But researchers are unsure of the clinical significance of the findings.
If an SGLT2 inhibitor increases haemoglobin and no one knows what it means, does it make a sound?
The use of sodium-glucose cotransporter 2 (SGLT2) inhibitors has previously been linked to lower rates of cardiovascular events, slowed progression of chronic kidney disease, and increases in haemoglobin and haematocrit – although these studies have been unable to confirm the magnitude and significance of these changes.
Now, a recent systematic review and meta-analysis, published in JAMA, has taken a closer look at the associations between SGLT2 inhibitor use and haemoglobin, haematocrit, and a range of markers of iron metabolism and erythropoiesis in people with cardiovascular disease, diabetes, heart failure, and chronic kidney disease.
Researchers included 70,543 participants (25,334 females, mean age 62.4 years) from 85 clinical trials in their primary analyses. Most trials involved people with type 2 diabetes (15,707 participants from 51 trials), but cardiovascular disease trials accounted for the largest proportion of participants (27,153 from nine trials). Other conditions included in the systematic review and meta-analysis were heart failure (14,499 participants from 14 trials), chronic kidney disease (10,355 participants from nine trials), and type 1 diabetes (2856 participants from two trials).
Eight different interventions were used: empagliflozin (23,363 participants from 30 trials), dapagliflozin (24,294 participants from 25 trials), canagliflozin (6164 participants from eight trials), ertugliflozin (10,266 from seven trials), ipragliflozin (1481 participants from six trials), luseogliflozin (841 participants from four trials), sotagliflozin (3643 participants from three trials), and bexagliflozin (286 participants from one trial).
Most trials compared an SGLT2 inhibitor to placebo (78 trials involving 67,208 participants). The remaining seven trials put an SGLT2 inhibitor up against an active comparator: sitagliptin (1158 participants across two trials), glipizide (814 participants across one trial), linagliptin (405 participants across one trial), vildagliptin (49 participants across one trial), glimepiride (538 participants across one trial), and metformin (371 participants across one trial).
The researchers focused on two primary outcomes: the mean change from baseline in haemoglobin (data available from 33,997 participants from 46 of the 85 trials) and haematocrit (67,370 participants from 74 of the 85 trials). The median time of follow-up was 24 weeks for both primary outcomes.
Participants in the SGLT2 inhibitor arms displayed a 0.55g/dL increase in their haemoglobin compared to a decrease of 0.08g/dL in patients in the control arms (overall difference 0.61g/dL, 95% confidence interval 0.55g/dL to 0.67g/dL). Similarly, participants treated with SGLT2 inhibitors saw a larger increase in their haematocrit compared to participants in the control arms (2.17% versus -0.13, overall difference of 2.36% [2.19% to 2.52%]).
The researchers noted the 2.36% increase in haematocrit was smaller than the 5-11% increase in haematocrit that is usually seen following the use of erythropoiesis-stimulating agents – such as EPO – that are used in the treatment of anaemia.
“Therefore, the point estimate reported in this meta-analysis was small and of uncertain clinical significance,” they wrote.
The associations between SGLT2 inhibitors and haemoglobin levels were maintained in subgroup analyses where participants were split by treatment duration, age, disease type, geographic region, study quality, and sample size.
However, significant interactions were observed for SGLT2 and haematocrit for treatment duration and disease type. The mean difference in haematocrit between active treatment and control was 1.96% (1.71%-2.20%) for short-term treatments and 2.67% (2.40%-2.94%) for long-term treatments, and 1.82% (1.62%-2.03%) for participants with type 1 diabetes and 2.52% (1.89%-3.16%) for participants with cardiovascular disease.
Related
With respect to secondary outcomes, SGLT2 inhibitors increased EPO levels (group difference and 95% CI 2.11mIU/mL, 0.50mIU/mL to 3.71mIU/mL) and the concentration of red blood cells (0.20 x 1012/L, 0.17 x 1012/L to 0.22 x 1012/L) but lowered ferritin (-15.05ng/mL, -22.75ng/mL to -7.36ng/mL) and transferrin saturation levels (-0.84%, -1.21ng/mL to -0.47ng/mL) compared to the control group. There was no association between SGLT2 use and changes in hepcidin (-9.16ng/mL, -18.49 ng/mL to 0.17 ng/mL) or serum iron (-0.20µmol/L, -0.46µmol/L to 0.06µmol/L) levels.
“This meta-analysis confirms quantitative heterogeneity between clinical trials, but on average showed significant increases in EPO and RBC concentration,” said the researchers. “Changes in iron-related biomarkers showed between-study heterogeneity, but mean values for ferritin and transferrin saturation were significantly reduced, suggesting increased iron utilisation.”
More than half the trials included in the systematic review and meta-analysis were deemed to be at low risk of bias (n = 48, 56.5%), with the remaining 37 showing some signs of bias. The certainty of evidence for the two primary outcomes was deemed to be moderate when evaluated using the Grading of Recommendations Assessment, Development and Evaluation (GRADE) framework.
The researchers identified numerous pathways that could explain the links between SGLT2 inhibitor use and changes in haemoglobin, haematocrit, and other markers of iron metabolism and erythropoiesis.
“Although an early rise in haemoglobin and haematocrit may partly reflect plasma volume contraction due to the diuretic effect of SGLT2 inhibitors, sustained increases may be due to enhanced erythropoiesis rather than haemoconcentration alone,” they said. “This is suggested by the higher EPO concentrations that were observed following initiation of SGLT2 inhibitors, potentially reflecting restoration of kidney EPO-producing cell function. EPO is a key regulator of red blood cell production and has been associated with biologic pathways involved in iron homeostasis, including hepcidin regulation.”
“In the present analysis, higher red blood cell concentration, haemoglobin, and haematocrit were accompanied by lower ferritin and transferrin saturation concentrations, whereas serum iron remained unchanged. This pattern in iron-related parameters is compatible with mobilisation and utilisation of stored iron to support erythropoiesis rather than depletion of circulating iron. Improvements in endothelial function and red blood cell deformability with SGLT2 inhibitor therapy may promote erythrocyte survival and prolong lifespan.”
Despite this, the researchers were unclear of the clinical significance of their findings and concluded that further studies were needed.
Writing in an accompanying editorial, Professor Kai-Uwe Eckardt, a nephrologist from Charité-Universitätsmedzin Berlin in Germany, and Dr Mark Koury, an internist and researcher from Vanderbilt University Medical Center in Tennessee, US, felt the small increases associated with SGLT2 inhibitor use could play a meaningful role in anaemia management.
“The current view is that anaemia should be managed cautiously with these agents, primarily guided by symptoms rather than laboratory values, aiming to avoid transfusions and targeting haemoglobin values below 11.5g/dL,” the editorial read. “The large dataset for SGLT2 inhibitors demonstrates clearly that cardiovascular and kidney benefits of this class of drugs occur along with the associated modest rise in haemoglobin.”
“In fact, mediation analyses provide statistical evidence that the rise of haemoglobin is the strongest mediator of these benefits.”
However, the pair cautioned that SGLT2 inhibitors could increase the likelihood of erythrocytosis, which could prove problematic for patients with heart failure.
“When erythrocytosis occurs, the benefits of continuing SGLT2 inhibitors should be carefully weighed against the risk for thromboembolic complications,” they said. “[The current study] observed that the increase in haemoglobin in their large cohort was dose-dependent, suggesting that SGLT2 inhibitor dose reduction or discontinuation with resumption at a lower dose may be options in patients who develop erythrocytosis.”
JAMA, 24 October 2026 (manuscript)
JAMA, 24 October 2026 (editorial)
