ABSTRACT
Aims/Introduction
Sodium–glucose cotransporter 2 inhibitors (SGLT2i) reduce the risk of acute kidney injury (AKI). However, an initial decline in estimated glomerular filtration rate (eGFR) may occur after initiation, and factors associated with sustained early eGFR decline remain unclear. We evaluated the temporal distribution of AKI‐related hospitalizations in older adults using claims data and, in an independent single‐center cohort, factors associated with sustained early eGFR decline.
Materials and Methods
A Japanese nationwide claims database identified patients aged ≥75 years emergently hospitalized with AKI after SGLT2i initiation. Intervals were categorized as 1–30, 31–90, and 91–180 days. Baseline characteristics were compared across groups, followed by adjusted comparisons of the 1–30‐day group with each later group. An independent single‐center analysis without age restriction evaluated early eGFR changes and associated factors as indirect complementary evidence.
Results
Among 190 patients, 59 (31.1%) were hospitalized within 1–30 days, 67 (35.3%) within 31–90 days, and 64 (33.7%) within 91–180 days. After inverse‐probability‐of‐treatment weighting, all propensity‐score covariates were balanced, and no additionally adjusted medication‐related factor was associated with 1‐ to 30‐day‐group assignment. Sustained early eGFR decline was associated with older age and use of one or two triple‐whammy (TW)‐related drug classes.
Conclusions
A substantial proportion of AKI‐related hospitalizations occurred within 30 days after SGLT2i initiation. In the single‐center analysis, sustained early eGFR decline was associated with older age and TW‐related drug use. These hypothesis‐generating findings require cautious interpretation because the case‐only design precludes estimation of early AKI risk.
Keywords: acute kidney injury, hospitalization, sodium‐glucose transporter 2 inhibitors
Among 190 older adults who experienced AKI‐related hospitalization after SGLT2 inhibitor initiation, 31.1% were hospitalized within 30 days. In an independent single‐center cohort, sustained early eGFR decline was associated with older age and use of one or two triple‐whammy‐related drug classes.
INTRODUCTION
Sodium–glucose cotransporter 2 inhibitors (SGLT2is) are oral antidiabetic agents that lower blood glucose independently of insulin by inhibiting glucose reabsorption in the proximal renal tubules and promoting urinary glucose excretion. 1 They also reduce intravascular volume through osmotic diuresis and lowering of intraglomerular pressure. Their clinical significance has expanded because they reduce heart‐failure events and slow chronic kidney disease (CKD) progression. 2 , 3 , 4 , 5 , 6 , 7 , 8
However, a transient decline in estimated glomerular filtration rate (eGFR), referred to as the “initial dip,” is well recognized in the early phase after SGLT2i initiation. 9 , 10 , 11 , 12 In most cases, this decline is reversible, with recovery occurring within several weeks to months, and contributes to long‐term renoprotective effects, as demonstrated in large‐scale clinical trials and meta‐analyses. 9 , 10 , 11 , 12 All six SGLT2i agents approved in Japan were associated with an initial dip shortly after initiation, followed by eGFR recovery at 6–12 months. 13
Most studies have focused on mid‐ to long‐term renal outcomes, with limited attention to early renal‐function changes following SGLT2i initiation. 9 , 10 , 11 , 12 Although the initial dip is generally considered a physiological response, its clinical significance remains unclear. Furthermore, this early decline in renal function may progress to levels meeting the clinical definition of acute kidney injury (AKI) and require hospitalization. Careful monitoring immediately after initiation is warranted, especially among older adults and those receiving multiple concomitant medications. 14
Although meta‐analyses have consistently shown that SGLT2is reduce AKI risk, an increased incidence of dehydration‐related adverse events has also been reported. 15 These findings highlight the difficulty of interpreting early renal function changes after initiation, particularly in vulnerable populations.
Therefore, we analyzed the temporal distribution of AKI‐related hospitalizations among older adults hospitalized with AKI after SGLT2i initiation using a large‐scale administrative claims database in Japan. We also conducted a complementary analysis using single‐center clinical data to evaluate early changes in eGFR after initiation and to examine the clinical characteristics of patients with sustained early renal function decline.
The aim of this study was to describe the temporal distribution of AKI‐related hospitalizations among identified cases after SGLT2i initiation and to explore, in a complementary single‐center analysis, patient characteristics associated with sustained early eGFR decline.
MATERIALS AND METHODS
Study design and data sources
The study comprised a retrospective case‐only analysis using a nationwide administrative claims database and an independent, complementary analysis using single‐center clinical data. The single‐center analysis explored early eGFR changes and associated clinical factors not available in the claims database and was intended to provide indirect complementary evidence rather than a mechanistic validation of the claims‐based findings.
For the claims‐database analysis, we used the Medical Data Vision (MDV) database (Medical Data Vision Co., Ltd., Tokyo, Japan). The MDV database collects claims and Diagnosis Procedure Combination data from over 540 hospitals in Japan and contains approximately 50 million patient records, including diagnoses, prescriptions, laboratory tests, procedures, surgeries, and hospitalization histories (as of February 2024). We analyzed anonymized data for 1,831,331 patients aged ≥75 years who were emergently hospitalized between September 1, 2018, and November 30, 2023. In the MDV database, hospitalization type is classified as “planned/other” or “emergency admission,” and the latter was defined as emergency hospitalization. Longitudinal patient data can be traced back to April 2008.
Ethics approval and informed consent
This study was conducted in accordance with the principles of the Declaration of Helsinki and the Ethical Guidelines for Medical and Biological Research Involving Human Subjects in Japan. The claims‐database analysis was approved by the Ethics Committee of Hokkaido University of Science (approval No. 25‐25). As fully anonymized secondary data were used, the requirement for informed consent was waived. 16
The single‐center clinical analysis was approved by the Ethics Committee of JA Hokkaido Koseiren Sapporo Kosei Hospital (approval No. 869). As this was a retrospective observational study using medical records, informed consent was obtained using an opt‐out approach.
Claims‐database analysis
Among 1,831,331 patients aged ≥75 years emergently hospitalized between September 1, 2018, and November 30, 2023, we identified those whose primary diagnosis at admission was AKI and who had a history of type 2 diabetes mellitus. Among these, 567 patients with a history of SGLT2i prescription were selected as candidate cases.
