SGLT2 inhibitor use associated with small rises in haemoglobin and haematocrit, 85 trials

The study

This is a systematic review and meta-analysis of randomised clinical trials, published in JAMA. The authors searched PubMed, Web of Science and CENTRAL from inception to July 22, 2026. They included randomised, double-masked trials that compared an SGLT2 inhibitor with placebo or an active comparator. Participants had one or more of these conditions: cardiovascular disease, diabetes, heart failure and chronic kidney disease.

Three reviewers extracted data independently. Results were pooled with random-effects meta-analysis. The primary outcomes were changes in haemoglobin and haematocrit. Secondary outcomes were red blood cell concentration, erythropoietin (EPO), serum iron, ferritin, transferrin saturation and hepcidin.

What they found

The analysis included 85 trials and 70 543 participants. Median trial duration was 24 weeks (IQR 12-32). Mean age was 62.4 years, and 25 334 participants (35.9%) were female. Mean baseline haemoglobin was 13.6 g/dL. Seventy-eight trials (67 208 participants) used placebo. Seven trials (3335 participants) used active comparators, most often sitagliptin (2 trials, 1158 participants).

Haemoglobin: It rose by 0.55 g/dL in the SGLT2 inhibitor group and fell in the control group (-0.08 g/dL). The difference was 0.60 g/dL (95% CI 0.55 to 0.66; P < .001).

Haematocrit: It rose by 2.17% with SGLT2 inhibitors. In the control group it fell (-0.13%), which was not statistically significant (P = .07). The difference was 2.36% (95% CI 2.19% to 2.52%; P < .001).

Secondary outcomes: Four of the 6 showed significant associations with SGLT2 inhibitors. EPO rose by 2.35 mIU/mL with SGLT2 inhibitors, against 0.04 mIU/mL in controls (difference 2.11 mIU/mL; 95% CI 0.5-3.71 mIU/mL; P = .01). Transferrin saturation fell with SGLT2 inhibitors (-2.2%), against -0.10% in controls (difference -0.84%; 95% CI -1.21% to -0.47%; P < .001). The abstract does not name the other two significant outcomes or the two non-significant ones.

How much weight to give it

The design is a meta-analysis of randomised, double-masked trials, and the sample is large. The direction of the haemoglobin and haematocrit results is clear.

The authors call the increases small and say their clinical significance is unclear. That caution is fair. These are laboratory markers, not clinical outcomes. The abstract does not report anaemia rates, transfusions, symptoms or other patient-centred outcomes.

The median trial duration was 24 weeks, so longer-term patterns are not described. The abstract also does not report heterogeneity between trials, risk of bias, funding, or results by individual drug, dose or patient group. The trials covered several conditions, so one pooled figure may not suit every patient group. Most trials used placebo, and only 7 used active comparators.

For your clinic

This review suggests that a modest rise in haemoglobin and haematocrit is seen in trials of SGLT2 inhibitors. Iron-related markers also moved, with a fall in transferrin saturation. Clinicians may want to keep this in mind when they read blood counts and iron tests in patients taking these drugs.

The abstract does not show that these changes help or harm patients. It does not say whether they alter the diagnosis or treatment of anaemia. Do not change practice on this evidence alone. Read the full paper, and check current guidelines, before using these findings in decisions about individual patients.