Does Jardiance Raise Hematocrit? Yes — by About Two Points

A November 2026 reel claimed SGLT2 inhibitors like Jardiance raise hematocrit, hemoglobin, and red blood cell production — and pointed to a 70,000-person meta-analysis as proof. The meta-analysis is real, recent, and peer-reviewed. The claim is mostly right, the mechanism is half right, and the part that matters most — whether the rise is dangerous — is where the reel stops short.

Evidence review · November 2026 · Prompted by a reel from @ddtmethod (Nov 5, 2026)

01The claim, as stated

In a reel posted November 5, 2026, @ddtmethod argued that SGLT2 inhibitors like Jardiance (empagliflozin) can increase hematocrit, hemoglobin, and red blood cell production. The proposed mechanism: by changing how the kidneys handle glucose, sodium, and oxygen, these drugs appear to stimulate EPO — the hormone that tells bone marrow to make more red blood cells — with SIRT1 activation possibly playing a role. The reel cited a large 2026 meta-analysis of over 70,000 participants reporting increases in hematocrit, hemoglobin, and EPO.

Three claims, three verdicts:

• Claim 1 — SGLT2 inhibitors raise hematocrit, hemoglobin, and red-cell production: VERIFIED (with the qualifier "modest").
• Claim 2 — mechanism via kidney glucose/sodium/oxygen handling → EPO, with SIRT1 possibly involved: PARTLY TRUE. The kidney/EPO part is well supported; SIRT1 is hypothesis only.
• Claim 3 — a large 2026 meta-analysis of >70,000 participants reported the rises: VERIFIED. It is real, peer-reviewed, and published in JAMA.

One honesty note up front: the reel's spoken audio was not independently transcribed — the claims above are as stated in the reel's caption. Nothing in this article is about the reel's tone or any claim beyond those three.

02The 2026 meta-analysis

The design is strong: 85 randomized, double-masked trials, 70,543 participants (mean age 62.4; 35.9% women) with cardiovascular disease, diabetes, heart failure, and/or chronic kidney disease. Median follow-up was 24 weeks. 78 trials were placebo-controlled; 7 used an active comparator. GRADE certainty for the primary outcomes was rated moderate. 48 of 85 trials were low risk of bias, 37 had some concerns, none were high risk.

Pooled differences versus control:

• Hemoglobin: +0.60 g/dL (95% CI 0.55–0.66)
• Hematocrit: +2.36 percentage points (95% CI 2.19–2.52)
• EPO: +2.11 mIU/mL (95% CI 0.50–3.71) — based on only 8 trials, with significant heterogeneity
• Red blood cell concentration: +0.20 ×10¹²/L (95% CI 0.17–0.22)
• Ferritin: −15.05 ng/mL and transferrin saturation: −0.84% — consistent with stored iron being used to make new cells
• Hepcidin and serum iron: no significant change

Two caveats from the paper itself. First, red blood cell mass was never measured directly — the erythropoiesis conclusion rests on EPO, reticulocytes, red cell count, and iron indices, not a gold-standard measurement. Second, the authors conclude the increases are "small and clinical significance remains uncertain." The accompanying JAMA editorial notes the 2.36-point hematocrit rise is smaller than the roughly 5–11-point rise seen with erythropoiesis-stimulating agents.

For scale: baseline hematocrit in the pooled population was about 41.5%, so the average person on an SGLT2 inhibitor lands around 43.7% — a move, not a transformation.

03Mechanism: two effects, not one

The reel's framing — new red blood cells, driven by EPO — is half the story. The consensus in 2024–2026 reviews is that both hemoconcentration and true erythropoiesis contribute, and hemoconcentration alone cannot explain the effect. The evidence that separates them:

• The time course doesn't match diuresis. The increase in urine output after starting an SGLT2 inhibitor is acute and transient — it normalizes within about a week. Hemoglobin, by contrast, rises gradually, peaks around 4 months, and plateaus by about 48 weeks.
• EPO rises, then settles at a new set point. A marked EPO rise occurs rapidly after initiation, then partially subsides during long-term therapy as a new equilibrium between EPO and the higher hemoglobin is established. In the EMPA-HEART mechanistic substudy, empagliflozin raised EPO by 3.86 mIU/mL versus placebo at 1 month and 1.91 at 6 months.
• Cell numbers go up, not just concentration. Reticulocyte counts rise, and red cell concentration rose in the meta-analysis — a rise in cell number, not just thicker blood.
• Iron is mobilized and used. Ferritin and transferrin saturation fall while serum iron is unchanged — compatible with stored iron being mobilized to make new red cells. In a 52-person randomized trial, dapagliflozin lowered hepcidin and ferritin and raised erythroferrone (the hepcidin inhibitor) with a transient EPO rise.
• Plasma-volume contraction is real but partial. In EMPA-REG OUTCOME, hematocrit rose from ~41% to ~44% in the first 12 weeks, corresponding to an estimated ~7% fall in plasma volume. That early component is hemoconcentration; the persistence for months to years after urine output normalizes is not.

