Clinical characteristics and outcomes of patients with type 2 diabetes mellitus and chronic kidney disease from two new-user medication cohorts: a retrospective cohort study using regional electronic health records database in China Authors: Xiaoqing Deng, Yanyan Chen, Zhe Xu, Huan Wei, Yuqin Liao, Fangfang Liu, Alfredo E Farjat, Nikolaus G Oberprieler, Liming Chen Published: 2025 ABSTRACT AbstractObjectivesGiven the rapidly evolving therapeutic landscape for type 2 diabetes (T2D) and chronic kidney disease (CKD), this study aimed to characterise the clinical profiles and real-world outcomes of patients with T2D and CKD in China who initiated sodium-glucose cotransporter 2 inhibitors (SGLT2i) or glucagon-like peptide-1 receptor agonists (GLP-1 RA).DesignRetrospective cohort study.SettingDemographic and clinical data of patients from a regional electronic health records database in Tianjin, China between 2012 and 2019 were used.ParticipantsAdult patients diagnosed with T2D and CKD who initiated SGLT2i or GLP-1 RA from 2012 to 2019.Primary and secondary outcome measuresBaseline demographic and disease characteristics, comorbidities and comedications, healthcare resource utilisation (HRU), and clinical outcomes were assessed using descriptive statistics.ResultsA total of 935 and 4821 patients were included in SGLT2i and GLP-1 RA cohorts, with the mean ages of 59 and 56 years, respectively. Both cohorts had similar durations of T2D (mean: 5 years) and CKD (mean: 3 years). In SGLT2i and GLP-1 RA cohorts, 54.4% and 56.9% of patients had hemoglobin A1c (HbA1c) >7%, and 50.5% and 54.1% were classified as CKD stage 1 at baseline. During the follow-up period (median 1.4 months for SGLT2i cohort and median 2.3 months for GLP-1 RA cohort), higher numbers of specialist visits compared with general practitioner visits were observed numerically for both cohorts. The incidence rates (95% CI) of kidney failure per 100 person-years were 3.1 (1.0, 7.3) for SGLT2i cohort, and 4.9 (3.9, 6.0) for GLP-1 RA cohort during follow-up.ConclusionsThis study provides descriptive evidence regarding the clinical characteristics and outcomes of patients with T2D and CKD who initiated SGLT2i or GLP-1 RA in China. The results are important for understanding the existing HRU and residual risk of severe clinical outcomes in such patient populations. The findings also provide a solid foundation for future research aimed at examining the clinical outcomes of new therapeutic options for T2D and CKD. STRENGTHS AND LIMITATIONS OF THIS STUDY Study datasets were standardised according to the Observational Medical Outcomes Partnership (OMOP) common data model, ensuring consistency and reliability in its analyses. Study datasets were standardised according to the Observational Medical Outcomes Partnership (OMOP) common data model, ensuring consistency and reliability in its analyses. The number of participants included for analysis was large for both sodium-glucose transporter 2 inhibitor (SGLT2i) and glucagon-like peptide-1 receptor agonist (GLP-1 RA) medication cohorts. The number of participants included for analysis was large for both sodium-glucose transporter 2 inhibitor (SGLT2i) and glucagon-like peptide-1 receptor agonist (GLP-1 RA) medication cohorts. The study utilized a regional database that captured comprehensive patient-level electronic medical records from healthcare organizations within the region, thereby limiting the generalizability of findings to broader populations in China. The study utilized a regional database that captured comprehensive patient-level electronic medical records from healthcare organizations within the region, thereby limiting the generalizability of findings to broader populations in China. New treatment options introduced after 2019, such as nonsteroidal mineralocorticoid receptor antagonist (nsMRA), were not included in the study due to the study period setting, which limits its applicability to current clinical practice. New treatment options introduced after 2019, such as nonsteroidal mineralocorticoid receptor antagonist (nsMRA), were not included in the study due to the study period setting, which limits its applicability to current clinical practice. Introduction According to the International Diabetes Federation, the global diabetes prevalence in 2019 is estimated to be 9.3% (463 million) and is expected to rise to 10.2% (578 million) by 2030.1 China is the country with the highest number of people living with diabetes of 116 million in 2019.1 Patients with type 2 diabetes mellitus (T2D) have a high prevalence and incidence of chronic kidney disease (CKD).2 The relationship between T2D and CKD is bidirectional and multifaceted, with an intricate network encompassing pathophysiology, clinical management and epidemiology.3 Approximately 40% of patients with diabetes worldwide have CKD.4 The prevalence of CKD among patients with diabetes in China is estimated to be 31%.5 Patients with T2D and CKD are at high risk of kidney failure, cardiovascular events and death. The risks of progression towards kidney failure and cardiovascular events increase with the severity and stage of CKD.6 Therefore, the major objectives in the management of patients with T2D and CKD are to preserve kidney function to reduce the risk of end-stage kidney disease, cardiovascular events and mortality. The clinical landscape for the treatment of patients with T2D and CKD is rapidly evolving with the introduction of new treatments. The consensus statements