FLOW is the trial that put a GLP-1 receptor agonist on the short list of drugs shown to slow kidney disease, and it was stopped before it was scheduled to finish because the effect was already there. It is also the trial most often cited to people it never studied. Its population had type 2 diabetes and established albuminuric kidney damage; almost nobody buying a weight-loss prescription from the semaglutide sellers has either. The clinical picture of GLP-1 drugs and the kidney is in the kidney article; what follows is what the trial itself did, and how far its result travels.
The endpoint, and the stop
Participants were assigned to subcutaneous semaglutide 1.0 mg weekly or placebo. The primary outcome was major kidney disease events, a composite of the onset of kidney failure — defined as dialysis, transplantation, or an eGFR below 15 ml per minute per 1.73 m² — at least a 50% reduction in eGFR from baseline, or death from kidney-related or cardiovascular causes. Confirmatory secondary outcomes were prespecified and tested hierarchically.[1]
Among the 3,533 participants randomized, 1,767 to semaglutide and 1,766 to placebo, median follow-up was 3.4 years, reached after early trial cessation was recommended at a prespecified interim analysis. The risk of a primary-outcome event was 24% lower on semaglutide, at 331 against 410 first events: hazard ratio 0.76 (95% CI, 0.66 to 0.88; P = 0.0003).[1]
The interim stop is worth naming rather than skipping. It means the published estimate rests on the events accumulated in the first part of the planned follow-up rather than all of it, which is the price of halting a trial once a benefit is established. That is a reasonable trade in a disease with few effective treatments, and it is a different kind of evidence from a trial that ran to its planned end.
The hierarchy then held. The mean annual eGFR slope was less steep by 1.16 ml per minute per 1.73 m² (P < 0.001), major cardiovascular events were 18% lower (hazard ratio 0.82; 95% CI, 0.68 to 0.98; P = 0.029) and death from any cause 20% lower (hazard ratio 0.80; 95% CI, 0.67 to 0.95; P = 0.01). Serious adverse events were reported in 49.6% of the semaglutide group against 53.8% on placebo.[1] Every confirmatory outcome cleared its test in order, which is what makes the secondary findings assertions rather than observations.
What a hierarchy buys, and what it costs
Hierarchical testing is the reason the secondary results above can be stated as findings rather than as observations. Each confirmatory outcome is tested only if the one before it succeeded, so the family-wise error rate stays controlled and no individual P value has to be discounted for multiplicity. FLOW's chain held end to end, which is uncommon and is the strongest structural feature of the trial.[1]
The cost is that anything outside the chain is exploratory, however interesting. A subgroup, a post hoc slice or an unlisted outcome carries no such protection, and the difference between a confirmatory result and an exploratory one is usually invisible in the way either gets quoted. That distinction decides how much weight the next section can bear.
The subgroup that complicates it
SGLT2 inhibitors independently reduce kidney and cardiovascular events, so the question of whether semaglutide adds anything on top of one is the question a nephrologist actually faces. A prespecified analysis split FLOW by baseline SGLT2 inhibitor use: 550 participants were taking one, 2,983 were not.[2]
Among those already on an SGLT2 inhibitor, the primary outcome occurred in 41 of 277 on semaglutide against 38 of 273 on placebo — hazard ratio 1.07 (95% CI, 0.69 to 1.67; P = 0.755). Among those not taking one, it occurred in 290 of 1,490 against 372 of 1,493 — hazard ratio 0.73 (95% CI, 0.63 to 0.85; P < 0.001). The interaction P value was 0.109.[2]
Neither figure should be quoted alone. A point estimate above 1.0 in 550 people, with a confidence interval running from 0.69 to 1.67 and a nonsignificant interaction, is consistent with a benefit as large as the one seen elsewhere in the trial and equally consistent with none. The eGFR slope difference pointed the same way in both subgroups, at 0.75 (95% CI, −0.01 to 1.5) with an SGLT2 inhibitor and 1.25 (95% CI, 0.91 to 1.58) without, interaction P = 0.237, and the cardiovascular and all-cause death benefits were similar regardless of use.[2] The analysis states its own limit plainly: power was insufficient to detect smaller but clinically relevant effects.
