1. Background
Type 2 diabetes mellitus (T2DM) is a common metabolic disease characterized by chronic hyperglycemia and microvascular and macrovascular complications (1). Empagliflozin, a sodium-glucose cotransporter 2 (SGLT2) inhibitor, is a relatively recent therapeutic agent with demonstrated cardiorenal benefits (2).
Empagliflozin reduces plasma glucose concentrations by blocking glucose reabsorption in the proximal renal tubules, thereby increasing urinary glucose excretion. Large randomized controlled trials have demonstrated that empagliflozin and dapagliflozin improve glycemic control while reducing hospitalizations for heart failure, slowing chronic kidney disease (CKD) progression, and lowering cardiovascular mortality (3, 4). Empagliflozin may also increase hemoglobin and hematocrit levels. One potential consequence of these changes is erythrocytosis, a complication that has received comparatively limited attention (5). Recent clinical evidence also suggests that SGLT2 inhibitors exert systemic effects beyond glycemic and cardiorenal outcomes. For example, empagliflozin has been shown to ameliorate hepatic inflammation and improve liver enzyme profiles in patients with T2DM, indicating that its physiological effects extend to multiple organ systems (6). These systemic effects may help contextualize the hematologic changes observed in patients receiving SGLT2 inhibitors.
Chronic kidney disease is defined by persistent structural or functional abnormalities of the kidneys lasting at least three months and represents a major global health burden (7). An estimated 850 million individuals worldwide have some form of kidney disease, and CKD is among the fastest-growing causes of mortality globally (8). Nearly one-third to one-half of patients with T2DM develop CKD, underscoring the importance of effective therapeutic strategies to slow disease progression and reduce complications (9).
Multiple investigations have reported that extended use of SGLT2 inhibitors, including empagliflozin, leads to modest increases in hemoglobin and hematocrit levels, generally reflecting hemoconcentration or enhanced erythropoiesis. In contrast, clinically defined erythrocytosis represents a pathological increase in red blood cell mass, typically diagnosed when hemoglobin exceeds 16.5 g/dL or hematocrit exceeds 49% in males, or when hemoglobin exceeds 16 g/dL or hematocrit exceeds 48% in females (10, 11). The reported prevalence of erythrocytosis varies widely, ranging from 2% to 19%, depending on the population studied (12). Although the mechanisms remain incompletely understood, proposed pathways include enhanced erythropoietin production, improved renal oxygenation, and reduced oxidative stress (13).
A rise in red blood cell mass may reflect hemoconcentration or erythropoiesis-related changes. Although erythrocytosis has been associated with increased blood viscosity in other clinical contexts, our study did not assess thrombotic or cardiovascular events. This distinction is particularly relevant in individuals with concurrent T2DM and CKD, who have impaired erythropoietic regulation, reduced renal oxygenation, chronic inflammation, and elevated cardiovascular risk, all of which may increase susceptibility to hematologic fluctuations. Clarifying the relationship between SGLT2 inhibitor therapy and erythrocytosis in patients with concurrent T2DM and CKD is therefore of considerable clinical and pathophysiological importance (14). Given the expanding use of SGLT2 inhibitors and the scarcity of data from Iranian CKD populations, further investigation is warranted.
2. Objectives
This study aimed to assess the association between empagliflozin use and the occurrence of erythrocytosis among patients with T2DM and CKD.
3. Methods
3.1. Study Design and Setting
This retrospective cohort study explored the association between empagliflozin use and erythrocytosis in patients with T2DM and CKD. Data were extracted from the medical records of patients treated at Shahid Beheshti and Alavi hospitals in Abadan, Iran, during 2023 - 2024.
3.2. Participants and Eligibility Criteria
Records were eligible for analysis if core laboratory data, including hemoglobin, hematocrit, serum creatinine, and estimated glomerular filtration rate, were complete. Missing data for other demographic or clinical variables did not preclude eligibility.
The inclusion criteria were as follows:
1) A confirmed diagnosis of T2DM based on the American Diabetes Association diagnostic criteria;
2) CKD, defined as an eGFR < 60 mL/min/1.73 m2 for at least three months;
3) Complete laboratory data for hemoglobin, hematocrit, serum creatinine, and eGFR.
