What was the primary endpoint in the FLOW trial?

A five-component composite: onset of kidney failure, sustained ≥50% decline in eGFR from baseline, sustained eGFR <15 mL/min/1.73 m², death from kidney causes, or death from cardiovascular causes. The sustained eGFR thresholds required confirmation at least 30 days later.

Why does a kidney trial include cardiovascular death in its primary endpoint?

Cardiovascular death is included to avoid informative censoring. If semaglutide prevents a CV death, that patient survives long enough to potentially reach kidney failure, which could mask a kidney benefit. Including CV death in the composite prevents this statistical distortion.

Was the FLOW trial stopped early?

Yes. An independent data monitoring committee recommended stopping after the pre-specified interim analysis showed efficacy crossing the O'Brien-Fleming boundary. Approximately 331 primary-endpoint events had occurred at the time of the recommendation.

How does early stopping affect the reliability of FLOW's results?

Trials stopped early for benefit tend to overestimate treatment effects by an average of roughly 29% based on meta-research. FLOW had accumulated a reasonable number of events (331), which provides more stability than very early stops, but the true long-term effect may be modestly smaller than HR 0.76.

Were patients on SGLT2 inhibitors included in FLOW?

Yes, but only about 15.6% at baseline. SGLT2 inhibitor use was permitted, not required. Subgroup analyses suggested consistent semaglutide benefit regardless of SGLT2i use, though the SGLT2i subgroup was small and not powered for definitive conclusions.

What dose of semaglutide was used in FLOW?

Subcutaneous semaglutide 1.0 mg once weekly, the same dose approved for T2D under the brand name Ozempic. The 2.4 mg dose used for obesity (Wegovy) was not tested in this trial.

Does FLOW apply to patients without type 2 diabetes?

No. All FLOW participants had T2D. Unlike DAPA-CKD and EMPA-KIDNEY, which enrolled patients with and without diabetes, FLOW's results are specific to the T2D-plus-CKD population.

What is the estimand framework and why does it matter for FLOW?

The ICH E9(R1) estimand framework requires trials to pre-specify how intercurrent events (like stopping the drug or starting rescue therapy) are handled. FLOW reported both a treatment-policy estimand (HR 0.76, reflecting all randomized patients regardless of adherence) and a trial-product estimand (HR 0.69, estimating the effect if all patients remained on drug).

Did semaglutide cause an initial drop in eGFR like SGLT2 inhibitors do?

Yes. A small acute hemodynamic dip in eGFR occurred in the first weeks, followed by a slower chronic decline. This pattern, seen with SGLT2 inhibitors as well, is thought to reflect reduced glomerular hyperfiltration and is considered a sign of the drug's mechanism rather than kidney damage.

Can FLOW's results be combined with SGLT2 inhibitor data to support dual therapy?

FLOW provides suggestive but not definitive evidence. The subgroup analysis by SGLT2i use was consistent, but underpowered. No dedicated trial has randomized patients to semaglutide plus an SGLT2 inhibitor vs. SGLT2 inhibitor alone with a kidney composite endpoint. Until that trial exists, dual use is a clinical judgment call rather than a protocol-grade recommendation.

Inside the FLOW Methodology: What Most Summaries Skip

Q: Why does a kidney trial include cardiovascular death in its primary endpoint?

Cardiovascular death is included to avoid informative censoring. If semaglutide prevents a CV death, that patient survives long enough to potentially reach kidney failure, which could mask a kidney benefit. Including CV death in the composite prevents this statistical distortion.

Q: Was the FLOW trial stopped early?

Yes. An independent data monitoring committee recommended stopping after the pre-specified interim analysis showed efficacy crossing the O'Brien-Fleming boundary. Approximately 331 primary-endpoint events had occurred at the time of the recommendation.

Q: How does early stopping affect the reliability of FLOW's results?

Trials stopped early for benefit tend to overestimate treatment effects by an average of roughly 29% based on meta-research. FLOW had accumulated a reasonable number of events (331), which provides more stability than very early stops, but the true long-term effect may be modestly smaller than HR 0.76.

Q: Were patients on SGLT2 inhibitors included in FLOW?