What it says about people without diabetes
Nothing, directly. Type 2 diabetes was an entry requirement, and both eGFR routes into the trial demanded measurable albumin in the urine. Every hazard ratio above describes slowing a disease that is already underway in people whose glucose is already abnormal, which is the distinction drawn in the two-indications article.
The nearest available answer comes from the cardiovascular outcomes trial in people without diabetes. Its prespecified kidney composite — death from kidney disease, initiation of chronic kidney replacement therapy, onset of persistent eGFR below 15, persistent reduction in eGFR of at least 50%, or onset of persistent macroalbuminuria — occurred in 1.8% on semaglutide against 2.2% on placebo: hazard ratio 0.78 (95% CI, 0.63 to 0.96; P = 0.02).[3]
Read the composite before reading the ratio. It counts onset of persistent macroalbuminuria, a laboratory finding rather than an event a person experiences, and FLOW's primary outcome does not. The absolute gap is 0.4 percentage points over roughly four years, which corresponds to about 250 people treated for one composite event avoided — and a meaningful share of those events are a changed urine test. The filtration-rate benefit is more tangible where kidneys are already struggling: 0.75 ml per minute per 1.73 m² at 104 weeks overall (95% CI, 0.43 to 1.06), rising to 2.19 (95% CI, 1.00 to 3.38) in participants whose baseline eGFR was below 60.[3] The rest of that trial is covered in the SELECT article.
Pooling the three trials
A 2026 prespecified pooled analysis combined participant-level data from three semaglutide trials at three different doses and routes: 1.0 mg subcutaneous in chronic kidney disease, 2.4 mg subcutaneous in atherosclerotic cardiovascular disease without diabetes, and 14 mg oral. Across 30,787 participants with a mean follow-up of 39.5 to 47.5 months, the kidney composite occurred as a first event in 973 on semaglutide against 1,134 on placebo, a hazard ratio of 0.84 (95% CI, 0.77 to 0.91).[4]
A narrower composite excluding cardiovascular death — the version that answers a kidney question rather than a mixed one — gave 347 events against 416, hazard ratio 0.80 (95% CI, 0.69 to 0.92). The authors note the analysis pools trials that differed in baseline characteristics, dose and route, and conclude that the benefit may not be explained by glycemic or weight effects alone.[4]
The reader who does match the trial
A minority of people shopping for a compounded prescription do have type 2 diabetes and reduced kidney function, and for them FLOW is the most relevant document on this site. It is also the strongest argument against buying from a checkout. The trial's benefit was measured at 1.0 mg weekly, a diabetes dose that weight-loss plans are not selling, on a background of standard care that included blood-pressure and albuminuria management. Reproducing the trial means reproducing the regimen, not the molecule.
Kidney function also changes what a titration should look like, because the vomiting and diarrhea that accompany a fast escalation are the route by which this class causes harm rather than benefit — the schedule in the titration article climbs slowly for that reason. An eGFR and a urine albumin-to-creatinine ratio are inexpensive tests, and an intake that asks for neither has not collected what this decision requires.
The population gap, stated plainly
Everyone in all three trials was already sick. Entry required type 2 diabetes with albuminuric kidney disease, or established atherosclerotic cardiovascular disease. A 39-year-old with a body-mass index of 32, a normal creatinine and no cardiac history is not in any of them, and that person is the typical customer of a cash-pay telehealth prescription.
What the evidence supports for that reader is narrower and worth stating without inflation: semaglutide slows kidney disease in people who have it, and in a population with heart disease it modestly reduced a composite that leans on a urine marker. It does not establish that the drug protects a healthy kidney, and no trial has asked. Nor has any kidney outcome trial reported for tirzepatide, so the comparison in the molecule article has nothing to weigh on this axis.
The dose is the other gap. FLOW used 1.0 mg weekly, the diabetes dose, not the 2.4 mg quoted for weight management, and the two results above came from different doses of the same molecule. And every one of these trials studied branded, FDA-approved product under supervision. Compounded semaglutide is not FDA-approved, and compounded drugs are not reviewed by the FDA for safety, efficacy or quality before they are dispensed — the difference is set out in the compounded-versus-brand article. How each figure here was established is described in the methodology.