from the American Diabetes Association (ADA) and the Kidney Disease: Improving Global Outcomes (KDIGO) provided evidence-based recommendations for the management of T2D and CKD in 2022,7 including renin-angiotensin system inhibitors (RASi) (eg, angiotensin-converting enzyme inhibitors (ACEi), angiotensin receptor blockers (ARB)) for patients with hypertension and albuminuria, statin for atherosclerotic cardiovascular disease prevention, metformin, sodium-glucose cotransporter-2 inhibitors (SGLT2i) (eg, dapagliflozin, empagliflozin), glucagon-like peptide-1 receptor agonists (GLP-1 RA) (eg, liraglutide, exenatide) as glucose-lowering agents, and nonsteroidal mineralocorticoid receptor antagonist (nsMRA) for kidney and cardiovascular protection. Specifically, SGLT2i lowers the glucose level without stimulating insulin release through inhibition of glucose reabsorption in the renal tubule.8 The DAPA-CKD trial showed that patients with CKD who were treated with dapagliflozin had a lower risk of kidney or cardiovascular events, compared with a placebo.9 The CREDENCE trial showed that long-term administration of canagliflozin conferred renal and cardiovascular protection in patients with T2D and CKD.10 Evidence from clinical trials and real-world studies also suggests a decreased risk of cardiovascular events associated with SGLT2i use.1115 GLP-1 RA mimics the action of endogenous GLP-1 by activating its receptor, thereby enhancing insulin secretion and suppressing glucagon release.16 GLP-1 RA has been recommended in adults with T2D and CKD who have not achieved individualized glycemic targets despite use of metformin and SGLT2i treatment.17 An exploratory analysis of the REWIND trial suggested that dulaglutide may reduce the progression of kidney disease in adults with T2D for about 5 years of exposure.18 In a secondary analysis of the LEADER trial,19 among patients with T2D at high risk for cardiovascular disease, a lower risk of a composite kidney outcome was observed with liraglutide compared with placebo. GLP-1 RAs have also demonstrated a reduction in cardiovascular events.2022 Characterizing a population of initiators of SGLT2i or GLP-1 RA and understanding the potential benefits and risks in real-world scenarios is crucial for clinical decision-making. To date, there has been no real-world study conducted in mainland China focusing on the profile of patients with T2D and CKD that were treated with the two classes of glucose-lowering agents. To better guide the disease management of T2D and CKD in China, it will be of interest and value for the healthcare practitioners to understand the characteristics of patients with usage of SGLT2i and GLP-1 RA in routine clinical practice. This study examined patient characteristics and clinical outcomes among new users of SGLT2i and GLP-1 RA with CKD and T2D in real-world clinical settings in China. Materials and methods Study design and population This study was conducted as part of the FOUNTAIN (FinerenOne mUltidatabase NeTwork for evidence generAtIoN) platform23 and designed as a retrospective study to evaluate the patient profile and clinical outcomes in adult patients in China with T2D and CKD using Tianjin Healthcare and Medical Big Data Platform. The study was approved by the Ethics Committee of Tianjin Medical University Chu Hsien-I Memorial Hospital under the number ZXYJNYYhMEC2023-4. Eligible patients were those who were aged ≥18 years when initiated one of the study medications of SGLT2i and GLP-1 RA for the first time in 12 months, had a diagnosis of T2D and CKD, and had continuous enrolment in the Tianjin database for at least 12 months prior to treatment initiation. Patients with a diagnosis of type 1 diabetes mellitus, kidney cancer or kidney failure on or before treatment initiation were excluded. Detailed definitions are provided in the online supplemental methods. New users were defined as patients receiving a prescription for one of the study medications during the study period, with no prior prescription of that medication class in the preceding 12 months. The eligibility for each medication-specific cohort was evaluated independently. These two cohorts were not mutually exclusive, allowing individual patients to be included in both if applicable. The study period was from 1 January 2012 to 31 December 2019. The end of the study period was selected to avoid any disruptions associated with the COVID-19 pandemic. The index date was defined as the date of the first eligible prescription that fulfilled the criteria for new use. The follow-up period began the day after the index date and continued until meeting the first censoring criteria: end of the study period, disenrollment from the database, death, discontinuation of the medication class (defined in Section 2.4.1), or development of kidney failure or cancer. Detailed definitions are provided in the online supplemental methods. Data collection and curation The Tianjin Healthcare and Medical Big Data Platform covers approximately 15 million residents in Tianjin, China. Patient information from 43 tertiary and 39 secondary hospitals covering both urban and rural areas was linked via unique identifier and was de-identified for analysis. The study team had access to the de-identified patient-level data related to the study. Given that electronic health records (EHR) from the Tianjin database are primarily collected for patient care rather than research purposes, there were potential biases related to incomplete, miscoded