Patients were excluded if they had acute kidney injury, end-stage renal disease requiring dialysis or kidney transplantation, known hematologic or bone marrow disorders, or were receiving medications known to stimulate erythropoiesis, such as erythropoietin or testosterone. Records with incomplete or unreliable core laboratory data were also excluded.
3.3. Exposure Definition
The exposed group included 100 patients who were prescribed empagliflozin at a daily dose of 25 mg by endocrinologists or nephrologists at the participating hospitals. Continuous use was confirmed through follow-up visits and prescription records, and patients were required to have received at least six months of uninterrupted therapy. The comparison group consisted of 200 patients who did not receive empagliflozin. Although limited medication availability contributed to nonuse, both groups were drawn from the same clinical settings, treated by the same endocrinologists and nephrologists, and evaluated during the same study period. Baseline demographic and clinical characteristics, including age, sex, duration of diabetes, CKD stage, comorbidities, and concomitant medications, were compared to assess whether nonuse reflected systematic differences in disease severity, prescribing patterns, or socioeconomic status. No significant differences were observed across these variables, minimizing the likelihood of selection bias.
3.4. Outcome Definition
The primary outcome was erythrocytosis, defined according to World Health Organization criteria and verified using local laboratory reference ranges. For empagliflozin users, hemoglobin and hematocrit values were obtained at baseline, before treatment initiation, and at follow-up six months after the index date. For nonusers, stable hemoglobin and hematocrit measurements were obtained at baseline and at follow-up six months after the index date. Laboratory results associated with transient hemoconcentration or dehydration, identified through clinical documentation, elevated serum sodium, or abnormal blood urea nitrogen/creatinine ratios, were excluded to ensure that only stable hematologic values were analyzed.
3.5. Data Collection and Variables
Demographic characteristics (age, sex, and body mass index [BMI]), clinical variables (duration of diabetes, duration of CKD, smoking status, and comorbid conditions), and laboratory findings (hemoglobin, hematocrit, serum creatinine, and eGFR) were extracted from patient records. Information on concomitant medications that could influence hemoglobin levels, including ACEi/ARB and diuretics, was also documented. Detailed iron studies, vitamin B12/folate levels, and hypoxia-related conditions were not consistently available and therefore could not be systematically analyzed.
For empagliflozin users, the index date was the date of treatment initiation. Baseline laboratory values were obtained before exposure, and follow-up ended after six months of continuous therapy. For nonusers, the index date was the most recent stable laboratory evaluation within the same study period, and follow-up ended six months later. Patients with baseline erythrocytosis were excluded to establish clear temporality between exposure and outcome. Data abstraction was performed using standardized forms by two independent investigators, with discrepancies resolved by reviewing the original records.
3.6. Sample Size Estimation
Sample size estimation was initially performed using FASS software; however, detailed assumptions regarding the expected event rate, effect size, and rationale for the 1:2 exposure ratio were not available in the study records. The final sample size was therefore determined by the number of eligible patients identified during the study period, resulting in 100 exposed and 200 unexposed individuals.
3.7. Statistical Analysis
Statistical analyses were conducted using SPSS software, version 26. Continuous variables were summarized as mean ± standard deviation, and categorical variables as frequencies and percentages. Between-group comparisons were performed using the independent t-test or Mann-Whitney U test for continuous variables and the chi-square or Fisher exact test for categorical variables, as appropriate. Given the limited number of erythrocytosis events (n = 13), a parsimonious logistic regression model was used to reduce the risk of overfitting. Age and sex were included as core covariates, and additional variables were incorporated only if clinically essential. Penalized or exact logistic regression methods were considered but could not be applied because of software limitations. Results were expressed as odds ratios (ORs) with 95% confidence intervals (CIs), and a two-sided P value < 0.05 was considered statistically significant. Sex was coded as female = 1 and male = 2; smoking status as smoker = 1 and nonsmoker = 2; empagliflozin use as user = 1 and nonuser = 2; and comorbidities as hypertension = 1, cardiovascular disease = 2, metabolic disease = 3, hematologic disease = 4, neurologic disease = 5, and pulmonary disease = 6.
