prostatitis (per 1000 person years [PYs]) was estimated for each treatment cohort (i.e., canagliflozin, empagliflozin, dapagliflozin, sitagliptin, and liraglutide; Fig 1b). To control for confounding, a large-scale regularized logistic regression (least absolute shrinkage and selection operator [LASSO]) was used to fit propensity score (PS) model [20], and a variable ratio matching (up to 100 maximum) on PS was employed for each pairwise comparison between the canagliflozin cohort and the comparator treatment cohort [21]. The Large Scale Propensity Scores (LSPS) models were based on a large-scale regularized regression, which included a large, comprehensive set of baseline covariates. Prior research has suggested that under typical circumstances, LSPS may also potentially account for unmeasured confounding indirectly through measured covariates that correlate with unmeasured factors [22]. The covariates used for PS matching included medical history, demographics (i.e., age, sex, race/ethnicity, index year, index month), 30-day preindex variables (i.e., all conditions, all medication use, all procedures, all measurements), 365-day preindex variables (i.e., same as 30-day variables plus measurement results where available), overlapping index date (i.e., drug groups, which are categorized by Anatomical Therapeutic Chemical code, with index dates overlapping with canagliflozin or comparator AHAs), and comorbidity indices calculated using all prior medical history.
Prior to formal analyses, every combination of time-at-risk, comparator, and database (N = 32) was required to pass diagnostic testing that included standardized difference of covariates (mean difference, < 0.1), equipoise (percentage of population with a preference of 0.3–0.7) [23], minimum detectable relative risks at 80% power, and expected absolute systematic error (negative controls; also used for empirical calibration of results). Note that the preference score was calculated based on a transformation of the PS (ranging from 0 to 1) and it reflected the degree of overlap in treatment choices for different therapies involved in the study. Per the research by Walker et al [23], preference score values within the range of 0.3 to 0.7 suggest that there is a significant degree of similarity in the characteristics of patients receiving each of the treatments under comparison.
A conditional Cox proportional hazard model based on time-to-event analysis was used to estimate the hazard ratio (HR) within each individual database, conditioned on the matched set, with no adjustments made to account for multiple testing. The effect-size estimates across databases were combined using Bayesian random-effects meta-analysis based on nonnormal likelihood approximation to avoid bias due to small counts [24]. Empirical calibration was performed by first computing meta-analytic estimates for all negative controls, using those to fit empirical null distributions, and finally calibrating the meta-analytic estimates for the outcomes of interest.
Results
Study population
Among the 8 databases that passed the diagnostic testing for inclusion in the statistical analyses comparison, the sample sizes were considerably larger for IQVIA™ Ambulatory EMR, IQVIA™ PharMetrics Plus, Optum® EHR, Optum® CDM, and Merative MarketScan® CCAE compared with JMDC, Merative MarketScan® MDCD, and Merative MarketScan® MDCR (Fig 2). A total of 388,893 adult male patients with type 2 diabetes mellitus received canagliflozin, and were matched with 657,134 patients who received empagliflozin, 340,539 who received dapagliflozin, 819,047 who received sitagliptin, and 278,684 who received liraglutide. Across the databases, the mean age for new users of canagliflozin ranged from 51.2 (Merative MarketScan® MDCD) to 71.7 years (Merative MarketScan® MDCR). The baseline demographics and clinical characteristics of the 5 treatment cohorts are provided in the supporting information (S1 Table [canagliflozin] and S2 to S5 Tables [comparator AHAs]). The median treatment duration ranged from 4 to 6 months for most databases, except for IQVIA™ Ambulatory EMR and Optum® EHR (approximately 1 month for each) and JMDC (approximately 10 months).
CDM, Clinformatics® Data Mart; CCAE, commercial claims and encounters; EHR, electronic health record; EMR, electronic medical record; MDC, medical data center; MDCD, multi-state Medicaid database; MDCR, Medicare supplemental.
Baseline patient characteristics from the 8 databases were analyzed based on individual (not aggregated) verbatim terms. Across the databases, most patients in the canagliflozin cohort had hypertension/hypertensive disorder and hyperlipidemia; other common comorbidities included obesity, chest pain, low back pain, and respiratory illnesses during the baseline period (S1 Table). Other AHAs, antihypertensive medications, and cholesterol-lowering drugs were common concomitant treatments.