or inaccurate data.24 25 The analysis dataset was cleaned and standardized onsite at the Tianjin platform following local regulations. The study-specific datasets were standardized according to the Observational Health Data Sciences and Informatics Observational Medical Outcomes Partnership (OMOP) common data model to align with international coding standards. During the data curation process, study variables were identified not only by structured codes like International Classification of Diseases for diagnoses, but also enhanced by free-text information extraction from sources such as diagnosis texts, medication names (both generic and brand), lab measurement names and clinician notes, which were carefully mapped to OMOP standard concepts to ensure consistency across datasets. Study measures Exposures Exposures to study medications were identified through prescriptions recorded in Tianjin database. Detailed drug list is provided in the online supplemental table 1. If multiple drugs from the same class were prescribed on the same index date, all were included to determine the follow-up. Current use periods of the study drug spanned from the day after the index date until the end of supply for consecutive prescriptions, plus a 30-day grace period. Any medication within the corresponding class were jointly considered. If consecutive prescriptions were separated by gaps of 30 days or less, gaps were included. Treatment discontinuation was defined as the last day of the grace period after final prescription of study medication in current use period. Detailed definitions are provided in the online supplemental methods. Variables Variables were evaluated separately during the baseline period and follow-up period. To focus on new-onset events for clinical outcomes, outcome-specific analysis groups were generated to exclude patients with a history of the events before or on the index date. The variables include in the analysis are listed below, and the specific assessment windows for each variable are shown in figure 1. Variable assessment windows relative to the study index date for medication-specific cohorts. CKD=chronic kidney disease; GLP-1 RA=glucagon-like peptide-1 receptor agonist; HCU=healthcare utilization; SGLT2i=sodium-glucose cotransporter 2 inhibitor; T1D=type 1 diabetes mellitus; T2D=type 2 diabetes mellitus. aStudy period: 01 January 2012 through 31 December 2019. bSGLT2i, GLP 1 RA, sMRA, ACEi, ARB. cConditions requiring ≥365 days for accurate assessment, including hypertension, glomerulonephritis, renovascular disease, polycystic kidney disease, autoimmune disease and amputations. dOutcome-specific exclusion criteria were assessed in outcome-specific washout periods; outcome-specific washout periods were applied separately before outcome-specific analyses. eCensored at the earliest of the end of the study period (31 December 2019), disenrollment, death, T1D diagnosis, discontinuation of medication class or development of kidney failure. Demographic and lifestyle factors: Demographic and lifestyle factors included variables such as age, sex, race and smoking status, etc. Clinical characteristics: Clinical characteristics included the variables related to the severity of T2D and CKD. Comorbidities: Comorbidities included macrovascular complications, cardiovascular risk factors and severe liver disease, etc. Comedications: Comedications included cardiovascular medications, anti-inflammatory drugs and chemotherapeutic agents, etc. Healthcare resource utilization (HRU): HRU variables included counts of different types of medical visits and the regular kidney function measurements counts. Cardiovascular outcomes: Cardiovascular outcomes included major events such as acute coronary syndrome, stroke, congestive heart failure and atrial fibrillation. Kidney outcomes: Kidney outcomes were the onset of kidney failure events, defined as any of the following: two outpatient estimated glomerular filtration rate (eGFR) measurements of <15 mL/min/1.73 m2 separated by at least 90 days, dependence on dialysis, a diagnosis of kidney failure or CKD stage 5, or kidney transplant. Statistical analysis Given the descriptive nature of this study, all statistical analyses focused on descriptive statistics, and no formal comparisons were made either between or within cohorts. The individuals in the two cohorts were not mutually exclusive by the cohort definition. Cardiovascular and kidney outcomes were identified during patient follow-up. Crude incidence rates (IR) were calculated for each outcome as the number of events occurring during the person-time at risk divided by the total person-time at risk (expressed in 100 person-years). Incidence rates were reported as point estimates and 95% confidence intervals (CIs). Sensitivity analysis was performed to evaluate the clinical outcomes with modified definitions of grace period and stockpiling to avoid potential bias. Missing values were reported as frequency and percentage. All statistical analyses were performed using SAS software V.9.4. Patient and public involvement statement Patients and/or the public were not involved in the design, conduct, reporting or dissemination plans of this research. Results Following eligibility screening (online supplemental figure 1 and 2), 935 patients were included in the SGLT2i cohort, and 4821 patients were included in the GLP-1 RA cohort. Patient characteristics Baseline demographic characteristics and lifestyle variables are summarized in Table 1 and online