The proposed kidney mechanism, as the reel describes it, is broadly accurate: SGLT2 blockade shifts glucose and sodium reabsorption downstream, increasing oxygen consumption in the deeper kidney and lowering local oxygen tension; oxygen sensors activate HIF-2α in peritubular fibroblasts, stimulating EPO synthesis. Reduced metabolic stress may also "rejuvenate" EPO-producing cells, and hepcidin suppression improves iron availability. Two of the four proposed sub-mechanisms are contested — SGLT2 inhibitors do not measurably alter regional tissue oxygen tension in the non-diabetic kidney, and acetazolamide, which also increases distal sodium delivery, does not stimulate EPO — which is why a hepatic contribution is also proposed.

04SIRT1: hypothesis, not proof

The SIRT1 connection comes principally from Milton Packer's reviews (European Heart Journal 2023; Cardiovascular Research 2021). Packer lists SIRT1 as one of four proposed hypotheses: SGLT2 inhibitors induce a fasting-like, nutrient-deprivation signal that up-regulates SIRT1; in hepatic cell lines, SIRT1 can directly activate HIF-2α by deacetylation, and via PGC-1α binding to hepatic nuclear factor 4 promote EPO gene transcription. SGLT2 inhibitors have been shown to stimulate SIRT1 and downstream effectors in a broad range of tissues — heart, kidney, liver — in experimental animal models.

But Packer's own figure marks these effects with dotted lines for effects "demonstrated in tissues other than the liver or kidney, and not fully evaluated in hepatic or renal cells." No randomized human study has measured SIRT1 activity and shown it mediates the EPO or hematocrit rise. The Yang 2026 meta-analysis does not test SIRT1. A 2025 review repeats the SIRT1 → HIF-2α pathway as a proposed mechanism, citing Packer. Plausible and mechanistically coherent — but it should be presented as speculation, not as an established human mechanism.

05How big, how fast, how long

• Average: hemoglobin +0.60 g/dL, hematocrit +2.36 percentage points versus control at median 24 weeks. Baseline hematocrit ~41.5% → ~43.7%.
• Early: rises are detectable within a few weeks; in EMPA-REG, hematocrit went from ~41% to ~44% in the first 12 weeks.
• Plateau: hemoglobin rises gradually and plateaus at ~48 weeks, reaching ~0.71 g/dL at 48 weeks in the meta-analysis duration analysis. The hematocrit difference was 1.96 points for short-term versus 2.67 points for long-term treatment. In EMPA-REG the plateau was sustained for the balance of the trial (~3 years).
• Dose: the effect is larger at higher doses — which is also why the JAMA editorial notes dose reduction as an option if erythrocytosis occurs.
• Distribution matters more than the mean. Most people have a small rise; a minority cross erythrocytosis thresholds. In one large cohort, erythrocytosis prevalence rose by 5.5–5.8 percentage points versus DPP-4/GLP-1 comparators; in a 9,646-patient observational study, polycythemia prevalence rose from 2.4% to 9.7%, with 1.4% severe.

The FDA Jardiance label lists the hematocrit rise as a laboratory finding: median hematocrit −1.3% on placebo versus +2.8% on empagliflozin, with 0.6% of placebo patients versus 2.7–3.5% of empagliflozin patients rising above the upper limit of normal. The US label mandates no routine hematocrit monitoring schedule; Health Canada's monograph says the clinician will monitor hematocrit among other labs.

06Does it differ by drug?

It is a class effect. There are no head-to-head randomized trials between SGLT2 inhibitors on this outcome, and the published meta-analysis does not report per-drug pooled effect sizes. A heart-failure-only meta-analysis (11 trials, 16,784 participants) found consistent hemoglobin (+0.68 g/dL) and hematocrit (+2.15%) effects across individual agents.

The only signals favoring one agent over another are weak and observational: one large cohort found empagliflozin associated with slightly higher erythrocytosis risk than dapagliflozin (adjusted odds ratio 1.16), and a small retrospective study found a significant hematocrit rise with empagliflozin but not dapagliflozin — with no significant between-drug difference in magnitude. Treat it as a class effect; any empagliflozin > dapagliflozin difference is small and from observational data.