Diagnoses were defined using the International Classification of Diseases, 10th Revision (ICD‐10), with AKI defined as code N17 and type 2 diabetes mellitus as code E11. SGLT2is included ipragliflozin, empagliflozin, canagliflozin, dapagliflozin, tofogliflozin, and luseogliflozin. The initiation date was defined as the first prescription date recorded in the MDV database. Patients who had initiated SGLT2is ≥181 days before hospitalization were excluded to restrict the observation window. To minimize confounding by contrast‐induced AKI, 377 patients who underwent contrast‐enhanced imaging within 30 days before admission were excluded. As a result, the data of 190 unique patients were included in the final analysis (Figure 1). Patients were categorized into three groups according to the interval from SGLT2i initiation to hospitalization: 1–30 days, 31–90 days, and 91–180 days.
Figure 1.
Flow diagram of patient selection. Flowchart illustrating the cohort selection process. Patients aged ≥75 years hospitalized for AKI were identified from the MDV database in Japan, and those with a diagnosis of T2DM were included. The date of AKI‐related emergency hospitalization was defined as the index date. Patients who initiated SGLT2is ≥181 days before hospitalization or underwent contrast‐enhanced imaging within 30 days prior to admission were excluded. After applying these criteria, 190 patients were included in the final analytic cohort. Patients were classified into three groups according to the interval from SGLT2i initiation to hospitalization: 1–30, 31–90, and 91–180 days. Diagnoses were defined using the International Classification of Diseases, 10th Revision (ICD‐10) (N17 for AKI and E11 for T2DM). AKI, acute kidney injury; SGLT2i, sodium–glucose cotransporter 2 inhibitors; T2DM, type 2 diabetes mellitus; MDV, Medical Data Vision.
Baseline characteristics
Baseline characteristics assessed at admission included age, sex, cognitive function status, 17 long‐term care level, home healthcare use prior to admission, comorbidities, and use of antidiabetic medications and suspected AKI‐related drugs. Cognitive function was classified according to the Ministry of Health, Labour and Welfare categories for the degree of independence in daily living for older adults with dementia (ranks I–IV and M). 17 Patients without dementia were categorized as “No impairment.” Long‐term care level was classified as “Independent,” “Support needed,” “Long‐term care,” or “Pending.”
Comorbidities included CKD, diabetic nephropathy, heart failure, hypertension, ischemic heart disease, and malignancy; the corresponding ICD‐10 codes are listed in Table S1.
Antidiabetic medications were categorized as biguanides, sulfonylureas, dipeptidyl peptidase‐4 inhibitors, glucagon‐like peptide‐1 receptor agonists, thiazolidinediones, alpha‐glucosidase inhibitors, imeglimin, and insulin; the drugs are listed in Table S2.
Suspected AKI‐related medications were identified based on the Clinical Practice Guideline for Drug‐Induced Kidney Injury (2016). 18 These included nonsteroidal anti‐inflammatory drugs (NSAIDs), diuretics (loop and thiazide diuretics), mineralocorticoid receptor antagonists (MRAs), vasopressin V2 receptor antagonists, renin–angiotensin–aldosterone system inhibitors (RAASi), calcium channel blockers (CCBs), β‐blockers, proton pump inhibitors (PPIs), allopurinol, calcineurin inhibitors, platinum‐based anticancer agents, alkylating agents, cephalosporins, penicillins, quinolones, tetracyclines, macrolides, carbapenems, monobactams, aminoglycosides, sulfonamides, vancomycin, azole antifungals, and amphotericin B. The full list of drugs is provided in Table S3.
Exposure was defined as a prescription within 1–30 days before admission for NSAIDs and antibiotics and 1–90 days for all other drugs.
Triple whammy (TW) exposure was defined as concurrent NSAID, diuretic, and RAASi use within 30 days before emergency hospitalization, consistent with its reported association with AKI risk. 19 , 20
Because the MDV database uses administrative claims data, unrecorded diagnoses or prescriptions were treated as absent.
This retrospective observational study included all eligible patients identified from the MDV database; therefore, no prior sample size calculation was performed.
Statistical analysis (claims database)
Patients were classified into three groups according to the interval from SGLT2i initiation to AKI‐related hospitalization: 1–30 days, 31–90 days, and 91–180 days. The time intervals were defined to distinguish the early post‐initiation period, during which the initial eGFR dip is generally observed, from subsequent periods. 9 , 10 , 11 , 12 Baseline characteristics were compared across the three groups using the Kruskal–Wallis test for continuous variables and the χ2 test or Fisher's exact test, as appropriate, for categorical variables.
To adjust for differences in baseline characteristics between groups, inverse probability of treatment weighting (IPTW) was applied separately to the comparison between the 1–30‐day and 31–90‐day groups and to the comparison between the 1–30‐day and 91–180‐day groups. 21 IPTW was used as a supplementary approach to balance baseline characteristics between the time‐based groups and not to estimate causal effects.
For each pairwise comparison, propensity scores for assignment to the 1–30‐day group were estimated using a logistic regression model, which included age, sex, cognitive function status, long‐term care level, home healthcare use prior to admission, comorbidities, and antidiabetic medications as covariates. Suspected AKI‐related medications were not included in the propensity‐score models because of the small number of patients exposed to each drug class. Baseline serum creatinine and eGFR values before SGLT2i initiation were not consistently available in the MDV database and therefore could not be included in the propensity‐score models or weighted logistic regression models described below.
Based on the estimated propensity scores, stabilized inverse‐probability‐of‐treatment weights were applied to balance baseline characteristics between groups. Covariate balance after IPTW was assessed using the absolute standardized mean difference (SMD), with an absolute SMD <0.1 indicating adequate balance. 22 For covariates with residual imbalance after IPTW, defined as an absolute SMD ≥0.1, additional adjustment was performed using weighted logistic‐regression models. Separate models were constructed for each pairwise comparison, and odds ratios (ORs) and 95% confidence intervals (CIs) for assignment to the 1–30‐day group were estimated.
All statistical analyses were performed using JMP® Student Edition 19 (SAS Institute Inc). A two‐sided P‐value <0.05 was considered statistically significant. The claims‐based analysis was a case‐only analysis, including only patients who experienced AKI‐related hospitalization after SGLT2i initiation. It therefore described the timing of hospitalization among the identified cases and did not estimate the incidence or risk of AKI among the overall population of SGLT2i users.
Single‐center clinical analysis
Study population
Patients who newly initiated SGLT2is at JA Hokkaido Koseiren Sapporo Kosei Hospital between July 2020 and August 2025 were included. The index date was defined as the date of SGLT2i initiation, and patients with available eGFR measurements at baseline (day 0), 30 days, and 60 days after initiation were included.
No exclusion criteria based on age were applied to capture real‐world renal function changes across diverse patients. The single‐center cohort comprised a separate, independent population and was analyzed to provide indirect complementary evidence on early eGFR changes rather than mechanistic validation of the claims‐based temporal findings.