During the follow-up period, the crude on-treatment incidence rates showed that prostatitis was uncommon among patients in the canagliflozin cohort in nearly all databases (range, 4.2–7.7 per 1000 PYs) and were similar to those observed for the 4 comparator cohorts. The exception to this was a higher crude incidence rate of prostatitis in the Merative MarketScan® MDCR (range, 10.1–12.1 per 1000 PYs; Fig 3). The crude incidence rates and event counts are provided in S8 Table.
CDM, Clinformatics® Data Mart; CCAE, commercial claims and encounters; EHR, electronic health record; EMR, electronic medical record; MDC, medical data center; MDCD, multi-state Medicaid database; MDCR, Medicare supplemental; PY, person-year.
Risk estimation
After PS matching, all available covariates were well-matched between cohorts (as an example, covariate balance before and after PS matching using the IQVIATM Ambulatory EMR database is presented in S1 Fig). Effect estimates across the databases for prostatitis risk varied on both sides of the null effect, with the HR relatively close to 1.0. The minimum detectable relative risks were low in most databases, often <1.5, and almost always <2.0. Estimates from the meta-analysis were very close to 1.0 with upper bounds <1.50 for all pairwise comparisons (HR and 95% confidence interval [CI]): canagliflozin versus empagliflozin (0.91 [0.75–1.09]), canagliflozin versus dapagliflozin (1.14 [0.87–1.50]), canagliflozin versus sitagliptin (1.06 [0.88–1.27]), and canagliflozin versus liraglutide (1.06 [0.81–1.37]). There was neither a protective nor harmful association for risk of prostatitis with canagliflozin versus the comparator AHAs (i.e., empagliflozin, dapagliflozin, sitagliptin, and liraglutide; Fig 4).
CDM, Clinformatics® Data Mart; CCAE, commercial claims and encounters; EHR, electronic health record; EMR, electronic medical record; MDC, medical data center; MDCD, multi-state Medicaid database; MDCR, Medicare supplemental.
Discussion
Using real-world clinical data from 8 global claims databases that met diagnostic criteria, this retrospective, observational study compared the incidence of prostatitis in adult male patients with type 2 diabetes mellitus who were new users of canagliflozin to that of 4 comparator AHA treatments (i.e., empagliflozin, dapagliflozin, sitagliptin, and liraglutide). During the follow-up period, the crude on-treatment incidence rates showed that prostatitis was uncommon in the canagliflozin cohort in nearly all databases (range, 4.2–7.7 per 1000 PYs) and were similar to those found among comparator treatment cohorts. The exception to this observation was a higher crude incidence of prostatitis in 1 database (Merative MarketScan® MDCR, range, 10.1–12.1 per 1000 PYs), which mainly consists of patients older than 65 years of age. After PS matching using all available covariates, there was no evidence of a statistically significantly increased risk of prostatitis in patients who received canagliflozin compared with the other AHAs evaluated in this study, including other SGLT2is. In a prior study using the Kaiser Permanente Northwest (Portland, Oregon) database, the incidence of physician-diagnosed prostatitis was 4.9 per 1000 PYs, although the study was not restricted to patients with type 2 diabetes mellitus [25].
While UTIs have been identified as adverse drug reactions associated with the use of SGLT2is, there is mixed support from other large data analyses for an association between SGLT2i use and an increased risk of genitourinary infections. Multiple meta-analyses have found SGLT2i use to be associated with an increased risk of genital mycotic infections but not UTIs [14,26], while a meta-analysis among older patients (≥65 years) found no increased incidence of genitourinary infections within 6 months of SGLT2i initiation [13]. It has been suggested that these disparities may be due to the specific SGLT2i molecule being studied, as initial meta-analyses were conducted using patient data from the earlier available drugs [27]. In more recent studies, the increased risk of UTI that was observed for dapagliflozin was not seen with either canagliflozin or empagliflozin [14,15,26,27]. Prostatitis has received limited attention in the literature separate from UTIs as a whole, but findings of the current study suggest that rates of prostatitis are low and similar in patients with type 2 diabetes mellitus who received treatment with the SGLT2i canagliflozin or 1 of the comparator drugs, with no statistically significantly increased risk of prostatitis from canagliflozin treatment versus any of the comparator drugs evaluated in this study.