Yes, but only about 15.6% at baseline. SGLT2 inhibitor use was permitted, not required. Subgroup analyses suggested consistent semaglutide benefit regardless of SGLT2i use, though the SGLT2i subgroup was small and not powered for definitive conclusions.

Q: What dose of semaglutide was used in FLOW?

Subcutaneous semaglutide 1.0 mg once weekly, the same dose approved for T2D under the brand name Ozempic. The 2.4 mg dose used for obesity (Wegovy) was not tested in this trial.

Q: Does FLOW apply to patients without type 2 diabetes?

No. All FLOW participants had T2D. Unlike DAPA-CKD and EMPA-KIDNEY, which enrolled patients with and without diabetes, FLOW's results are specific to the T2D-plus-CKD population.

Q: What is the estimand framework and why does it matter for FLOW?

The ICH E9(R1) estimand framework requires trials to pre-specify how intercurrent events (like stopping the drug or starting rescue therapy) are handled. FLOW reported both a treatment-policy estimand (HR 0.76, reflecting all randomized patients regardless of adherence) and a trial-product estimand (HR 0.69, estimating the effect if all patients remained on drug).

Q: Did semaglutide cause an initial drop in eGFR like SGLT2 inhibitors do?

Yes. A small acute hemodynamic dip in eGFR occurred in the first weeks, followed by a slower chronic decline. This pattern, seen with SGLT2 inhibitors as well, is thought to reflect reduced glomerular hyperfiltration and is considered a sign of the drug's mechanism rather than kidney damage.

Q: Can FLOW's results be combined with SGLT2 inhibitor data to support dual therapy?

FLOW provides suggestive but not definitive evidence. The subgroup analysis by SGLT2i use was consistent, but underpowered. No dedicated trial has randomized patients to semaglutide plus an SGLT2 inhibitor vs. SGLT2 inhibitor alone with a kidney composite endpoint. Until that trial exists, dual use is a clinical judgment call rather than a protocol-grade recommendation.

By HealthRX.com Medical Team

Published May 25, 2026Updated May 25, 2026Last reviewed May 25, 2026

Clinical image for Inside the FLOW Methodology: What Most Summaries Skip Image: HealthRX.com clinical image

At a glance

| Parameter | Detail | |---|---| | N | 3,533 (1,767 semaglutide, 1,766 placebo) | | Intervention | Subcutaneous semaglutide 1.0 mg once weekly | | Comparator | Matched placebo injection once weekly | | Duration | Median 3.4 years (event-driven; stopped early May 2023) | | Primary endpoint | Composite: onset of kidney failure, sustained ≥50% eGFR decline from baseline, sustained eGFR <15 mL/min/1.73 m², renal death, or cardiovascular death | | Key result | HR 0.76 (95% CI 0.66-0.88; P = 0.0003) favoring semaglutide |

Why the Design Choices Matter More Than the Headline Number

The 24% risk reduction from FLOW is the number that gets cited. But how much weight that number should carry in clinical decision-making depends on a series of design decisions made years before the first patient was dosed. This page unpacks those decisions.

Randomization and Blinding

FLOW used a centralized, interactive web-response system for 1:1 randomization, stratified by baseline eGFR (25 to <50 vs. 50 to <75 mL/min/1.73 m²) and urine albumin-to-creatinine ratio (UACR; 300 to <1,000 vs. ≥1,000 mg/g). This two-factor stratification is important. Prior kidney outcome trials (CREDENCE, DAPA-CKD) stratified similarly, but the UACR threshold matters because patients with UACR ≥1,000 mg/g carry a steeper trajectory of kidney function decline.

Both patients and investigators were blinded. The semaglutide pen and the placebo pen were visually identical. Because semaglutide produces well-known gastrointestinal side effects, functional unblinding is a real concern. Nausea occurred in 21.4% of semaglutide patients vs. 7.0% on placebo. Whether site investigators could guess allocation based on GI complaints is not formally tested in the published paper, and this is a limitation worth flagging.

Inclusion and Exclusion: Who Was Actually in This Trial?