Clinical characteristics
Baseline characteristics included age, sex, functional status, 23 level of care dependency, home‐healthcare use, comorbidities, and use of antidiabetic medications and suspected AKI‐related drugs. All variables, except for functional status and TW exposure, were defined in the same manner as in the claims‐database analysis. As no cases met the criteria for TW exposure in the single‐center dataset, the number of TW‐related drug classes used was evaluated as a surrogate indicator. Functional status was assessed using the Japanese scale for independence in daily living for elderly individuals with physical disabilities (bedriddenness scale), 23 which classifies patients as J, A, B, and C, where rank J indicates near independence and rank C indicates a condition requiring constant care.
Based on changes in eGFR after SGLT2i initiation, patients were classified into sustained‐decline and non‐sustained‐decline groups. The sustained‐decline group included patients with ≥10% decline in eGFR from baseline at both 30 and 60 days after SGLT2i initiation. The non‐sustained‐decline group comprised patients with a transient dip, delayed decline, or no dip. A transient dip was defined as ≥10% decline at 30 days but not at 60 days, delayed decline as ≥10% decline at 60 days but not at 30 days, and no dip as the absence of ≥10% decline at either time point. The 10% threshold was based on prior SGLT2i initial‐dip studies and was considered a clinically meaningful indicator of early renal function decline. 24 , 25 Assessment at 30 days was used to identify early decline, whereas assessment at 60 days was used to determine whether the decline persisted.
Statistical analysis (clinical data)
Baseline characteristics and clinical variables were compared between the sustained‐decline and non‐sustained‐decline groups. Continuous variables are presented as the mean ± standard deviation and were compared using the Wilcoxon rank‐sum test. Categorical variables are presented as n (%) and were compared using the χ2 test or Fisher's exact test, as appropriate.
To examine factors associated with sustained eGFR decline, multivariable logistic regression analysis was performed including age and number of TW‐related drug classes, which differed between the groups in the univariate analyses, together with baseline eGFR (day 0) as a clinically relevant adjustment variable. Age and baseline eGFR were treated as continuous variables, and ORs were expressed per 5‐year increase and per 10 mL/min/1.73 m2 increase, respectively. The number of TW‐related drug classes was treated as a categorical variable with three levels: 0, 1, and 2 classes. Results are presented as ORs with 95% CIs.
As a sensitivity analysis, patients with a transient dip or delayed decline were excluded, and the same multivariable logistic‐regression model was applied to the sustained‐decline and no‐dip groups.
All analyses were performed using JMP® Student Edition 19 (SAS Institute Inc). A two‐sided P‐value <0.05 was considered statistically significant.
RESULTS
Claims‐database analysis
Temporal pattern of AKI‐related hospitalizations
The distribution of SGLT2i agents included in this analysis included ipragliflozin (n = 5), empagliflozin (n = 80), canagliflozin (n = 15), dapagliflozin (n = 84), tofogliflozin (n = 3), and luseogliflozin (n = 3). The interval from SGLT2i initiation to AKI‐related emergency hospitalization among the identified cases is shown in Figure 2. Of the 190 cases, 59 (31.1%) occurred within 1–30 days, 67 (35.3%) within 31–90 days, and 64 (33.7%) within 91–180 days after initiation. Among the 59 cases in the 1–30‐day group, 49 occurred during days 11–30. Because the intervals differed in duration and the number of patients at risk during each interval was unavailable, period‐specific incidence rates and relative risks could not be estimated.
Figure 2.
Temporal distribution of AKI‐related hospitalizations after initiation of SGLT2is. Histogram showing the number of AKI‐related hospitalizations according to 10‐day intervals between SGLT2i initiation and hospitalization. The early period (1–30 days) is highlighted in black. The figure descriptively presents the timing of hospitalization among the identified cases; no statistical comparison across time intervals was performed. AKI, acute kidney injury; SGLT2i, sodium–glucose cotransporter 2 inhibitors.
Baseline characteristics and adjusted comparisons between groups
Baseline characteristics were compared across the 1–30‐day group (n = 59), 31–90‐day group (n = 67), and 91–180‐day group (n = 64) (Table 1). Only medications with observed use were included in the table. The prevalence of diabetic nephropathy significantly differed among the three groups (13.6%, 13.4%, and 28.1%, respectively; P < 0.05), whereas no significant differences were observed in the other baseline characteristics.
Table 1.
Baseline characteristics of patients with AKI‐related emergency hospitalization according to the interval from SGLT2i initiation to hospitalization in the claims‐database cohort
| 1–30‐day group (n = 59) | 31–90‐day group (n = 67) | 91–180‐day group (n = 64) | P‐value | |
|---|---|---|---|---|
| Demographics and functional status | ||||
| Age (years) | 82.1 ± 4.8 | 83.0 ± 5.1 | 83.0 ± 5.3 | 0.53 † |
| Male | 44 (74.6) | 46 (68.7) | 42 (65.6) | 0.55 ‡ |
| Cognitive ADL independence level (No impairment/I–II/III–IV・M) | 36/15/8 | 42/15/10 | 37/19/8 | 0.92 ‡ |
| Long‐term care level (Independent/Support needed/Long‐term care/Pending) | 28/6/13/12 | 37/7/14/9 | 31/11/14/8 | 0.74 ‡ |
| Home healthcare before admission | 5 (8.5) | 6 (9.0) | 8 (12.5) | 0.71 ‡ |
| Comorbidities | ||||
| CKD | 18 (30.5) | 24 (35.8) | 26 (40.6) | 0.50 ‡ |
| DN | 8 (13.6) | 9 (13.4) | 18 (28.1) | < 0.05 ‡ |
| HF | 54 (91.5) | 57 (85.1) | 54 (84.4) | 0.44 ‡ |
| HTN | 54 (91.5) | 57 (85.1) | 59 (92.2) | 0.34 ‡ |
| IHD | 49 (83.1) | 46 (68.7) | 52 (81.3) | 0.10 ‡ |