The strengths of the current analysis include its use of well-established LSPS matching methods to control confounding. The results were robust, including very large patient cohorts and evaluating treatment consistency based on 9 years of data from routine clinical practice from 8 large global administrative claims databases. Analyses were required to pass stringent prespecified diagnostic testing and negative control outcomes were also implemented to evaluate any potential systematic study design bias and used for empirical calibration of the overall treatment effect estimates. Additionally, this analysis specifically evaluated the risk for prostatitis among patients with type 2 diabetes mellitus, whereas prior analyses have focused on risk assessments across broader classifications of infection, such as overall genitourinary infections.
However, this analysis has several limitations. Although the results showed no statistically significant association of prostatitis with canagliflozin over comparator AHAs, the observational design precludes completely ruling out a causal effect. While the Bayesian meta-analytical estimates for the risk of prostatitis with canagliflozin versus comparator AHAs showed point estimates very close to 1.0, a moderate risk increase cannot be ruled out as the upper bounds of the 95% CI were <1.50. The study used observational retrospective data that were originally collected for insurance claims purposes and medical care processing from routine clinical practice. Accordingly, data accuracy and the level of clinical details may be different than those from a data source that has been prospectively collected or specifically designed to address certain clinical questions. Data-related limitations include dependency on the accuracy of codes and algorithms to identify at risk conditions and the lack of details regarding the circumstances related to prostatitis diagnosis (e.g., urinalysis and urine culture results, prostate-specific variables like prostate-specific antigen levels or imaging results, and alternative etiologies). The use of a 1-year baseline period to capture covariates may not have accounted for long-term chronic conditions that could predispose patients to prostatitis. As the study includes data spanning from 2013 to 2022, the potential for temporal bias (changes in clinical guidelines, prescribing habits, or diagnostic criteria over time) cannot be excluded; this is an inherent limitation of real-world studies. The median follow-up period of 4–6 months for most databases is also relatively short for detecting prostatitis. Misclassification of participants with type 2 diabetes mellitus or prostatitis cannot be ruled out due to the use of ICD codes for diagnosis of these conditions. Bias due to non-random censoring is also a limitation of our study as treatment exposure was determined based on pharmacy drug dispensing records rather than actual administration. In addition, treatment discontinuation due to UTIs (a known SGLT2i risk) might have censored prostatitis events (if the prostatitis event proceeded shortly after a UTI event), potentially biasing the HRs for the risk of prostatitis. The study findings may also be subject to residual confounding effects due to its observational nature and unmeasured confounding such as the frequency of prostate examinations, antibiotic use, sexual history, and lifestyle factors (e.g., smoking, alcohol consumption). Therefore, it was critically essential that our analysis used LSPS methods to minimize the residual confounding and the implementation of negative controls further reassured that no systematic bias or residual confounding were observed, and the results were calibrated based on the empirical distributions of the negative controls [22]. Clinical interpretability of the findings is limited by the lack of stratification of prostatitis by severity (acute versus chronic) and whether prostatitis presents differently in SGLT2i users (e.g., bacterial versus nonbacterial). A broader definition of prostatitis was used in this study to address the specific Health Authority query; further stratification would have been limited by the use of ICD diagnosis codes. The analyses were conducted for 4 specific AHAs using data from 7 US databases and 1 Japanese database, with the majority of patients coming from the US (particularly from the Optum® EHR, IQVIATM PharMetrics Plus, and IQVIATM Ambulatory EMR databases); therefore, the findings are not generalizable to other geographic regions and other AHAs. Regarding statistical methodology, although the Cox proportional hazard model that was used to estimate the HR for prostatitis within each database assumes proportionality of hazards, a quantitative evaluation of proportionality of hazards was not conducted. Finally, while there is an established relationship between lower urinary tract dysfunction or prostate enlargement and the development of prostatitis, these risk factors are beyond the scope of this study and would be important endpoints for future analyses.
This study, using a comparative cohort design with active comparators, found no evidence of a statistically significantly increased risk of prostatitis in adult male patients with type 2 diabetes mellitus who received canagliflozin compared with the other AHAs evaluated in this study, providing evidence for the lack of a safety issue for this hypothesized association. Future studies based on clinical datasets are needed to validate these findings. In addition, the potential for age-related effect modification (if any) confounding on the risk of prostatitis among SGLT2i users could be explored by future research.