Eligible patients had T2D (HbA1c ≥6.5% or on glucose-lowering therapy), CKD defined by eGFR 50-75 with UACR 300-5,000 mg/g or eGFR 25-50 with UACR 100-5,000 mg/g, and were on a stable dose of a renin-angiotensin system (RAS) inhibitor (or had a documented reason for not taking one).

Three points clinicians miss:

The RAS-inhibitor requirement. Roughly 96% of patients were on an ACE inhibitor or ARB at baseline. FLOW was not testing semaglutide against nothing. It was testing semaglutide added to standard nephrology care. The result applies to the add-on scenario, not monotherapy.
The eGFR floor of 25. Patients with eGFR <25 were excluded. Whether semaglutide helps patients approaching dialysis remains unknown from this trial. The KDIGO 2024 guidelines acknowledge this gap.
SGLT2 inhibitor use was permitted but not required. At baseline, about 15.6% of participants were taking an SGLT2 inhibitor. By the end of the trial, that rose to approximately 22%. Because SGLT2 inhibitors have their own kidney-protective signal (demonstrated in DAPA-CKD and EMPA-KIDNEY), FLOW's result reflects a population where most patients were not on an SGLT2 inhibitor. The interaction analyses by SGLT2i use showed consistent benefit, but the subgroup on SGLT2 inhibitors was small.

The Primary Endpoint: What "Kidney Composite" Actually Means

FLOW's primary composite included five components:

| Component | What it measures | Events (sema vs. placebo) | |---|---|---| | Onset of kidney failure (dialysis, transplant, or sustained eGFR <15) | Hard clinical outcome | 30 vs. 50 | | Sustained ≥50% eGFR decline | Surrogate accepted by FDA | 41 vs. 64 | | Sustained eGFR <15 mL/min/1.73 m² | Overlap with kidney failure | Included in kidney failure count | | Renal death | Extremely rare | 1 vs. 3 | | Cardiovascular death | Not a kidney endpoint | 60 vs. 65 |

The inclusion of cardiovascular death in a "kidney" composite is a design choice inherited from the CREDENCE trial. It prevents a bias where a treatment that prevents CV death could appear neutral on kidney outcomes simply because patients survived long enough to reach kidney failure. But it also means the composite is not purely a measure of kidney biology. In FLOW, 60 of 331 primary-endpoint events in the semaglutide group (18%) were CV deaths. Readers should check the individual-component breakdown rather than taking the composite at face value.

The word "sustained" requires confirmation at least 30 days later. This is stricter than a single lab draw and reduces false positives from acute illness or dehydration.

The Estimand Framework

FLOW was one of the first major kidney-outcome trials to adopt the ICH E9(R1) estimand framework. The primary estimand was a "treatment policy" estimand: events counted regardless of whether the patient stopped the study drug. This is an intention-to-treat philosophy taken one step further by explicitly defining how intercurrent events (treatment discontinuation, use of rescue therapy, death) are handled.

A second "trial product" estimand attempted to estimate the effect of semaglutide if all patients had stayed on treatment. The HR for this estimand was 0.69 (95% CI 0.57-0.83), a larger effect size, which makes intuitive sense because patients who tolerate the drug benefit more than the average randomized population. Clinicians should cite the treatment-policy estimand (HR 0.76) for population-level policy questions and the trial-product estimand (HR 0.69) when counseling an individual patient who plans to stay on therapy.

Statistical Design: Event-Driven With a Single Interim Look

FLOW was designed to run until 571 primary-endpoint events accumulated. One pre-specified interim analysis was planned after approximately 50% of events. The independent data monitoring committee reviewed that interim in October 2023 and recommended stopping the trial early for efficacy. Final analyses were conducted on data collected through the cutoff.

The alpha-spending function used an O'Brien-Fleming boundary, which is conservative at the interim look and spends very little alpha early. The final two-sided significance level after early stopping was P < 0.001. A Cox proportional hazards model stratified by baseline eGFR and UACR category served as the primary analysis.

Early stopping deserves a caution flag. Trials stopped early for benefit tend to overestimate treatment effects by roughly 29% on average (Bassler et al., JAMA 2010). The FLOW steering committee was aware of this and noted that 331 events had already accumulated, which provides more statistical stability than a trial stopped after, say, 100 events. Still, the true long-term effect may be modestly smaller than the reported HR of 0.76.