| Malignancy | 42 (71.2) | 43 (64.2) | 43 (67.2) | 0.70 ‡ |
| Antidiabetic medications | ||||
| Biguanides | 11 (18.6) | 7 (10.4) | 10 (15.6) | 0.42 ‡ |
| DPP‐4i | 21 (35.6) | 18 (26.9) | 22 (34.4) | 0.52 ‡ |
| GLP‐1RA | 3 (5.1) | 4 (6.0) | 5 (7.8) | 0.82 ‡ |
| TZDs | 4 (6.8) | 4 (6.0) | 5 (7.8) | 0.92 ‡ |
| α‐GI | 6 (10.2) | 4 (6.0) | 8 (12.5) | 0.43 ‡ |
| Insulin preparations | 18 (30.5) | 25 (37.3) | 20 (31.3) | 0.66 ‡ |
| Suspected AKI‐related medications | ||||
| NSAIDs | 11 (18.6) | 8 (11.9) | 11 (17.2) | 0.55 ‡ |
| Loop diuretics | 47 (79.7) | 52 (77.6) | 50 (78.1) | 0.96 ‡ |
| Thiazide diuretics | 8 (13.6) | 13 (19.4) | 11 (17.2) | 0.68 ‡ |
| MRAs | 31 (52.5) | 31 (46.3) | 27 (42.2) | 0.51 ‡ |
| RAASi | 48 (81.4) | 51 (76.1) | 53 (82.8) | 0.60 ‡ |
| CCBs | 30 (50.8) | 35 (52.2) | 35 (54.7) | 0.91 ‡ |
| β‐blockers | 35 (59.3) | 45 (67.2) | 42 (65.6) | 0.63 ‡ |
| PPIs | 24 (40.7) | 35 (52.2) | 26 (40.6) | 0.31 ‡ |
| Allopurinol | 4 (6.8) | 5 (7.5) | 3 (4.7) | 0.81 § |
| Cephalosporins | 8 (13.6) | 4 (6.0) | 5 (7.8) | 0.31 ‡ |
| Penicillins | 5 (8.5) | 2 (3.0) | 1 (1.6) | 0.17 § |
| Quinolone antibiotics | 2 (3.4) | 4 (6.0) | 1 (1.6) | 0.45 § |
| TW exposure | 7 (11.9) | 3 (4.5) | 6 (9.4) | 0.31 ‡ |
Baseline characteristics of patients aged ≥75 years emergently hospitalized with AKI, stratified according to the interval from SGLT2i initiation to hospitalization (1–30, 31–90, and 91–180 days). Continuous variables are presented as mean ± standard deviation, and categorical variables as n (%). P‐values represent overall comparisons among the three groups. Cognitive ADL independence level was classified according to the Japanese dementia independence scale, and long‐term care level was classified according to the Japanese long‐term care insurance system. ADL, activities of daily living; AKI, acute kidney injury; MDV, Medical Data Vision; SMD, standardized mean difference; IPTW, inverse probability of treatment weighting; SGLT2i, sodium–glucose cotransporter 2 inhibitors; CKD, chronic kidney disease; DN, diabetic nephropathy; HF, heart failure; HTN, hypertension; IHD, ischemic heart disease; DPP‐4i, dipeptidyl peptidase‐4 inhibitor; GLP‐1RA, glucagon‐like peptide‐1 receptor agonist; TZDs, thiazolidinediones; α‐GI, alpha‐glucosidase inhibitor; NSAIDs, nonsteroidal anti‐inflammatory drugs; MRAs, mineralocorticoid receptor antagonists; RAASi, renin–angiotensin–aldosterone system inhibitors; CCBs, calcium channel blockers; PPIs, proton pump inhibitors; TW, triple whammy.
Kruskal–Wallis test.
χ2 test.
Fisher's exact test.
Before weighting, multiple variables related to demographics and functional status, comorbidities, and antidiabetic medications showed absolute SMDs ≥0.1 in both the comparison between the 1–30‐day and 31–90‐day groups and that between the 1–30‐day and 91–180‐day groups. After applying IPTW, all covariates included in the propensity‐score models achieved adequate balance, with absolute SMDs <0.1 in both comparisons (Figure 3a,c).
Figure 3.
Absolute SMDs before and after IPTW. This Love plot displays the absolute SMDs of covariates before (●) and after (■) applying IPTW. Panels a and b compare the 1–30‐day group with the 31–90‐day group, whereas Panels c and d compare the 1–30‐day group with the 91–180‐day group. Panels a and c show demographics and functional status, comorbidities, and antidiabetic medications, whereas Panels b and d show suspected AKI‐related medications. The vertical dashed line represents the threshold of an SMD of 0.1, with values below 0.1 indicating adequate balance. AKI, acute kidney injury; SMD, standardized mean difference; IPTW, inverse probability of treatment weighting; CKD, chronic kidney disease; DN, diabetic nephropathy; HF, heart failure; HTN, hypertension; IHD, ischemic heart disease; DPP‐4i, dipeptidyl peptidase‐4 inhibitor; GLP‐1RA, glucagon‐like peptide‐1 receptor agonist; TZDs, thiazolidinediones; α‐GI, alpha‐glucosidase inhibitor; NSAIDs, nonsteroidal anti‐inflammatory drugs; MRAs, mineralocorticoid receptor antagonists; RAASi, renin–angiotensin–aldosterone system inhibitors; CCBs, calcium channel blockers; PPIs, proton pump inhibitors; TW, triple whammy.
Because suspected AKI‐related medications were excluded from the propensity‐score models, residual imbalance remained for some variables after IPTW. In the comparison between the 1–30‐day and 31–90‐day groups, residual imbalance was observed for RAASi, PPIs, cephalosporins, penicillins, and TW exposure (Figure 3b). In the comparison between the 1–30‐day and 91–180‐day groups, residual imbalance was observed for NSAIDs, MRAs, PPIs, cephalosporins, penicillins, and quinolones (Figure 3d). These variables were therefore additionally included in the respective weighted logistic‐regression models. In both comparisons, none of these variables was significantly associated with membership in the 1–30‐day group (Figure 4).
Figure 4.
Factors associated with 1–30‐day group membership after IPTW. ORs with 95% CIs are shown for each variable after IPTW. Panel a compares the 1–30‐day group with the 31–90‐day group, whereas Panel b compares the 1–30‐day group with the 91–180‐day group. In Panel a, variables that remained imbalanced after IPTW—RAASi, PPIs, cephalosporins, penicillins, and TW exposure—were additionally included in the outcome model. In Panel b, NSAIDs, MRAs, PPIs, cephalosporins, penicillins, and quinolone antibiotics were additionally included. None of these variables showed a significant association with membership in the 1–30‐day group. The vertical dashed line represents an OR of 1.0. Arrows indicate that the upper limit of the 95% CI extends beyond the displayed range. AKI, acute kidney injury; OR, odds ratio; CI, confidence interval; IPTW, inverse probability of treatment weighting; NSAIDs, nonsteroidal anti‐inflammatory drugs; MRAs, mineralocorticoid receptor antagonists; RAASi, renin–angiotensin–aldosterone system inhibitors; PPIs, proton pump inhibitors; TW, triple whammy.