Supporting information
S1 Fig. Covariate Balance Before and After Propensity Score Matching (IQVIATM Ambulatory EMR Database).
EMR, electronic medical record.
https://doi.org/10.1371/journal.pone.0341745.s001
(PDF)
S1 Table. Baseline Demographics and Clinical Characteristics of the Canagliflozin Cohort.
CCAE, Commercial Claims and Encounters; CDM, Clinformatics® Data Mart; EHR, electronic health record; EMR, electronic medical record; JMDC, Japan medical data center; MDCD, multi-state Medicaid database; MDCR, Medicare supplemental; NA, not available; SD, standard deviation. a Select comorbidities and concomitant medications are those with a prevalence of at least 20% in any treatment cohort. Concomitant medications exclude those that were evaluated in this study (i.e., canagliflozin, dapagliflozin, empagliflozin, liraglutide, and sitagliptin).
https://doi.org/10.1371/journal.pone.0341745.s002
(XLSX)
S2 Table. Baseline Demographics and Clinical Characteristics of the Dapagliflozin Cohort.
CCAE, Commercial Claims and Encounters; CDM, Clinformatics® Data Mart; EHR, electronic health record; EMR, electronic medical record; JMDC, Japan medical data center; MDCD, multi-state Medicaid database; MDCR, Medicare supplemental; NA, not available; SD, standard deviation. a Select comorbidities and concomitant medications are those with a prevalence of at least 20% in any treatment cohort. Concomitant medications exclude those that were evaluated in this study (i.e., canagliflozin, dapagliflozin, empagliflozin, liraglutide, and sitagliptin).
https://doi.org/10.1371/journal.pone.0341745.s003
(XLSX)
S3 Table. Baseline Demographics and Clinical Characteristics of the Empagliflozin Cohort.
CCAE, Commercial Claims and Encounters; CDM, Clinformatics® Data Mart; EHR, electronic health record; EMR, electronic medical record; JMDC, Japan medical data center; MDCD, multi-state Medicaid database; MDCR, Medicare supplemental; NA, not available; SD, standard deviation. a Select comorbidities and concomitant medications are those with a prevalence of at least 20% in any treatment cohort. Concomitant medications exclude those that were evaluated in this study (i.e., canagliflozin, dapagliflozin, empagliflozin, liraglutide, and sitagliptin).
https://doi.org/10.1371/journal.pone.0341745.s004
(XLSX)
S4 Table. Baseline Demographics and Clinical Characteristics of the Liraglutide Cohort.
CCAE, Commercial Claims and Encounters; CDM, Clinformatics® Data Mart; EHR, electronic health record; EMR, electronic medical record; JMDC, Japan medical data center; MDCD, multi-state Medicaid database; MDCR, Medicare supplemental; NA, not available; SD, standard deviation. a Select comorbidities and concomitant medications are those with a prevalence of at least 20% in any treatment cohort. Concomitant medications exclude those that were evaluated in this study (i.e., canagliflozin, dapagliflozin, empagliflozin, liraglutide, and sitagliptin).
https://doi.org/10.1371/journal.pone.0341745.s005
(XLSX)
S5 Table. Baseline Demographics and Clinical Characteristics of the Sitagliptin Cohort.
CCAE, Commercial Claims and Encounters; CDM, Clinformatics® Data Mart; EHR, electronic health record; EMR, electronic medical record; JMDC, Japan medical data center; MDCD, multi-state Medicaid database; MDCR, Medicare supplemental; NA, not available; SD, standard deviation. a Select comorbidities and concomitant medications are those with a prevalence of at least 20% in any treatment cohort. Concomitant medications exclude those that were evaluated in this study (i.e., canagliflozin, dapagliflozin, empagliflozin, liraglutide, and sitagliptin).
https://doi.org/10.1371/journal.pone.0341745.s006
(XLSX)
S6 Table. Summary of Administrative Claims Databases Included in This Analysis.