supplemental table 2. For both the SGLT2i and GLP-1 RA cohorts, the proportion of male patients was slightly higher, as of 61.4% and 55.7%, respectively. The mean ages at the index date were 59.1 (SD: 12.7) years for the SGLT2i cohort and 55.7 (SD: 12.1) years for the GLP-1 RA cohort. Table 1Patient profile of new users for study medications at baselineSGLT2i cohort (n=935)GLP-1 RA cohort (n=4821)Age at index date (years) n (%)935 (100.0%)4821 (100.0%) Mean (SD)59.1 (12.7)55.7 (12.1)Sex,* n (%) Male574 (61.4%)2685 (55.7%) Female361 (38.6%)2136 (44.3%)Race,* n (%) Asian935 (100.0%)4821 (100.0%)Obesity,† n (%) Yes (by diagnosis or BMI‡ ≥30 kg/m²)42107 (11.4%)992 (20.6%) No203 (21.7%)811 (16.8%) Missing625 (66.8%)3018 (62.6%)Duration of T2D* (years) n(%)935 (100.0%)4821 (100.0%) Mean (SD)5.2 (2.9)5.0 (3.1)Medications for T2D (hypoglycaemic agents),§ n (%) GLP-1 RA and fixed-dose combinations155 (16.6%)NA SGLT2i and fixed-dose combinationsNA44 (0.9%) Metformin and fixed-dose combinations590 (63.1%)3531 (73.2%) Sulfonylureas and fixed-dose combinations254 (27.2%)1466 (30.4%) Alpha glucosidase inhibitors502 (53.7%)3009 (62.4%) Thiazolidinediones142 (15.2%)927 (19.2%) Dipeptidyl peptidase-4 inhibitors and fixed-dose combinations412 (44.1%)1438 (29.8%) Meglitinides (including repaglinide, nateglinide and mitiglinide)162 (17.3%)1008 (20.9%)Insulin use recorded,§ n (%)682 (72.9%)3853 (79.9%)HbA1c,† n (%) HbA1c ≤53 mmol/mol or ≤7%157 (16.8%)534 (11.1%) HbA1c >53 mmol/mol and ≤63.9 mmol/mol or >7% and ≤8%173 (18.5%)709 (14.7%) HbA1c >63.9 mmol/mol and ≤74.9 mmol/mol or >8% and ≤9%143 (15.3%)731 (15.2%) HbA1c >74.9 mmol/mol or >9%193 (20.6%)1305 (27.1%) HbA1c missing269 (28.8%)1542 (32.0%)The diabetes severity complications index† Key diagnoses for scoring of index score, n (%) Retinopathy288 (30.8%)1174 (24.4%) Nephropathy696 (74.4%)2951 (61.2%) Neuropathy20 (2.1%)66 (1.4%) Cerebrovascular257 (27.5%)992 (20.6%) Cardiovascular553 (59.1%)2431 (50.4%) Peripheral vascular disease537 (57.4%)2532 (52.5%) Metabolic complications6 (0.6%)64 (1.3%) Index score n(%)935 (100.0%)4821 (100.0%) Mean (SD)3.2 (2.0)2.6 (1.9)Duration of CKD* (years) n(%)935 (100.0%)4821 (100.0%) Mean (SD)3.2 (2.8)3.3 (2.6)CKD stage based on eGFR† or diagnosis code,† n (%) Stage 1: ≥90, normal or high472 (50.5%)2607 (54.1%) Stage 2: 60–89, mildly decreased166 (17.8%)641 (13.3%) Stage 3: mildly to severely decreased98 (10.5%)274 (5.7%) Stage 3a: 45–59, mildly to moderately decreased65 (7.0%)220 (4.6%) Stage 3b: 30–44, moderately to severely decreased33 (3.5%)52 (1.1%) Stage 3 without specification of substage0 (0.0%)2 (0.0%) Stage 4: 15–29, severely decreased7 (0.7%)15 (0.3%) Stage 5: <15 OR treated by dialysis, kidney failure12 (1.3%)33 (0.7%) Stage unspecified121 (12.9%)606 (12.6%) No assessment of eGFR or diagnosis code in the year before index date59 (6.3%)645 (13.4%)‘Recent use’¶ drug classes Drug classes used, n (%) SGLT2i and fixed-dose combinations0 (0.0%)27 (0.6%) GLP-1 RA and fixed-dose combinations129 (13.8%)0 (0.0%) sMRA85 (9.1%)198 (4.1%) ACEi/ARB388 (41.5%)1858 (38.5%) No medication record458 (49.0%)2896 (60.1%)Clinical conditions associated with risk of CKD,* n (%) Glomerulonephritis (all causes)197 (21.1%)705 (14.6%) Renovascular disease4 (0.4%)4 (0.1%) Autoimmune disease2 (0.2%)11 (0.2%) Polycystic kidney disease0 (0.0%)2 (0.0%) Gout or hyperuricaemia,† (n%)109 (11.7%)515 (10.7%)eGFR** Count with baseline value, n (%)626 (67.0%)2908 (60.3%) Baseline value, mean (SD)90.6 (24.4)97.2 (22.5)UACR** Count with baseline value, n (%)115 (12.3%)1128 (23.4%) Median (Q1, Q3)42.5 (13, 261)40.1 (13, 184)Serum potassium** Count with baseline value, n (%)468 (50.1%)2333 (48.4%) Baseline value, mean (SD)4.3 (1.4)4.1 (0.6)Cardiovascular medications,* n (%) Thiazide-like diuretics24 (2.6%)108 (2.2%) Loop diuretics209 (22.4%)621 (12.9%) Potassium-sparing diuretics2 (0.2%)17 (0.4%) Beta blockers426 (45.6%)1670 (34.6%) Direct renin inhibitors0 (0.0%)0 (0.0%) Angiotensin receptor-neprilysin inhibitor0 (0.0%)0 (0.0%) Non-dihydropyridine calcium channel blockers488 (52.2%)2294 (47.6%) Other antihypertensives46 (4.9%)148 (3.1%) Statins590 (63.1%)2686 (55.7%) Anticoagulants206 (22.0%)607 (12.6%) Digoxin31 (3.3%)96 (2.0%) Nitrates and other vasodilators427 (45.7%)1493 (31.0%) Aspirin and other antiplatelet agents614 (65.7%)2880 (59.7%) Lipid-lowering drugs other than statins142 (15.2%)853 (17.7%)Macrovascular complications,† n (%) Coronary heart disease623 (66.6%)2460 (51.0%) Cerebrovascular disease257 (27.5%)989 (20.5%) Peripheral vascular disease537 (57.4%)2528 (52.4%)Cardiovascular risk factors,† n (%) Hypertension653 (69.8%)2791 (57.9%) Hypercholesterolaemia508 (54.3%)2092 (43.4%) Congestive heart failure6 (0.6%)9 (0.2%) Left ventricular hypertrophy0 (0.0%)2 (0.0%)*The lookback period for this variable used all available data (study days [-all available, 0]).†The lookback period for this variable used data in the 365 days before and including the index date (study days [−365, 0]).‡The BMI cut-off for obesity was based on the WHO definition for obesity.§The lookback period for this variable used data in the 180 days before and including the index date (study days [−180, 0]).¶The lookback period for this variable used data in the 90 days before the index date (study days [−90 to –1]).