Single‐center clinical analysis
Early eGFR changes and clinical correlates
In the single‐center cohort of 114 patients who newly initiated SGLT2is, patients were classified into four groups according to changes in eGFR at 30 and 60 days after initiation: sustained decline (n = 32), transient dip (n = 14), delayed decline (n = 14), and no dip (n = 54). In the primary analysis, the transient‐dip, delayed‐decline, and no‐dip groups were combined into the non‐sustained‐decline group (n = 82). In the overall cohort, mean eGFR was 61.2 ± 2.7 mL/min/1.73 m2 at baseline, 60.7 ± 2.6 mL/min/1.73 m2 at 30 days, and 61.4 ± 2.8 mL/min/1.73 m2 at 60 days (values are presented as mean ± standard error), indicating minimal overall change during the observation period. Conversely, stratification according to early eGFR changes revealed distinct longitudinal patterns not apparent from the overall mean trajectory (Figure 5).
Figure 5.
Longitudinal changes in eGFR according to early decline patterns after initiation of SGLT2i. Mean eGFR values at baseline (day 0), 30 days, and 60 days after SGLT2i initiation are shown for the sustained decline group (n = 32), transient dip group (n = 14), delayed decline group (n = 14), and no dip group (n = 54). Error bars indicate standard errors. The sustained decline group included patients with ≥10% decline in eGFR from baseline at both 30 and 60 days. The transient dip group included patients with ≥10% decline at 30 days but not at 60 days. The delayed decline group included patients with ≥10% decline at 60 days but not at 30 days. The no‐dip group included patients without ≥10% decline at either 30 or 60 days. eGFR, estimated glomerular filtration rate; SGLT2i, sodium–glucose cotransporter 2 inhibitors.
Baseline characteristics of the sustained‐decline and non‐sustained‐decline groups are shown in Table 2, with only medications used by at least one patient included. Compared with the non‐sustained‐decline group, the sustained‐decline group was older and had higher frequencies of NSAID and RAASi use (all P < 0.01). The distribution of the number of TW‐related drug classes also differed significantly between the groups, with greater use in the sustained‐decline group (P < 0.01). In the sensitivity analysis restricted to the sustained‐decline and no‐dip groups, patients in the sustained‐decline group remained significantly older (76.5 ± 9.9 vs. 70.0 ± 11.5 years, P < 0.01), had more frequent NSAID use (31.3% vs. 7.4%, P < 0.01), and showed a significantly different distribution of TW‐related drug classes (P < 0.01). Conversely, the difference in RAASi use did not reach significance (50.0% vs. 27.8%, P = 0.06) (Table S4).
Table 2.
Baseline characteristics of the single‐center clinical cohort stratified by sustained eGFR decline status
| Sustained decline (n = 32) | Non‐sustained decline (n = 82) | P‐value | |
|---|---|---|---|
| Demographics and functional status | |||
| Age (years) | 76.5 ± 9.9 | 66.7 ± 12 | <0.01 † |
| Male | 16 (50.0) | 57 (69.5) | 0.08 ‡ |
| Physical ADL independence level (J/A/B/C) | 15/11/4/2 | 39/16/7/20 | 0.09 § |
| Long‐term care level (Independent/Support needed/Long‐term care) | 18/5/9 | 57/10/15 | 0.39 § |
| Comorbidities | |||
| CKD | 2 (6.3) | 1 (1.2) | 0.19 ‡ |
| DN | 0 (0.0) | 3 (3.7) | 0.37 ‡ |
| HF | 5 (15.6) | 10 (12.2) | 0.76 ‡ |
| HTN | 19 (59.4) | 35 (42.7) | 0.16 ‡ |
| IHD | 3 (9.4) | 6 (7.3) | 0.71 ‡ |
| Malignancy | 14 (43.8) | 32 (39.0) | 0.80 ‡ |
| Antidiabetic medications | |||
| Biguanides | 2 (6.3) | 10 (12.2) | 0.50 ‡ |
| SU | 1 (3.1) | 6 (7.3) | 0.67 ‡ |
| DPP‐4i | 7 (21.9) | 25 (30.5) | 0.49 § |
| GLP‐1RA | 0 (0.0) | 4 (4.9) | 0.58 ‡ |
| TZDs | 0 (0.0) | 1 (1.2) | 1.00 ‡ |
| α‐GI | 1 (3.1) | 2 (2.4) | 1.00 ‡ |
| Insulin preparations | 1 (3.1) | 7 (8.5) | 0.44 ‡ |
| Suspected AKI‐related drugs | |||
| NSAIDs | 10 (31.3) | 8 (9.8) | <0.01 ‡ |
| Loop diuretics | 13 (40.6) | 31 (37.8) | 0.83 § |
| Thiazide diuretics | 2 (6.3) | 2 (2.4) | 0.31 ‡ |
| MRAs | 11 (34.4) | 17 (20.7) | 0.15 § |
| RAASi | 16 (50.0) | 18 (22.0) | <0.01 ‡ |
| CCBs | 11 (34.4) | 22 (26.8) | 0.49 § |
| β‐blockers | 5 (15.6) | 13 (15.9) | 1.00 ‡ |
| PPIs | 9 (28.1) | 23 (28.0) | 1.00 ‡ |
| Allopurinol | 2 (6.3) | 1 (1.2) | 0.19 ‡ |
| Cephalosporins | 0 (0.0) | 2 (2.4) | 1.00 ‡ |
| Macrolides | 1 (3.1) | 0 (0.0) | 0.28 ‡ |
| Sulfonamides | 5 (15.6) | 6 (7.3) | 0.29 ‡ |
| TW‐related drug classes (0/1/2) | 2/17/13 | 36/31/15 | <0.01 § |
Baseline characteristics of the single‐center clinical cohort stratified by sustained eGFR decline status. Continuous variables are presented as mean ± standard deviation, and categorical variables as n (%). The sustained‐decline group (n = 32) included patients with ≥10% decline in eGFR from baseline at both 30 and 60 days after SGLT2i initiation. The non‐sustained‐decline group (n = 82) comprised patients with a transient dip (n = 14), delayed decline (n = 14), or no dip (n = 54). Physical ADL independence level was classified using the Japanese bedriddenness scale, and long‐term care level was classified according to the Japanese long‐term care insurance system. ADL, activities of daily living; AKI, acute kidney injury; eGFR, estimated glomerular filtration rate; SGLT2i, sodium–glucose cotransporter 2 inhibitors; CKD, chronic kidney disease; DN, diabetic nephropathy; HF, heart failure; HTN, hypertension; IHD, ischemic heart disease; SU, sulfonylurea; DPP‐4i, dipeptidyl peptidase‐4 inhibitor; GLP‐1RA, glucagon‐like peptide‐1 receptor agonist; TZDs, thiazolidinediones; α‐GI, alpha‐glucosidase inhibitor; NSAIDs, nonsteroidal anti‐inflammatory drugs; MRAs, mineralocorticoid receptor antagonists; RAASi, renin–angiotensin–aldosterone system inhibitors; CCBs, calcium channel blockers; PPIs, proton pump inhibitors; TW, triple whammy.