CCAE, commercial claims and encounters; CDM, Clinformatics® Data Mart; EHR, electronic health record; EMR, electronic medical record; HIPAA, Health Insurance Portability and Accountability Act; MDC, medical data center; MDCD, multi-state Medicaid database; MDCR, Medicare supplemental; NLP, natural language processing.
https://doi.org/10.1371/journal.pone.0341745.s007
(DOCX)
S7 Table. International Classification of Diseases Codes Used to Define Prostatitis.
CM, Clinical Modification; ICD, International Classification of Diseases.
https://doi.org/10.1371/journal.pone.0341745.s008
(XLSX)
S8 Table. Crude Incidence Rates and Event Counts of Prostatitis.
CCAE, Commercial Claims and Encounters; CDM, Clinformatics® Data Mart; EHR, electronic health record; EMR, electronic medical record; IR, incidence rate; JMDC, Japan medical data center; MDCD, multi-state Medicaid database; MDCR, Medicare supplemental. Note: Data for the following comparisons are not presented because they failed pre-specified diagnostic testing: 1) canagliflozin vs. dapagliflozin – Optum CDM; 2) canagliflozin vs. liraglutide – JMDC.
https://doi.org/10.1371/journal.pone.0341745.s009
(XLSX)
Acknowledgments
Medical writing support was provided by Alex Pilote, PhD, of Lumanity Communications Inc.
References
- 1. Chaplin S. SGLT2 inhibitors and risk of genitourinary infections. Prescriber. 2016;27(12):26–30.
- 2. Geerlings S, Fonseca V, Castro-Diaz D, List J, Parikh S. Genital and urinary tract infections in diabetes: impact of pharmacologically-induced glucosuria. Diabetes Res Clin Pract. 2014;103(3):373–81. pmid:24529566
- 3. Geerlings SE, Brouwer EC, Gaastra W, Verhoef J, Hoepelman AIM. Effect of glucose and pH on uropathogenic and non-uropathogenic Escherichia coli: studies with urine from diabetic and non-diabetic individuals. J Med Microbiol. 1999;48(6):535–9. pmid:10359302
- 4. Turan H, Serefhanoglu K, Torun AN, Kulaksizoglu S, Kulaksizoglu M, Pamuk B, et al. Frequency, risk factors, and responsible pathogenic microorganisms of asymptomatic bacteriuria in patients with type 2 diabetes mellitus. Jpn J Infect Dis. 2008;61(3):236–8. pmid:18503181
- 5. Hoepelman AIM, Meiland R, Geerlings SE. Pathogenesis and management of bacterial urinary tract infections in adult patients with diabetes mellitus. Int J Antimicrob Agents. 2003;22 Suppl 2:35–43. pmid:14527769
- 6. de Lastours V, Foxman B. Urinary tract infection in diabetes: epidemiologic considerations. Curr Infect Dis Rep. 2014;16(1):389. pmid:24407547
- 7. Nitzan O, Elias M, Chazan B, Saliba W. Urinary tract infections in patients with type 2 diabetes mellitus: review of prevalence, diagnosis, and management. Diabetes Metab Syndr Obes. 2015;8:129–36. pmid:25759592
- 8. Gandhi J, Dagur G, Warren K, Smith NL, Khan SA. Genitourinary complications of diabetes mellitus: an overview of pathogenesis, evaluation, and management. Curr Diabetes Rev. 2017;13(5):498–518. pmid:27774877
- 9. Saijo Y, Okada H, Hata S, Nakajima H, Kitagawa N, Okamura T, et al. Reasons for discontinuing treatment with sodium-glucose cotransporter 2 inhibitors in patients with diabetes in real-world settings: the KAMOGAWA-A study. J Clin Med. 2023;12(22):6993. pmid:38002608
- 10. Plodkowski RA, McGarvey ME, Huribal HM, Reisinger-Kindle K, Kramer B, Solomon M, et al. SGLT2 inhibitors for type 2 diabetes mellitus treatment. Fed Pract. 2015;32(Suppl 11):8S-15S. pmid:30766102
- 11. Simhadri PK, Vaitla P, Sriperumbuduri S, Chandramohan D, Singh P, Murari U. Sodium-glucose co-transporter-2 inhibitors causing Candida tropicalis fungemia and renal abscess. JCEM Case Rep. 2024;2(2):luae010. pmid:38304006