3.8. Ethical Considerations
The study adhered to the ethical principles of the Declaration of Helsinki and complied with institutional research ethics regulations. All patient information was anonymized and coded to ensure confidentiality, and access to the data was restricted to the research team. Because this was a retrospective study with no direct patient intervention, no changes were made to clinical management. The research protocol was approved by the Ethics Committee of Abadan University of Medical Sciences (approval code: IR.ABADANUMS.REC.1404.049), and the Ethics Committee waived the requirement for informed consent.
4. Results
The baseline demographic and clinical characteristics of the participants are summarized in Table 1. Overall, the two groups were well balanced for most baseline variables, including age, sex distribution, BMI, smoking status, duration of diabetes, duration of CKD, serum creatinine, and eGFR, with no statistically significant differences between empagliflozin users and nonusers (all P > 0.05). However, baseline hemoglobin levels were significantly higher among empagliflozin users than among nonusers (11.88 ± 1.91 vs 11.07 ± 1.91 g/dL, P = 0.001). Hematocrit levels were also significantly higher among users (35.6 ± 5.7% vs 33.2 ± 5.7%, P = 0.002). These imbalances were accounted for by including baseline hemoglobin in the fully adjusted logistic regression models. The prevalence of comorbidities was comparable between the groups, with hypertension being the most common comorbidity in both groups (62% vs 65%, P = 0.61). Missing data were minimal across all variables, with the highest proportion observed for total comorbidities (2.0%); complete-case analysis was used to handle missing data.
Table 1.
Baseline Demographic and Clinical Characteristics of Study Participants a
| Characteristics | Empagliflozin (n = 100) | Non-users (n = 200) | P value | Missing |
|---|---|---|---|---|
| Demographic characteristics | ||||
| Age (y) | 63.67 ± 11.75 | 64.95 ± 11.75 | 0.38 | 0 (0) |
| Gender (male) | 45 (45) | 108 (54) | 0.14 | 0 (0) |
| BMI (kg/m2) | 28.4 ± 5.2 | 27.9 ± 5.6 | 0.45 | 4 (1.3) |
| Smoking | 29 (29) | 80 (40) | 0.06 | 2 (0.7) |
| Diabetes and CKD history | ||||
| Duration of diabetes (y) | 10.2 ± 6.8 | 10.8 ± 7.1 | 0.48 | 3 (1.0) |
| Duration of CKD (y) | 4.5 ± 3.2 | 4.8 ± 3.5 | 0.47 | 5 (1.7) |
| Laboratory parameters | ||||
| Baseline hemoglobin (g/dL) | 11.88 ± 1.91 | 11.07 ± 1.91 | 0.001 | 0 (0) |
| Hematocrit (%) | 35.6 ± 5.7 | 33.2 ± 5.7 | 0.002 | 0 (0) |
| Serum creatinine (mg/dL) | 1.98 ± 0.68 | 2.09 ± 0.71 | 0.17 | 0 (0) |
| eGFR (mL/min/1.73 m2) | 37.01 ± 11.87 | 35.83 ± 11.87 | 0.42 | 0 (0) |
| Comorbidities | ||||
| Hypertension | 62 (62) | 130 (65) | 0.61 | 3 (1.0) |
| Cardiovascular disease | 28 (28) | 60 (30) | 0.72 | 3 (1.0) |
| Metabolic disease | 15 (15) | 35 (17.5) | 0.58 | 4 (1.3) |
| Hematologic disease | 4 (4) | 6 (3) | 0.64 | 5 (1.7) |
| Neurologic disease | 8 (8) | 18 (9) | 0.77 | 4 (1.3) |
| Pulmonary disease | 6 (6) | 14 (7) | 0.74 | 4 (1.3) |
| Total comorbidities | 1.23 ± 0.95 | 1.29 ± 0.98 | 0.61 | 6 (2.0) |
a Values are expressed as mean ± SD or No. (%). Abbreviations: BMI, Body Mass Index; CKD, chronic kidney disease; eGFR, estimated glomerular filtration rate; SD, standard deviation.
In the univariate logistic regression model, empagliflozin therapy was associated with higher estimated odds of erythrocytosis (OR = 3.64; 95% CI, 0.59 - 22.20; P = 0.161). However, this association did not reach statistical significance, and the wide confidence interval indicated substantial uncertainty in the estimate (Table 2).
Table 2.