Wilcoxon rank‐sum test.
Fisher's exact test.
χ2 test.
To identify factors associated with sustained eGFR decline, multivariable logistic regression was performed using age, number of TW‐related drug classes, and baseline eGFR (day 0) as covariates (Figure 6). Older age and the use of one or two TW‐related drug classes, compared with no use, were significantly associated with sustained eGFR decline, whereas baseline eGFR was not. The odds of sustained eGFR decline did not differ significantly between using one and two TW‐related drug classes, indicating no clear incremental association with the number of drug classes. In the sensitivity analysis restricted to the sustained‐decline and no‐dip groups, the associations for the use of one or two TW‐related drug classes versus no use remained significant. Older age showed a nonsignificant trend toward an association with sustained eGFR decline (P = 0.06), whereas baseline eGFR remained nonsignificant (P = 0.10) (Figure S1).
Figure 6.
Factors associated with sustained eGFR decline identified by multivariable logistic regression analysis. ORs with 95% CIs are shown for age, baseline eGFR, and the number of TW‐related drug classes. The sustained‐decline group included patients with ≥10% decline in eGFR from baseline at both 30 and 60 days after SGLT2i initiation, whereas the non‐sustained‐decline group comprised patients with a transient dip, delayed decline, or no dip. Age and baseline eGFR were modeled as continuous variables, with ORs expressed per 5‐year increase and per 10 mL/min/1.73 m2 increase, respectively. For the number of TW‐related drug classes, ORs represent comparisons between categories (0 vs. 1, 0 vs. 2, and 1 vs. 2). The vertical dashed line indicates an OR of 1.0. eGFR, estimated glomerular filtration rate; OR, odds ratio; CI, confidence interval; SGLT2i, sodium–glucose cotransporter 2 inhibitor; TW, triple whammy.
DISCUSSION
In the claims‐based case‐only analysis, 59 of the 190 patients who experienced AKI‐related emergency hospitalization after SGLT2i initiation were hospitalized within 30 days. No measured patient characteristic or medication‐related factor included in the adjusted models was significantly associated with classification into the 1–30‐day group, suggesting that this early temporal distribution was not fully explained by the measured patient characteristics and medication‐related factors. In the independent single‐center clinical analysis, sustained early eGFR decline was associated with older age and use of TW‐related drug classes. These single‐center findings provide complementary clinical information on patient factors associated with early renal‐function changes after SGLT2i initiation but should not be interpreted as mechanistic validation of the temporal pattern observed in the claims data because the two analyses involved different patient populations.
One possible explanation for this temporal pattern is the initial dip following SGLT2i initiation. The initial dip typically occurs within 2–4 weeks after initiation, followed by stabilization or partial recovery and subsequent long‐term renoprotection. 24 , 26 , 27 , 28 Based on these temporal characteristics, we categorized the interval from SGLT2i initiation to hospitalization into 1–30, 31–90, and 91–180 days, with the 1–30‐day interval corresponding to the expected timing of the initial dip and later intervals capturing subsequent changes over time. The timing of AKI‐related hospitalizations observed in the claims data was temporally compatible with early glomerular hemodynamic changes after SGLT2i initiation.
However, because baseline renal function, serial laboratory data, and signs of dehydration were not available in the claims database, we could not determine whether these hospitalizations were directly related to the physiological initial dip or assess other potential contributing mechanisms. Moreover, because this was a case‐only analysis including only patients with AKI‐related hospitalization, the analysis did not account for the size of the underlying population or person‐time during each period after SGLT2i initiation and could not estimate the absolute risk of AKI. Accordingly, the observed temporal pattern should be interpreted as the distribution of hospitalization timing among identified cases, rather than as evidence of increased AKI risk during the early period.
In the single‐center analysis, patients in the sustained‐decline group were older and had a higher prevalence of RAASi and NSAID use, as well as a greater number of TW‐related drug classes, whereas baseline eGFR was not significantly associated with the sustained‐decline status. The sustained decline in eGFR observed in this group may reflect interindividual variability in early glomerular hemodynamic responses after SGLT2i initiation, suggesting that a physiological initial dip may manifest as more sustained renal‐function decline.
In particular, the odds of sustained eGFR decline were higher among patients receiving one or two TW‐related drug classes compared with those receiving no TW‐related drug classes; however, no clear stepwise association according to the number of TW‐related drug classes was identified. These findings suggest that early renal‐function decline may be influenced by hemodynamic stress related to aging and concomitant medications, rather than by baseline renal function alone. NSAIDs, diuretics, and RAASi are all known to affect glomerular hemodynamics. In patients exposed to these agents, compensatory regulation of renal hemodynamics may already be engaged at the time of SGLT2i initiation; therefore, adaptation to SGLT2i‐induced reductions in intraglomerular pressure may be attenuated, leading to more sustained decline in eGFR. The findings from the single‐center analysis may provide complementary clinical information for interpreting the temporal distribution of early AKI hospitalizations observed in the claims‐database analysis, although they do not establish a causal or mechanistic relationship between sustained early eGFR decline and AKI‐related hospitalization. Because randomized trials and meta‐analyses consistently show lower overall AKI risk with SGLT2is, 15 , 29 , 30 these findings do not contradict the established long‐term renoprotective effects of SGLT2is but rather highlight that a substantial proportion of the identified AKI‐related hospitalizations occurred during the early phase after initiation.
SGLT2is are established cardiorenal protective therapies, with benefits for glycemic control, heart failure, and CKD. Their efficacy is evident from the early phase after initiation and is sustained over the long term with continued treatment. 31 , 32 , 33 However, some patients discontinue treatment or exhibit poor adherence, with common reasons including polyuria, infections, and declines in eGFR. 34 , 35 Although the safety profile of SGLT2is in older adults is generally comparable to that in younger populations, 36 , 37 treatment initiation is often approached with caution in the presence of multimorbidity and polypharmacy. Moreover, product labeling for SGLT2is includes changes in renal–function parameters and renal impairment as potential adverse events, which may create challenges in interpreting renal–function changes despite their established renoprotective effects. The present findings should be regarded as hypothesis‐generating and suggest that early renal–function changes after SGLT2i initiation warrant further investigation, particularly in older adults receiving concomitant medications that may affect renal hemodynamics. Prospective studies are needed to determine whether early monitoring can prevent unnecessary treatment discontinuation or improve the safety of SGLT2i use.
