FLOW: Semaglutide and the Kidneys
Last verified May 2026 · Phase 3b · Halted early because the benefit was clear; headline results reported May 2024 · NCT03819153 ↗
Chronic kidney disease in diabetes has few treatments that change its trajectory, and the ones that exist are largely blood-pressure and SGLT2 drugs. FLOW tested whether a GLP-1 belongs on that list. The population was specifically selected for kidney disease — either reduced eGFR with albuminuria, or heavy albuminuria with preserved function — rather than for weight, and the endpoint was hard: kidney failure and death, not a laboratory marker. Stopping early for efficacy is uncommon and is the strongest procedural signal a trial can produce.
- Enrollment
- 3,533
- Duration
- A median 3.4 years of follow-up, cut short at a planned interim look — the original plan ran to five years
- Drug
- Semaglutide 1.0 mg (Ozempic)
- Population
- Adults 18 and over whose type 2 diabetes ran at an HbA1c of 10% or under, with kidney disease at one of two defined severities: filtration moderately down alongside heavy urine protein, or filtration further down with a lower protein bar. ★ Everyone had to already be on the maximum tolerated dose of a standard kidney-protecting blood-pressure drug unless they could not take one — so this measures what semaglutide adds on top of proper care, not instead of it. Average age 66.6, 70.1% men, average HbA1c 7.8%, and about 17 years since diagnosis. At the start, just 15.6% were taking an SGLT2 inhibitor.
Primary endpoint
Kidney failure, losing half of kidney function, or death from kidney or cardiac causes — whichever came first
Treatment arm
331 of 1,767 (18.7%)
Comparator
410 of 1,766 (23.2%)
Treatment difference: Hazard ratio 0.76 (95% CI 0.66 to 0.88); P=0.0003 for superiority
★ A 24% reduction, driven by both the kidney components and by fewer cardiac deaths. In plain terms: about 20 people treated for three and a half years to prevent one of these outcomes, in a group where these outcomes are common.
Secondary endpoints
| Endpoint | Treatment | Comparator | Difference |
|---|---|---|---|
| How fast kidney function declined each year ★ Roughly a third slower. For someone whose filtration is falling toward dialysis, slowing the slope is the whole game — it is measured in years of independence, not in a lab value. | −2.19 mL/min/1.73 m² per year | −3.36 mL/min/1.73 m² per year | Between-group difference +1.16 mL/min/1.73 m² per year (95% CI 0.86 to 1.47); P<0.001 |
| Heart attack, stroke or cardiac death 18% fewer, in people carrying two separate multipliers of cardiac risk at once. Kidney disease and diabetes together make this a very high-risk group, and the drug helped anyway. | 189 of 1,767 (10.7%) | 236 of 1,766 (13.4%) | Hazard ratio 0.82 (95% CI 0.68 to 0.98); P=0.029 |
| Death from any cause at all ★ 20% fewer deaths, and — unlike in most trials on this site — this one cleared the formal testing sequence. It is among the strongest mortality findings anywhere in this drug class. | 227 of 1,767 (12.8%) | 279 of 1,766 (15.8%) | Hazard ratio 0.80 (95% CI 0.67 to 0.95); P=0.01 |
| Death from a cardiac cause 29% fewer, the single largest contributor to both the headline result and the mortality finding. | 123 of 1,767 (7.0%) | 169 of 1,766 (9.6%) | Hazard ratio 0.71 (95% CI 0.56 to 0.89) |
| Kidney outcomes only, with cardiac death removed ★ The check that matters if you are reading this for kidney reasons: strip out cardiac death and the kidney benefit is still there, at 21%. The headline is not simply a heart result wearing a kidney label. | Reported in publication with hazard ratio 0.79 | Reference | Hazard ratio 0.79 (95% CI 0.66 to 0.94) |
| Protein leaking into the urine, at two years Down about 40% against 8% on placebo — the earliest and most sensitive sign that kidney damage is slowing. | Ratio to baseline 0.60 (40% reduction) | Ratio to baseline 0.92 (8% reduction) | Estimated treatment ratio 0.66 (95% CI 0.61 to 0.71) |
| The rate of decline after the first 12 weeks, once the early dip has passed ⚠ This row exists to answer a real objection. Kidney drugs often cause a small immediate drop in filtration that looks like harm and is not. Excluding that initial dip, the long-term decline is still meaningfully slower — so the benefit is genuine and not an artifact of where the line starts. | Approximately −2.4 mL/min/1.73 m² per year | Approximately −3.3 mL/min/1.73 m² per year | Between-group difference roughly +0.9 mL/min/1.73 m² per year |
Adverse events
| Event | Treatment rate | Comparator rate |
|---|---|---|
| Serious adverse events of any kind ★ Fewer on the drug than on placebo, because the drug prevented more cardiac and kidney events than it caused problems. | 49.6% (876/1,767) | 53.8% (950/1,766) |
| Stopped the drug for good over a side effect A modest excess — 13.2% against 11.9% — mainly stomach and bowel trouble while the dose was climbing. | 13.2% (233/1,767) | 11.9% (210/1,766) |