Univariate Logistic Regression Analysis of Empagliflozin Use and Erythrocytosis
| Variables | Odds Ratio | 95% Confidence Interval | P Value |
|---|---|---|---|
| Empagliflozin use | 3.64 | 0.59 - 22.20 | 0.161 |
After adjustment for age and sex, the positive association between empagliflozin use and erythrocytosis persisted (OR = 4.38; 95% CI, 0.66 - 28.90; P = 0.125). Age showed an inverse association with erythrocytosis in the age- and sex-adjusted model (OR = 0.91; 95% CI, 0.84 - 0.97; P = 0.012); however, this finding should be interpreted cautiously because it was not replicated in the fully adjusted model. Sex was not significantly associated with the outcome (OR = 1.63; 95% CI, 0.25 - 10.44; P = 0.605) (Table 3).
Table 3.
Multivariable Logistic Regression Adjusted for Age and Sex
| Variables | Odds Ratio | 95% Confidence Interval | P Value |
|---|---|---|---|
| Empagliflozin use | 4.38 | 0.66 - 28.90 | 0.125 |
| Age | 0.91 | 0.84 - 0.97 | 0.012 |
| Sex | 1.63 | 0.25 - 10.44 | 0.605 |
In the fully adjusted logistic regression model, the estimate for empagliflozin use was extremely imprecise (OR = 41.73; 95% CI, 0.11 - 15061.54; P = 0.214), reflecting sparse data and a very wide confidence interval. This result should be considered exploratory and does not provide reliable evidence of increased risk (Table 4).
Table 4.
Fully Adjusted Multivariable Logistic Regression Analysis for Erythrocytosis a
| Variables | Odds Ratio | 95% Confidence Interval | P Value |
|---|---|---|---|
| Empagliflozin use | 41.73 | 0.11 - 15061.54 | 0.214 |
| Age | 0.85 | 0.55 - 1.33 | 0.492 |
| Gender | 0.0019 | 0.0000000068 - 540.40 | 0.328 |
| Smoking | 0.0049 | 0.000000079 - 308.30 | 0.346 |
| Other comorbidities | 1.04 | 0.91 - 1.18 | 0.552 |
| Duration of diabetes | 0.0007 | 0.000000034 - 12.94 | 0.147 |
| eGFR | 0.93 | 0.53 - 1.67 | 0.828 |
| Creatinine | 0.43 | 0.0005 - 346.14 | 0.803 |
| Baseline hemoglobin (Hb1) | 16.31 | 0.71 - 373.90 | 0.081 |
a Abbreviations: eGFR, estimated glomerular filtration rate; Hb1, baseline hemoglobin.
Other covariates, including sex (OR = 0.0019, P = 0.328), smoking (OR = 0.0049, P = 0.346), eGFR (OR = 0.93, P = 0.828), and creatinine (OR = 0.43, P = 0.803), were not significantly associated with erythrocytosis. Baseline hemoglobin showed a borderline positive association (OR = 16.31; 95% CI, 0.71 - 373.90; P = 0.081), suggesting that higher initial hemoglobin levels may be associated with a greater probability of erythrocytosis.
Given the small number of erythrocytosis events (n = 13), the fully adjusted model may be unstable. These results should be interpreted cautiously.
5. Discussion
This retrospective cohort study investigated the relationship between empagliflozin treatment and the development of erythrocytosis in individuals with T2DM and CKD. In both unadjusted and adjusted models, empagliflozin use was associated with higher estimated odds of erythrocytosis; however, these associations did not reach statistical significance, and the wide confidence intervals indicated substantial uncertainty in the estimates (15).
In unadjusted analyses, empagliflozin users showed a trend toward increased odds of erythrocytosis (OR = 3.64), but this finding did not reach statistical significance. This pattern is consistent with earlier reports indicating that SGLT2 inhibitors can raise hemoglobin and hematocrit levels (15). The lack of significance is likely related to the modest sample size and resulting wide confidence intervals. Nonetheless, this pattern underscores the potential clinical relevance of the findings and warrants further investigation in larger populations.
Adjustment for age and sex maintained the positive association (OR = 4.38). Age emerged as a significant factor, with younger patients showing increased susceptibility to erythrocytosis (OR = 0.91, P = 0.012). Sex did not appear to modify the association, consistent with earlier studies suggesting that the erythropoietic effects of these drugs are not sex-specific.