07Does the rise cause clots?

At the population level, the large outcome trials show net cardiovascular benefit and no thrombotic safety signal attributable to the hematocrit rise:

• EMPA-REG OUTCOME (empagliflozin, n=7,020): major adverse cardiovascular events HR 0.86, cardiovascular death HR 0.62.
• DECLARE-TIMI 58 (dapagliflozin, n=17,160): MACE HR 0.93, with reduced heart-failure hospitalization.
• CANVAS Program (canagliflozin, n=10,142): MACE HR 0.86.
• DAPA-HF (dapagliflozin, n=4,744): cardiovascular death or worsening heart failure HR 0.74.
• Pooled outcome-trial meta-analyses show ~10% lower MACE, ~30% lower heart-failure hospitalization, no significant effect on MI or stroke individually, and no venous-thromboembolism signal.

The picture is less settled for the small subgroup with marked or pre-existing erythrocytosis, where observational studies disagree:

• Reassuring: Lewis et al., JAMA Network Open 2025 (137,552 new users) — new-onset erythrocytosis on an SGLT2 inhibitor was not associated with MI (HR 0.92), venous thromboembolism (HR 1.56, not significant), or stroke (HR 1.26, not significant). A Danish nationwide heart-failure register study found erythrocytosis incidence 4.1× higher on SGLT2 inhibitors but no association with 1-year thromboembolic risk (HR 0.85).
• Cautionary: a TriNetX study (Scientific Reports 2025) found that among SGLT2-inhibitor users, those who developed erythrocytosis had substantially higher risks — in men, stroke/TIA HR 1.89, MI 2.49, venous thrombosis 2.09. This is observational electronic-health-record data with likely residual confounding (people who develop erythrocytosis differ in smoking, hypoxia, and comorbidity) and it conflicts with Lewis et al. It cannot prove causation. A post hoc analysis of CANVAS + CREDENCE found thrombotic risk concentrated in men with erythrocytosis at baseline, not women.

08The testosterone stacking problem

Testosterone alone raises hemoglobin by ~0.8 g/dL and hematocrit by ~3 percentage points in pooled placebo-controlled trials — so the two drugs' average effects point the same way and add up. The observational evidence that the combination stacks is consistent:

• Gosmanov et al., 2024 (US Veterans, n=53,971 men with type 2 diabetes): combination SGLT2 inhibitor + testosterone versus SGLT2 inhibitor alone, adjusted odds ratio 3.80 for new hematocrit >54% within a year; versus testosterone alone, OR 2.49. Overall 1.4% developed erythrocytosis. Route of testosterone (topical vs injectable) did not modify the risk.
• Earlier VA analysis (n=9,402): concomitant testosterone associated with erythrocytosis with adjusted OR 7.26 — a large relative odds on a low absolute risk (0.65% of the cohort overall).
• Clalit Health Services study, 2025 (n=5,235 men with hypogonadism on testosterone): SGLT2-inhibitor users had OR 4.85 for hematocrit >54% compared with testosterone users not on an SGLT2 inhibitor.

No randomized trial and no thrombosis-outcome data exist for the combination. The same risk factors recur across the studies — male sex, smoking, sleep apnea, hypoxia, high baseline hematocrit — so the stacking is not unique to testosterone, but testosterone is the best-quantified co-factor. Reviews recommend reassessing the indication, addressing modifiable risks (smoking, sleep apnea, obesity), and considering a switch to transdermal testosterone and/or a lower dose. Venesection is not routinely recommended except for clonal erythrocytosis.

Try it: the stacking projector

The averages above add up. This projector applies the pooled numbers — SGLT2 inhibitor +2.36 points (Yang et al. 2026), testosterone ~+3 points (pooled placebo-controlled trials) — to a baseline hematocrit you set. It is arithmetic on published averages, not a prediction for any individual.

09Bottom line

The reel's central claims are substantially correct in direction and magnitude, and the cited meta-analysis is real, recent, and peer-reviewed. The main overstatement risks are threefold: presenting SIRT1 as an established mechanism rather than a hypothesis; implying the hematocrit rise is entirely "new red blood cells" when early hemoconcentration contributes; and implying that a ~2.4-point hematocrit rise is large or dangerous per se. On average it is small, of uncertain clinical significance per the authors, and the outcome trials show net benefit. Risk, if any, is concentrated in the minority who develop marked erythrocytosis — especially men with pre-existing erythrocytosis or on testosterone. That is the group where monitoring actually matters.