**The lookback period for this variable used data in the 90 days before and including the index date (study days [−90, 0]).BMI, body mass index; HbA1c, haemoglobin A1c; NA, not applicable; GLP-1 RA, glucagon like peptide-1 receptor agonist; SGLT2i, sodium-glucose cotransporter 2 inhibitor; sMRA, steroidal mineralocorticoid receptor antagonist; T2D, type 2 diabetes; UACR, urinary albumin-to-creatinine ratio. Patient profile of new users for study medications at baseline The lookback period for this variable used all available data (study days [-all available, 0]). The lookback period for this variable used data in the 365 days before and including the index date (study days [−365, 0]). The BMI cut-off for obesity was based on the WHO definition for obesity. The lookback period for this variable used data in the 180 days before and including the index date (study days [−180, 0]). The lookback period for this variable used data in the 90 days before the index date (study days [−90 to –1]). The lookback period for this variable used data in the 90 days before and including the index date (study days [−90, 0]). BMI, body mass index; HbA1c, haemoglobin A1c; NA, not applicable; GLP-1 RA, glucagon like peptide-1 receptor agonist; SGLT2i, sodium-glucose cotransporter 2 inhibitor; sMRA, steroidal mineralocorticoid receptor antagonist; T2D, type 2 diabetes; UACR, urinary albumin-to-creatinine ratio. T2D and CKD clinical characteristics are presented in table 1 and online supplemental table 2. Regarding the treatment duration, patients in the SGLT2i and GLP-1 RA cohorts had a mean T2D duration of 5 years (SD: 3 years) and a mean CKD duration of 3 years (SD: 3 years). Among the nine categories of T2D medications listed, metformin and its fixed-dose combinations were the most prescribed treatment (63.1%; 73.2%) during baseline lookback period, followed by alpha-glucosidase inhibitors (53.7%; 62.4%), and dipeptidyl peptidase-4 inhibitors (44.1%; 29.8%). Insulin use was recorded for 73% of patients in the SGLT2i cohort and 80% of the GLP-1 RA cohort during the 180-day lookback period. Around 40% of the patients were prescribed ACEi/ARB during the 90-day lookback period. The Diabetes Severity Complications Index (DSCI), which quantifies the presence and severity of diabetes complications,26 indicated that nephropathy, cardiovascular disease, and peripheral vascular disease as the most common complications for both SGLT2i and GLP-1 RA cohorts, with a mean DSCI score of 3.2 (SD: 2.0) and 2.6 (SD: 1.9), respectively. As for the severity of CKD, more than half of the patients were in Stage 1. There were 67% and 60% of the patients with recorded baseline eGFR values in the SGLT2i and GLP-1 RA cohort, with mean eGFR values of 90.6 (SD: 24.4) mL/min/1.73 m2 and 97.2 (SD: 22.5) mL/min/1.73 m2, respectively. As for the urinary albumin-to-creatinine ratio (UACR), only 12% of the SGLT2i cohort and 23% of the GLP-1 RA cohorts had baseline values retrieved. Around 50% of the patients were found with baseline serum potassium value, with a mean of 4 mmol/L for both medication cohorts. For the usage of cardiovascular medications within the 180-day baseline period, around 50%–65% of the patients in both medication cohorts were prescribed with aspirin and other antiplatelet agents, statins and non-dihydropyridine calcium channel blockers, jointly as the three cardiovascular medication classes with the highest proportions of patients (table 1). The disease burden of macrovascular and cardiovascular complications was heavy for both medication cohorts at baseline. Specifically, the most common comorbidities were hypertension, coronary heart disease and peripheral vascular disease. Healthcare resource utilization HRU during baseline and follow-up periods is shown in table 2. During the baseline period, over 96% of the patients in the SGLT2i and GLP-1 RA cohorts had at least one specialist visit, while around 25% had general practitioner (GP) visits. During the follow-up period (median: 1.4 months) for SGLT2i medication cohort, and the follow-up period (median: 2.3 months) for GLP-1 RA medication cohort, approximately 80% for patients in both cohorts had at least one specialist visit, while only 12%–19% of them had at least one GP visit. Among those with specialist visits, patients in the GLP-1 RA had a mean of 15 specialist visits, and patients in the SGLT2i cohort had a mean of 5 specialist visits during follow-up period. Table 2Healthcare resource utilization recorded during the baseline and follow-up period of new users for study medicationsBaseline*Follow-upSGLT2i cohort (n=935)GLP-1 RA cohort (n=4821)SGLT2i cohort (n=935)GLP-1 RA cohort (n=4821)General practitioner (or primary care) visits† n (%)231 (24.7%)1239 (25.7%)116 (12.4%)897 (18.6%) Mean (SD)3.8 (5.5)4.4 (5.7)2.6 (2.9)4.7 (7.8)Outpatient visits n (%)917 (98.1%)4690 (97.3%)780 (83.4%)4307 (89.3%) Mean (SD)13.1 (11.5)13.2 (11.5)5.8 (7.1)16.0 (22.5)Inpatient hospital admissions n (%)552 (59.0%)2915 (60.5%)97 (10.4%)690 (14.3%) Mean (SD)1.4 (0.7)1.2 (0.5)1.2 (0.6)1.3 (0.8)Specialist visits‡ n (%)899 (96.1%)4650 (96.5%)748 (80.0%)4258 (88.3%) Mean (SD)12.0 (10.4)12.0 (10.5)5.4 (6.7)15.2 (20.9)Emergency department visits n (%)235 (25.1%)897 (18.6%)83 (8.9%)615 (12.8%) Mean (SD)2.1 (2.2)2.0 (1.8)1.9 (2.4)2.2 (2.6)*The lookback period for this variable used data in the 180 days before and including the index date (study days [−180, 0]).†General practitioner (or primary care) visits were derived in the Tianjin database as i) For hospital visits (ie, primary, secondary and tertiary hospitals), records from the department of general practice were categorised as general practitioner visits and; ii) For community health service institution visits: all records were categorised as general practitioner visits.