The eGFR Slope Analysis: A Secondary Endpoint Worth Studying

Beyond the composite, FLOW reported total eGFR slope from baseline to week 104. Semaglutide patients lost approximately 1.16 mL/min/1.73 m² per year. Placebo patients lost approximately 2.19 mL/min/1.73 m² per year. The between-group difference was 1.16 mL/min/1.73 m² per year (95% CI 0.86-1.47; P <0.001).

What makes this interesting is the initial acute eGFR dip. Semaglutide caused a small hemodynamic reduction in eGFR during the first weeks (similar to what SGLT2 inhibitors do), followed by a slower chronic slope. The "chronic slope" analysis, which excludes the first few months, showed an even larger separation. This pattern mirrors the SGLT2 inhibitor literature and suggests a shared hemodynamic mechanism involving reduced glomerular hyperfiltration.

Limitations the Authors Acknowledged (and One They Didn't)

Acknowledged:

Early stopping limits long-term safety and durability data.
The trial was not powered for individual composite components.
Low SGLT2 inhibitor use limits conclusions about combination therapy.
The population was ≥95% on RAS blockade; results cannot be extended to patients off these agents without extrapolation.

Not explicitly discussed:

Weight loss as confounder. Semaglutide patients lost approximately 4.7 kg more than placebo. Weight loss improves blood pressure, glycemia, and albuminuria independently. FLOW did not include a mediation analysis to determine how much of the kidney benefit was a direct GLP-1 receptor effect vs. an indirect metabolic improvement from weight loss. This is a genuine gap. Until a study uses a GLP-1 receptor agonist without significant weight-loss effects (or adjusts for weight change as a time-varying covariate), the mechanistic question remains open.

How FLOW Compares to Other Kidney-Outcome Trials

| Trial | Drug class | N | Population | Primary HR | Stopped early? | |---|---|---|---|---|---| | FLOW | GLP-1 RA | 3,533 | T2D + CKD | 0.76 | Yes | | CREDENCE | SGLT2i | 4,401 | T2D + CKD | 0.70 | Yes | | DAPA-CKD | SGLT2i | 4,304 | CKD +/- T2D | 0.61 | Yes | | EMPA-KIDNEY | SGLT2i | 6,609 | CKD +/- T2D | 0.72 | Yes |

All four were stopped early for benefit, a pattern that raises the collective question of whether kidney-outcome trials in the SGLT2i/GLP-1 era are systematically over-powered or systematically over-estimated. Pragmatic registry data over the next five years will clarify.

What This Means for Prescribers

FLOW establishes semaglutide as the first GLP-1 receptor agonist with a positive kidney-outcome trial. For patients with T2D and CKD stage 3-4 who are already on RAS blockade, adding semaglutide 1.0 mg weekly reduced the composite kidney endpoint by 24%. The evidence is strongest for eGFR 25-75 with UACR ≥300 mg/g. It does not replace SGLT2 inhibitors, which have independent evidence, but it adds a second drug class with renal outcome data. Whether to use one or both is a question FLOW was not designed to answer definitively.

Frequently asked questions

›

References

Perkovic V, Tuttle KR, Gerdstein H, et al. Effects of semaglutide on chronic kidney disease in patients with type 2 diabetes. N Engl J Med. 2024;391(2):109-121. PubMed
Heerspink HJL, Stefansson BV, Correa-Rotter R, et al. Dapagliflozin in patients with chronic kidney disease. N Engl J Med. 2020;383(15):1436-1446. PubMed
Perkovic V, Jardine MJ, Neal B, et al. Canagliflozin and renal outcomes in type 2 diabetes and nephropathy. N Engl J Med. 2019;380(24):2295-2306. PubMed
Novo Nordisk. Ozempic (semaglutide) prescribing information. FDA. Label
Bassler D, Briel M, Montori VM, et al. Stopping randomized trials early for benefit and estimation of treatment effects: systematic review and meta-regression analysis. JAMA. 2010;303(12):1180-1187. PubMed
KDIGO 2024 Clinical Practice Guideline for the Evaluation and Management of Chronic Kidney Disease. Kidney Int. 2024;105(4S):S117-S314. PubMed