- 12. Bartolo C, Hall V, Friedman ND, Lanyon C, Fuller A, Morrissey CO, et al. Bittersweet: infective complications of drug-induced glycosuria in patients with diabetes mellitus on SGLT2-inhibitors: two case reports. BMC Infect Dis. 2021;21(1):284. pmid:33743624
- 13. Varshney N, Billups SJ, Saseen JJ, Fixen CW. Sodium-glucose cotransporter-2 inhibitors and risk for genitourinary infections in older adults with type 2 diabetes. Ther Adv Drug Saf. 2021;12:2042098621997703. pmid:33854754
- 14. Puckrin R, Saltiel M-P, Reynier P, Azoulay L, Yu OHY, Filion KB. SGLT-2 inhibitors and the risk of infections: a systematic review and meta-analysis of randomized controlled trials. Acta Diabetol. 2018;55(5):503–14. pmid:29484489
- 15. Liu J, Li L, Li S, Jia P, Deng K, Chen W, et al. Effects of SGLT2 inhibitors on UTIs and genital infections in type 2 diabetes mellitus: a systematic review and meta-analysis. Sci Rep. 2017;7(1):2824. pmid:28588220
- 16. Hripcsak G, Duke JD, Shah NH, Reich CG, Huser V, Schuemie MJ, et al. Observational Health Data Sciences and Informatics (OHDSI): opportunities for observational researchers. Stud Health Technol Inform. 2015;216:574–8. pmid:26262116
- 17. Voss EA, Makadia R, Matcho A, Ma Q, Knoll C, Schuemie M, et al. Feasibility and utility of applications of the common data model to multiple, disparate observational health databases. J Am Med Inform Assoc. 2015;22(3):553–64. pmid:25670757
- 18. Rao GA, Shoaibi A, Makadia R, Hardin J, Swerdel J, Weaver J, et al. CohortDiagnostics: Phenotype evaluation across a network of observational data sources using population-level characterization. PLoS One. 2025;20(1):e0310634. pmid:39820599
- 19. Schuemie MJ, Ryan PB, DuMouchel W, Suchard MA, Madigan D. Interpreting observational studies: why empirical calibration is needed to correct p-values. Stat Med. 2014;33(2):209–18. pmid:23900808
- 20. Tian Y, Schuemie MJ, Suchard MA. Evaluating large-scale propensity score performance through real-world and synthetic data experiments. Int J Epidemiol. 2018;47(6):2005–14. pmid:29939268
- 21. Rassen JA, Shelat AA, Myers J, Glynn RJ, Rothman KJ, Schneeweiss S. One-to-many propensity score matching in cohort studies. Pharmacoepidemiol Drug Saf. 2012;21 Suppl 2:69–80. pmid:22552982
- 22. Zhang L, Wang Y, Schuemie MJ, Blei DM, Hripcsak G. Adjusting for indirectly measured confounding using large-scale propensity score. J Biomed Inform. 2022;134:104204. pmid:36108816
- 23. Walker AM, Patrick AR, Lauer MS, Hornbrook MC, Marin MG, Platt R, et al. A tool for assessing the feasibility of comparative effectiveness research. Comparat Effect Res. 2013;3:11–20. https://doi.org/10.2147/CER.S40357
- 24. Schuemie MJ, Chen Y, Madigan D, Suchard MA. Combining cox regressions across a heterogeneous distributed research network facing small and zero counts. Stat Methods Med Res. 2022;31(3):438–50. pmid:34841975
- 25. Clemens JQ, Meenan RT, O’Keeffe Rosetti MC, Gao SY, Calhoun EA. Incidence and clinical characteristics of National Institutes of Health type III prostatitis in the community. J Urol. 2005;174(6):2319–22. pmid:16280832
- 26. Dave CV, Schneeweiss S, Kim D, Fralick M, Tong A, Patorno E. Sodium-glucose cotransporter-2 inhibitors and the risk for severe urinary tract infections: a population-based cohort study. Ann Intern Med. 2019;171(4):248–56. pmid:31357213
- 27. Fisher A, Fralick M, Filion KB, Dell’Aniello S, Douros A, Tremblay É, et al. Sodium-glucose co-transporter-2 inhibitors and the risk of urosepsis: a multi-site, prevalent new-user cohort study. Diabetes Obes Metab. 2020;22(9):1648–58. pmid:32383792