‡Specialist visits were derived in the Tianjin database for hospital visits (ie, primary, secondary and tertiary hospitals): records other than from the department of general practice were categorised as specialist visits.GLP-1 RA, glucagon-like peptide-1 receptor agonist; SGLT2i, sodium-glucose cotransporter 2 inhibitor. Healthcare resource utilization recorded during the baseline and follow-up period of new users for study medications The lookback period for this variable used data in the 180 days before and including the index date (study days [−180, 0]). General practitioner (or primary care) visits were derived in the Tianjin database as i) For hospital visits (ie, primary, secondary and tertiary hospitals), records from the department of general practice were categorised as general practitioner visits and; ii) For community health service institution visits: all records were categorised as general practitioner visits. Specialist visits were derived in the Tianjin database for hospital visits (ie, primary, secondary and tertiary hospitals): records other than from the department of general practice were categorised as specialist visits. GLP-1 RA, glucagon-like peptide-1 receptor agonist; SGLT2i, sodium-glucose cotransporter 2 inhibitor. Regarding laboratory measurements, the reported number and percentage of patients with available measurements for eGFR, UACR, serum potassium during follow-up and at specific timepoints are presented in table 3. Around 50% of the patients with baseline eGFR and serum potassium had measurements during follow-up periods. The number of patients with UACR at baseline and follow-up periods was consistently low. As for the laboratory measurements at pre-defined time points, few of them had measurements at 4 months and 1 year after index date. Table 3Kidney function laboratory values recorded during follow-up of new users for study medicationsSGLT2i cohortGLP-1 RA cohorteGFRUACRSerum PotassiumeGFRUACRSerum PotassiumNumber of patients in medication-specific analysis cohort935935935482148214821Patients with any measurement* during follow-up, n (%)326 (51.5%)26 (22.6%)240 (51.2%)1510 (51.7%)412 (36.5%)968 (41.5%) Number of measurements through end of follow-up Mean (SD)2.5 (2.4)1.3 (0.6)2.9 (3.1)2.4 (2.5)1.6 (1.1)2.1 (2.6) Median (Q1, Q3)2.0 (1.0, 3.0)1.0 (1.0, 1.0)2.0 (1.0, 3.0)2.0 (1.0, 3.0)1.0 (1.0, 2.0)1.0 (1.0, 2.0)Patients with measurements at predefined time points† during follow-up, n (%) 4 months88 (13.9%)7 (6.1%)29 (6.2%)685 (23.4%)92 (8.2%)202 (8.7%) 1 year2 (0.3%)0 (0.0%)0 (0.0%)237 (8.1%)44 (3.9%)99 (4.2%) 2 years0 (0.0%)0 (0.0%)0 (0.0%)12 (0.4%)1 (0.1%)2 (0.1%)*The denominator is the total count of patients in the medication-specific analysis cohort.†Defined as having a measurement within 90 days before or after the time point.eGFR, estimated glomerular filtration rate; GLP-1 RA, glucagon-like peptide-1 receptor agonist; Q1, Q3, first and third quartiles; SGLT2i, sodium-glucose cotransporter 2 inhibitor; UACR, urinary albumin-to-creatinine ratio. Kidney function laboratory values recorded during follow-up of new users for study medications The denominator is the total count of patients in the medication-specific analysis cohort. Defined as having a measurement within 90 days before or after the time point. eGFR, estimated glomerular filtration rate; GLP-1 RA, glucagon-like peptide-1 receptor agonist; Q1, Q3, first and third quartiles; SGLT2i, sodium-glucose cotransporter 2 inhibitor; UACR, urinary albumin-to-creatinine ratio. Clinical outcomes Table 4 shows the crude incidence rates for each outcome. In the SGLT2i cohort, the incidence rates with 95% CIs of kidney failure, acute coronary syndrome, stroke and atrial fibrillation were 3.1 (1.0, 7.3), 36.4 (27.0, 47.9), 1.9 (0.4, 5.5) and 4.0 (1.5, 8.6) per 100 person-years, respectively. In the GLP-1 RA cohort, the incidence rates with 95% CIs of kidney failure, acute coronary syndrome, stroke, heart failure and atrial fibrillation were 4.9 (3.9, 6.0), 12.4 (10.8, 14.2), 0.7 (0.4, 1.2), 0.1 (0.0, 0.3) and 1.9 (1.3, 2.7) per 100 person-years, respectively. Table 4Event counts, follow-up time and incidence rates of new users for study medicationsOutcomesSGLT2i cohortGLP-1 RA cohortKidney failure Patients in the analysis cohort9354821 Person-years of follow-up160.121849.86 Cases590 Incidence rate* and 95% CI3.1 (1.0, 7.3)4.9 (3.9, 6.0)Acute coronary syndrome Patients in the analysis cohort8154450 Person-years of follow-up137.531664.00 Cases50206 Incidence rate* and 95% CI36.4 (27.0, 47.9)12.4 (10.8, 14.2)Stroke Patients in the analysis cohort9244786 Person-years of follow-up158.311832.68 Cases313 Incidence rate* and 95% CI1.9 (0.4, 5.5)0.7 (0.4, 1.2)Heart failure Patients in the analysis cohort9144787 Person-years of follow-up157.151839.92 Cases01 Incidence rate* and 95% CI0.00 (NA)0.1 (0.0, 0.3)Atrial fibrillation Patients in the analysis cohort8774638 Person-years of follow-up151.81774.18 Cases634 Incidence rate* and 95% CI4.0 (1.5, 8.6)1.9 (1.3, 2.7)Crude incidence rates (IRs) were calculated for each outcome as the number of events (E) occurring during the person-time at risk (Y) divided by the total person-time at risk. IRs were reported as point estimates and 95% CIs. Overall, the robustness of the study results was reflected in the unchanged trend of sensitivity analyses (ie, by adjusting the grace period and stockpiling) compared with the main analyses results.