This study has several limitations. First, as a retrospective observational study based on administrative claims data, we were unable to adequately capture clinical parameters such as serial eGFR values, serum creatinine levels, and urine output. Therefore, the underlying mechanisms and severity of renal‐function changes could not be directly evaluated, and the precise pathophysiology of AKI or its direct relationship with the initial dip could not be determined. Additionally, we could not distinguish whether AKI observed in the early phase after initiation was directly attributable to SGLT2is or to other factors.
Second, AKI‐related emergency hospitalizations were defined based on ICD‐10 codes, which may be subject to limited diagnostic accuracy and potential misclassification bias. Differences in diagnostic and coding practices across institutions may also have influenced the results. Furthermore, because this study focused on emergency hospitalization cases, mild AKI or transient declines in renal function managed as outpatient cases may not have been captured.
Third, SGLT2i initiation was based on physicians' clinical judgment, and patients selected for treatment may have systematically differed from those not selected for treatment, raising the possibility of indication bias. Although we adjusted for measured confounders using propensity‐score methods, residual confounding due to unmeasured factors cannot be excluded. Furthermore, SGLT2is may have been initiated in response to worsening renal function or overall clinical status already deteriorating before initiation, which could have subsequently led to AKI hospitalization (protopathic bias). Such reverse causation may partially explain the observed temporal pattern of AKI‐related hospitalizations. However, this temporal pattern is likely multifactorial, including early hemodynamic changes after SGLT2i initiation.
Fourth, this study did not estimate the absolute incidence of AKI among all SGLT2i users but rather evaluated the temporal distribution of AKI‐related hospitalizations among cases. Finally, because the claims‐based analysis was conducted in an older Japanese population, the findings may not be directly generalizable to younger populations or other healthcare settings.
CONCLUSION
This study evaluated the temporal distribution of AKI‐related emergency hospitalizations following SGLT2i initiation in older adults. Among the identified cases, a substantial proportion of hospitalizations occurred within 30 days after initiation. However, because this was a case‐only analysis, the findings should be interpreted as a temporal distribution rather than as evidence of increased early risk of AKI. These findings are hypothesis‐generating and warrant further investigation using a new‐user cohort design that includes all SGLT2i users and accounts for person‐time after initiation.
Disclosure
The authors declare no conflict of interest.
Approval of the research protocol: This study was conducted in accordance with the Declaration of Helsinki and the Ethical Guidelines for Medical and Biological Research Involving Human Subjects in Japan. The analysis using the administrative claims database was approved by the Ethics Committee of Hokkaido University of Science, Hokkaido, Japan (approval No.: 25‐25). The single‐center clinical data analysis was approved by the Ethics Committee of JA Hokkaido Koseiren Sapporo Kosei Hospital, Hokkaido, Japan (approval No.: 869).
Informed consent: For the claims database analysis, fully anonymized secondary data were used, and the requirement for informed consent was waived. For the single‐center clinical data analysis, this retrospective observational study used medical record data, and informed consent was obtained using an opt‐out approach.
Approval date of registry and the registration no. of the study/trial: N/A.
Animal studies: N/A.
Supporting information
Figure S1. Factors associated with sustained eGFR decline in the sensitivity analysis restricted to the sustained‐decline and no‐dip groups.
Table S1. Definitions of Comorbidities Based on ICD‐10 Codes.
Table S2. Antidiabetic medications.
Table S3. Concomitant medications potentially associated with AKI.
Table S4. Baseline characteristics of the single‐center clinical cohort comparing the sustained‐decline and no‐dip groups.
Acknowledgments
We would like to thank Editage (www.editage.jp) for its English language editing services. An artificial intelligence tool (ChatGPT, OpenAI, San Francisco, CA, USA) was also used to improve the clarity and language of the manuscript. All AI‐assisted content was carefully reviewed, verified, and revised by the authors, who take full responsibility for the integrity and accuracy of the final manuscript.
DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.
References
- 1. Rajasekeran H, Lytvyn Y, Cherney DZ. Sodium‐glucose cotransporter 2 inhibition and cardiovascular risk reduction in patients with type 2 diabetes: The emerging role of natriuresis. Kidney Int 2016; 89: 524–526. [DOI] [PubMed] [Google Scholar]
- 2. Neal B, Perkovic V, Mahaffey KW, et al. Canagliflozin and cardiovascular and renal events in type 2 diabetes. N Engl J Med 2017; 377: 644–657. [DOI] [PubMed] [Google Scholar]
- 3. Perkovic V, Jardine MJ, Neal B, et al. Canagliflozin and renal outcomes in type 2 diabetes and nephropathy. N Engl J Med 2019; 380: 2295–2306. [DOI] [PubMed] [Google Scholar]
- 4. Shiraishi Y, Kohsaka S, Sato N, et al. 9‐year trend in the Management of Acute Heart Failure in Japan: A report from the National Consortium of acute heart failure registries. J Am Heart Assoc 2018; 7: e008687. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Uchmanowicz I, Kuśnierz M, Wleklik M, et al. Frailty syndrome and rehospitalizations in elderly heart failure patients. Aging Clin Exp Res 2018; 30: 617–623. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Wanner C, Inzucchi SE, Lachin JM, et al. Empagliflozin and progression of kidney disease in type 2 diabetes. N Engl J Med 2016; 375: 323–334. [DOI] [PubMed] [Google Scholar]
- 7. Wiviott SD, Raz I, Bonaca MP, et al. Dapagliflozin and cardiovascular outcomes in type 2 diabetes. N Engl J Med 2019; 380: 347–357. [DOI] [PubMed] [Google Scholar]
- 8. Zinman B, Wanner C, Lachin JM, et al. Empagliflozin, cardiovascular outcomes, and mortality in type 2 diabetes. N Engl J Med 2015; 373: 2117–2128. [DOI] [PubMed] [Google Scholar]
- 9. Takahashi K, Nakamura A, Furusawa S, et al. Initial dip predicts renal protective effects after the administration of sodium‐glucose cotransporter 2 inhibitors in patients with type 2 diabetes and chronic kidney disease with normoalbuminuria. J Clin Transl Endocrinol 2020; 22: 100244. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Oshima M, Jardine MJ, Agarwal R, et al. Insights from CREDENCE trial indicate an acute drop in estimated glomerular filtration rate during treatment with canagliflozin with implications for clinical practice. Kidney Int 2021; 99: 999–1009. [DOI] [PubMed] [Google Scholar]
- 11. Kraus BJ, Weir MR, Bakris GL, et al. Characterization and implications of the initial estimated glomerular filtration rate 'dip' upon sodium‐glucose cotransporter‐2 inhibition with empagliflozin in the EMPA‐REG OUTCOME trial. Kidney Int 2021; 99: 750–762. [DOI] [PubMed] [Google Scholar]
- 12. Adamson C, Docherty KF, Heerspink HJL, et al. Initial decline (dip) in estimated glomerular filtration rate after initiation of Dapagliflozin in patients with heart failure and reduced ejection fraction: Insights from DAPA‐HF. Circulation 2022; 146: 438–449. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Hiura K, Suzuki C, Kubo J, et al. Comparison of kidney and hepatic outcomes among sodium‐glucose cotransporter‐2 inhibitors: A retrospective study using multiple propensity scores. J Pharm Health Care Sci 2024; 10: 57. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Szalat A, Perlman A, Muszkat M, et al. Can SGLT2 inhibitors cause acute renal failure? Plausible role for altered glomerular hemodynamics and medullary hypoxia. Drug Saf 2018; 41: 239–252. [DOI] [PubMed] [Google Scholar]
- 15. Menne J, Dumann E, Haller H, et al. Acute kidney injury and adverse renal events in patients receiving SGLT2‐inhibitors: A systematic review and meta‐analysis. PLoS Med 2019; 16: e1002983. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Eba J, Nakamura K. Overview of the ethical guidelines for medical and biological research involving human subjects in Japan. Jpn J Clin Oncol 2022; 52: 539–544. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Ministry of health law . Guidelines for the utilization of the independence degree of daily living for elderly with dementia. Notification No. 0403003. 2006. https://www.mhlw.go.jp/stf/shingi/2r9852000001hi4o‐att/2r9852000001hi8n.pdf.