| Stomach and bowel problems Nausea, vomiting, diarrhea and reduced appetite, all more common than on placebo and matching what this drug does at the 1.0 mg dose everywhere else. | Higher with semaglutide (nausea, diarrhea, vomiting, constipation; class-typical profile) | Lower |
| Sudden kidney injury ★ Lower on the drug, not higher — the result that most needed checking, since vomiting and diarrhea in someone with failing kidneys is exactly how sudden injury happens. | Numerically lower with semaglutide despite GI losses | Higher with placebo |
| Diabetic eye disease getting worse ★ No difference between the arms, in contrast to the eye signal seen in an earlier semaglutide trial. Eye treatment was required by the protocol here, which may be part of why. | 4.0% (71/1,767) | 4.1% (72/1,766) |
| Blood sugar dropping dangerously low Rare and equal in both arms. Semaglutide does not force insulin out when glucose is already normal. | Low and similar between arms | Low and similar between arms |
| Pancreatitis, confirmed by independent review No increase — in fact numerically lower than placebo. | 0.6% | 1.0% |
| Cancers of any kind Identical in both arms over the trial period. | 5.4% | 5.5% |
Subgroup analyses
- How much kidney function people had left when they started: The same benefit whether filtration was moderately or severely reduced
★ The trial deliberately recruited across the full range so this question could be answered. It means the drug is not only for people caught early.
- Whether someone was already taking an SGLT2 inhibitor, the other kidney-protecting class: Similar benefit either way
★ The most practically useful subgroup here: the two classes appear to add rather than overlap, which supports using both. ⚠ Only 15.6% were on one, so this is directionally reassuring rather than settled.
- How much protein was leaking into the urine at the start: Consistent benefit across every band of severity
Not concentrated in the worst cases — people across the range benefited.
- Whether blood sugar was reasonably controlled at the start: Same benefit whether it was or was not
★ Direct evidence that the kidney protection is not just a consequence of better blood sugar. Something else is doing the work.
- Under 65 against 65 and over: Consistent reduction in both
Older participants carry the highest absolute risk, so the same relative benefit means more prevented events.
- Women against men: Similar in both
⚠ Women made up under a third of the trial, well below their share of this disease. The finding holds directionally; it rests on fewer participants.
Clinical significance
FLOW drove the January 2025 FDA label expansion for Ozempic to include kidney-disease progression, and it is the reason nephrologists now discuss this class at all. For readers, its practical use is coverage: a kidney indication is concrete and documented in a way obesity often is not. ⚠ The dose was 1.0 mg, the diabetes dose, in a population with both diabetes and established CKD. It does not establish kidney benefit for people without diabetes, and it says nothing about compounded semaglutide.
Who sells semaglutide, and for how much
This trial tested Semaglutide 1.0 mg (Ozempic). Across the 556 sellers on this register, 283 publish a standing monthly cash price for compounded semaglutide, from $79 a month, with a median of $174.
⛔ None of those sellers was in this trial. A compounded preparation is not the product studied here: no agency has reviewed that specific preparation for safety, effectiveness or manufacturing quality, and its concentration is set by whichever pharmacy filled it. The result above is the strongest evidence available for semaglutide and it was generated on something else. Every seller, with its price record.
Frequently Asked Questions
References
- 1.Perkovic V, Tuttle KR, Rossing P, et al.; FLOW Trial Committees and Investigators. Effects of Semaglutide on Chronic Kidney Disease in Patients with Type 2 Diabetes. N Engl J Med. 2024. PMID: 38785209.
- 2.U.S. National Library of Medicine. A Research Study to See How Semaglutide Works Compared to Placebo in People With Type 2 Diabetes and Chronic Kidney Disease (FLOW). ClinicalTrials.gov, NCT03819153. 2024. https://clinicaltrials.gov/study/NCT03819153
- 3.U.S. Food and Drug Administration. FDA Approves Treatment to Reduce Risk of Serious Kidney and Cardiovascular Events in Adults With Chronic Kidney Disease and Type 2 Diabetes (Ozempic label expansion). FDA Drug Safety and Availability Announcement, January 28, 2025. 2025. https://www.fda.gov/drugs/news-events-human-drugs/fda-approves-treatment-reduce-risk-serious-kidney-and-cardiovascular-events-adults-chronic-kidney
- 4.Marso SP, Bain SC, Consoli A, et al.; SUSTAIN-6 Investigators. Semaglutide and Cardiovascular Outcomes in Patients with Type 2 Diabetes. N Engl J Med. 2016. PMID: 27633186.
- 5.Lincoff AM, Brown-Frandsen K, Colhoun HM, et al.; SELECT Trial Investigators. Semaglutide and Cardiovascular Outcomes in Obesity without Diabetes. N Engl J Med. 2023. PMID: 37952131.