*Incidence rates are per 100 person-years.GLP-1 RA, glucagon-like peptide-1 receptor agonist; NA, not applicable; SGLT2i, sodium-glucose cotransporter 2 inhibitor. Event counts, follow-up time and incidence rates of new users for study medications Crude incidence rates (IRs) were calculated for each outcome as the number of events (E) occurring during the person-time at risk (Y) divided by the total person-time at risk. IRs were reported as point estimates and 95% CIs. Overall, the robustness of the study results was reflected in the unchanged trend of sensitivity analyses (ie, by adjusting the grace period and stockpiling) compared with the main analyses results. Incidence rates are per 100 person-years. GLP-1 RA, glucagon-like peptide-1 receptor agonist; NA, not applicable; SGLT2i, sodium-glucose cotransporter 2 inhibitor. Discussion This study is by far the largest observational analysis in China examining SGLT2i/GLP-1 RA users with T2D and CKD, using OMOP-standardized data to ensure reliability. It fills a key evidence gap by providing real-world data on patients in China with a high burden of T2D and CKD. Of all patients with study drug prescriptions identified in the Tianjin database, 0.36% of patients from the SGLT2i cohort and 4.01% of patients from the GLP-1 RA cohort met the study’s inclusion and exclusion criteria and were included in the analysis. The primary reason for exclusion was the potential index dates of study drugs reported outside the study period of 2012–2019. More than 80% of the potential index dates were removed due to late drug launches in China. Another reason for eligible patients dropping out was that our study included only new users of the two medication cohorts, with a 12-month washout period, which excluded patients with any drug used in the same medication cohorts to mitigate misclassification. For the patients included in this study, although no formal comparison was conducted between the two medication cohorts, similar trends in their baseline characteristics were observed. The range of the mean age at index date was 55 to 59, which is consistent with the range reported globally in studies among patients with T2D and CKD.27 No gender distribution disparity was found between this study and other global studies from Europe, Asia and North America.27 Patients across the two medication cohorts exhibited a prolonged duration of T2D relative to CKD, which aligns with the literature identifying CKD as a comorbidity in individuals with T2D.2 28 As suggested in the Chinese Expert consensus on clinical management for chronic kidney disease in diabetes mellitus in China,29 eGFR and UACR should be jointly considered when determining CKD severity. However, in Tianjin database, because of the high missingness of baseline UACR in both cohorts, more than 50% of the patients in the SGLT2i and GLP-1 RA cohorts were classified as stage 1 CKD based on CKD diagnosis and eGFR value. The mean eGFR was 90.6 and 97.2 mL/min/1.73 m2 at baseline for SGLT2i and GLP-1 RA cohorts, respectively. This trend should be mainly affected by the clinical prescription practice during study period. China expert consensus for the prevention and treatment of diabetic kidney disease (2014 edition) only recommended GLP-1 RA for the treatment of CKD stage 1–3, and SGLT2i was not clearly recommended. For the updated guidance in 2024,30 both SGLT2i and GLP-1 RA are applicable to CKD stage 4 patients. Comorbidities and comedications profile of the study population align with the findings of patients from other studies.3134 The high proportions of patients with hypertension echo with that the blood pressure management is essential for reducing CKD progression as agreed by international guidelines.28 The study indicated that both medication cohorts had a greater number of specialist visits than general practitioner visits for HRU. This trend was also verified by the kidney function measurements captured, including eGFR, UACR and serum potassium. There was only a small amount of kidney function measurement results retrieved for patients during the study follow-up period, which might also indicate that regular GP visits after diagnosis were not the standard clinical practice during the study period. In general, this trend can be attributed to the healthcare-seeking pattern with greater reliance on specialist visits during the study period in China. Specifically, the General Office of the State Council of China published a guideline to urge the establishment of a tiered medical system in 2015,35 and the General Office of Tianjin Municipal People’s Government published the Long-Term Plan for the prevention and treatment of Chronic Diseases in Tianjin (2017–2025) in 201736 to emphasize that the implementation of the tiered diagnosis and treatment system should be one of the major goals for chronic disease management in Tianjin. For clinical outcomes, our study results are comparable with similar studies conducted for the progression of kidney outcomes in Chinese population. A study using Hong Kong database,37 with 80% of patients with eGFR ≥60 mL/min/1.73 m2 from SGLT2i cohort and 69% of patients with eGFR ≥60 mL/min/1.73 m2 from GLP-1 RA cohort, reported the cumulative incidence of composite kidney outcomes as 3.9 per 100 person-years, and 5.2 per 100 person-years for SGLT2i and GLP-1 RA cohort, respectively. In our study, with slightly lower eGFR baseline characteristics of the study population, the cumulative incidence