In the fully adjusted multivariable model, which accounted for multiple potential confounders, empagliflozin use remained associated with substantially increased odds of erythrocytosis (OR = 41.73), although statistical significance was not reached (P = 0.214). The wide confidence interval reflects sample variability and limited power. None of the examined covariates were significantly associated with erythrocytosis, although baseline hemoglobin showed a borderline association (OR = 16.31, P = 0.081), suggesting that patients with higher initial hemoglobin levels may be more likely to develop erythrocytosis.
Overall, these findings suggest a possible association between empagliflozin therapy and increased odds of erythrocytosis, although statistical significance was not demonstrated. Research on the broader metabolic and inflammatory aspects of T2DM has also examined the effects of lifestyle interventions, including exercise timing, on inflammatory and cognitive outcomes in affected populations (16). However, these findings do not establish a direct relationship between exercise-related changes in inflammation and erythrocytosis associated with SGLT2 inhibitors. Previous studies have reported metabolic and anti-inflammatory effects of empagliflozin in extrarenal organs, including the liver (6), suggesting that SGLT2 inhibitors may simultaneously influence multiple physiological pathways. Although these studies do not directly evaluate hematologic parameters, they support the concept that SGLT2 inhibitors can induce widespread biological changes that may contribute to alterations in hemoglobin and hematocrit levels. The primary limitation was the relatively small sample size, which constrained the precision of the effect estimates. Larger prospective studies are needed to confirm these associations and clarify the hematologic effects of these agents in patients with CKD (17, 18).
From a clinical perspective, these findings highlight the importance of monitoring hemoglobin and hematocrit levels in patients receiving empagliflozin (19). Because our study did not assess thromboembolic events or management outcomes, clinical recommendations should be limited to careful monitoring. Further research is needed to clarify the clinical significance of these hematologic changes.
The very broad confidence intervals and unstable estimates, notably the OR of 41.73 in the fully adjusted model, probably result from sparse-data bias and overfitting rather than reflecting a genuine biological phenomenon. Because major trials such as EMPA-REG OUTCOME, CANVAS, and CREDENCE did not report erythrocytosis outcomes comparable to ours, direct numerical comparison is inappropriate. The present findings should therefore be interpreted as hypothesis-generating rather than confirmatory.
Several hypothetical mechanisms have been proposed to explain SGLT2 inhibitor-associated erythrocytosis, including enhanced erythropoietin (EPO) production, improved renal tubular oxygenation, and reduced oxidative stress (15). Prior studies, including those by Sano et al., suggest that increases in hematocrit during SGLT2 inhibitor therapy may be related to elevated EPO levels and possible HIF stabilization (15). Additionally, reduced proximal tubular sodium reabsorption could theoretically decrease oxygen consumption in the renal cortex, potentially alleviating chronic hypoxia and stimulating EPO secretion from peritubular fibroblasts. From a health care management perspective, effective monitoring of potential medication-related complications requires coordination of clinical processes, information sharing, and organizational learning (20). These mechanisms remain speculative in the context of our dataset because none of these pathways were measured, and their relative contributions require further investigation.
5.1. Limitations
Several limitations should be acknowledged. First, the modest sample size reduced statistical power and widened confidence intervals. Second, the retrospective observational design precludes causal inference. Third, residual confounding by unmeasured factors, including nutritional status, iron metabolism, and other clinical variables, cannot be excluded. Fourth, conducting the study in teaching hospitals in a single regional setting may limit generalizability. Fifth, the small number of erythrocytosis events relative to the number of covariates may have contributed to sparse-data bias, model instability, and overfitting. Additionally, the retrospective design introduces the potential for immortal-time bias, and exposure and outcome misclassification may have occurred because of reliance on routine clinical documentation. Missing-data mechanisms were not fully characterized, and the absence of propensity-score balancing may have allowed systematic differences between exposed and unexposed groups to persist. Finally, residual confounding by anemia-related diseases and medications cannot be ruled out. Because the exposed group consisted exclusively of empagliflozin users, the findings may not be generalizable to all SGLT2 inhibitors.
5.2. Conclusions
This study did not demonstrate a statistically significant association between empagliflozin use and erythrocytosis in individuals with T2DM and CKD. All effect estimates were imprecise, with wide confidence intervals indicating substantial uncertainty. The findings should therefore be interpreted cautiously and considered exploratory rather than confirmatory.