- 18. Japanese Society of Nephrology . Clinical Practice Guideline for Drug‐Induced Kidney Injury 2016. Tokyo: Japanese Society of Nephrology, 2016. [Google Scholar]
- 19. Leete J, Wang C, López‐Hernández FJ, et al. Determining risk factors for triple whammy acute kidney injury. Math Biosci 2022; 347: 108809. [DOI] [PubMed] [Google Scholar]
- 20. Calvo DM, Saiz LC, Leache L, et al. Effect of the combination of diuretics, renin‐angiotensin‐aldosterone system inhibitors, and non‐steroidal anti‐inflammatory drugs or metamizole (triple whammy) on hospitalisation due to acute kidney injury: A nested case‐control study. Pharmacoepidemiol Drug Saf 2023; 32: 898–909. [DOI] [PubMed] [Google Scholar]
- 21. Xu S, Ross C, Raebel MA, et al. Use of stabilized inverse propensity scores as weights to directly estimate relative risk and its confidence intervals. Value Health 2010; 13: 273–277. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. Austin PC. An introduction to propensity score methods for reducing the effects of confounding in observational studies. Multivar Behav Res 2011; 46: 399–424. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Tamiya N, Noguchi H, Nishi A, et al. Population ageing and wellbeing: Lessons from Japan's long‐term care insurance policy. Lancet 2011; 378: 1183–1192. [DOI] [PubMed] [Google Scholar]
- 24. Heerspink HJL, Cherney DZI. Clinical implications of an acute dip in eGFR after SGLT2 inhibitor initiation. Clin J Am Soc Nephrol 2021; 16: 1278–1280. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Sugiyama S, Yoshida A, Hieshima K, et al. Initial acute decline in estimated glomerular filtration rate after sodium‐glucose Cotransporter‐2 inhibitor in patients with chronic kidney disease. J Clin Med Res 2020; 12: 724–733. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Meraz‐Muñoz AY, Weinstein J, Wald R. eGFR decline after SGLT2 inhibitor initiation: The tortoise and the hare reimagined. Kidney360 2021; 2: 1042–1047. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Chan YH, Chao TF, Chen SW, et al. The Association of Background Medications on initial eGFR change and kidney outcomes in diabetic patients receiving SGLT2 inhibitor. Clin J Am Soc Nephrol 2023; 18: 858–868. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Chuang MH, Tang YS, Chen JY, et al. Abrupt decline in estimated glomerular filtration rate after initiating sodium‐glucose cotransporter 2 inhibitors predicts clinical outcomes: A systematic review and meta‐analysis. Diabetes Metab J 2024; 48: 242–252. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29. Dong Z, Mo W, Ling Z, et al. Efficacy of SGLT2 inhibitors on acute kidney injury in patients with chronic kidney disease, cardiovascular disease, and type 2 diabetes: A meta‐analysis. J Natl Med Assoc 2025; 117: 458–469. [DOI] [PubMed] [Google Scholar]
- 30. Wang Q, Yu J, Deng W, et al. Influence of sodium/glucose cotransporter‐2 inhibitors on the incidence of acute kidney injury: A meta‐analysis. Front Pharmacol 2024; 15: 1372421. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31. Zou X, Shi Q, Vandvik PO, et al. Sodium–glucose Cotransporter‐2 inhibitors in patients with heart failure. Ann Intern Med 2022; 175: 851–861. [DOI] [PubMed] [Google Scholar]
- 32. Chen K, Nie Z, Shi R, et al. Time to benefit of sodium‐glucose Cotransporter‐2 inhibitors among patients with heart failure. JAMA Netw Open 2023; 6: e2330754. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33. Noiri JI, Tanaka H, Odajima S, et al. Efficacy of sodium‐glucose cotransporter 2 inhibitors for super‐aged heart failure population. Int J Cardiol 2025; 439: 133647. [DOI] [PubMed] [Google Scholar]
- 34. Saijo Y, Okada H, Hata S, 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: 6993. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35. Hopf M, Kloos C, Wolf G, et al. Effectiveness and safety of SGLT2 inhibitors in clinical routine treatment of patients with diabetes mellitus type 2. J Clin Med 2021; 10: 571. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36. Scheen AJ, Bonnet F. Efficacy and safety profile of SGLT2 inhibitors in the elderly: How is the benefit/risk balance? Diabetes Metab 2023; 49: 101419. [DOI] [PubMed] [Google Scholar]
- 37. Wang Y, Shao X, Liu Z. Efficacy and safety of sodium‐glucose co‐transporter 2 inhibitors in the elderly versus non‐elderly patients with type 2 diabetes mellitus: A meta‐analysis. Endocr J 2022; 69: 669–679. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Figure S1. Factors associated with sustained eGFR decline in the sensitivity analysis restricted to the sustained‐decline and no‐dip groups.
Table S1. Definitions of Comorbidities Based on ICD‐10 Codes.
Table S2. Antidiabetic medications.
Table S3. Concomitant medications potentially associated with AKI.
Table S4. Baseline characteristics of the single‐center clinical cohort comparing the sustained‐decline and no‐dip groups.
Data Availability Statement
The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.