of kidney failure was 3.1 per 100 person-years and 4.9 per 100 person-years for SGLT2i and GLP-1 RA cohorts. Acute coronary syndrome displayed a high incidence rate in both medication cohorts. Several factors may account for the increased incidence rate. During the baseline period, both cohorts demonstrated elevated rates of cardiovascular medication use and a high prevalence of cardiovascular risk factors, including hypertension and hypercholesterolemia. In addition, cardiorenal syndrome is commonly known as representing the confluence of heart-kidney interactions such as the hemodynamic crosstalk between the failing heart and the response of the kidney and vice versa.38 It is widely acknowledged that cardiovascular disease and CKD have interrelated mechanisms that exacerbate each other’s progression.39 Additional factors may include the possibility of biased results due to potential misclassification of the respective outcomes and limited follow-up duration. Although the study utilized data from existing electronic medical records and applied stringent quality control and standardization procedures, there remains the possibility of inaccuracies in diagnosis dates and records. Such discrepancies may result in misclassification of clinical outcomes, particularly among records associated with a single hospitalization episode. The short follow-up duration (with end date of the study period set on 31 December 2019 to avoid COVID-19 interruptions) also limits the interpretability of these incidence estimates for long-term outcomes. From a different perspective, the results also highlight the importance of close monitoring, early intervention and follow-up care for patients with T2D and CKD, especially for those with heavy comorbidities. In a consensus report by the ADA and KDIGO,7 comprehensive, holistic, patient-centered medical care was recommended, in which patients may need to be seen frequently to identify and implement multiple therapies to avoid therapeutic inertia. Despite the availability of established treatments such as SGLT2i and GLP-1 RA for patients with T2D and CKD investigated in the current study, our findings demonstrate that a residual risk of cardiovascular events and kidney failure persists. This underscores the need for a more comprehensive and integrated approach to disease management, including efficient intervention and proactive monitoring. Recent advancements such as the introduction of nsMRA, e.g., finerenone, offer a promising avenue for combination therapy. The current Chinese guideline recommends finerenone for patients with T2D and CKD (with albuminuria).30 Furthermore, a previous study showed that the combination treatment of SGLT2i, GLP-1 RA and finerenone has the potential to afford substantial gains in cardiovascular, kidney and overall survival.40 The CONFIDENCE trial indicated definite data on the effects of combining SGLT2i and finerenone in slowing disease progression and offering long-term benefits.41 These findings emphasize the need for clinicians to consider a broader treatment spectrum to optimize clinical outcomes. Our study has several limitations. First, the Tianjin database is a regional EHR database, with information extracted from covered hospitals in Tianjin mainly for patient care. Consequently, the database did not include patients’ medical information outside Tianjin, which may undermine the generalizability of the study results to broader population in China. However, the observed trends regarding patient characteristics were generally comparable with previous evidence. Second, the study’s follow-up period was generally limited due to patients discontinuing their medication classes. Further long-term studies are needed to confirm these results. The timing of drug launches—SGLT2i after 2017 and most GLP-1 RAs after 2018 (except liraglutide, launched in 2011)—may have influenced the findings. Also, it is unfeasible to capture all actual intake of the medication based solely on the prescription information in medical records. Given the possibility of multiple refills of the medication for prescriptions from hospitals or clinics not included in the database, there may be an underestimation of the medication usage. Third, our study was descriptive in nature, and there were no formal comparisons between the two medication cohorts, and hence no comparative evidence was generated regarding the effectiveness of study medications. Finally, since the end date of the study period was set on 31 December 2019 to avoid COVID-19 interruptions, new treatment options launched after 2019, such as nsMRA, were not included in the current study. Conclusion The study findings fill the knowledge gap regarding patient characteristics among individuals with T2D and CKD who had continuous usage of SGLT2i or GLP-1 RA in real-world settings in China. The results highlight the residual risk of severe clinical outcomes for these patients and support tailored disease management strategies, including regular monitoring of kidney function and early initiation of cardio-renal protective agents. Additionally, the study provides a solid foundation for future research aimed at evaluating the clinical outcomes of new therapeutic options for T2D and CKD. Further studies in other regions of China with extended follow-up would help to validate the observed trends and evaluate new treatments, including nsMRA, offering a broader understanding of patients’ needs for T2D and CKD treatment in China. Supplementary material