The GLP-1 receptor agonists are a widely-prescribed standard drug class, so their efficacy and limits on hard outcomes are a stratum-level decision in their own right (the Pharmacotherapy taper). This page holds the class-level pooled evidence — the cardiovascular-outcome-trial (CVOT) programme read together — which the single-agent pages in this cluster cannot supply: Semaglutide for Cardiovascular Risk in Obesity is one agent in one non-diabetic population (SELECT), and Semaglutide and Kidney Outcomes in Chronic Kidney Disease is one agent in one CKD population (FLOW). Badve 2024 pools 11 trials to answer the class question and, for the first time, shows the class moves a hard kidney outcome rather than a surrogate.
Scope. This is the class efficacy/limitations appraisal (IN, per the Pharmacotherapy taper: a class efficacy comparison is admissible). Per-agent selection, dosing/titration, and the management of CKD or diabetes are OUT (prescriber/frontier zone) and not covered here.
What the meta-analysis pooled
A gold-tier aggregate-data meta-analysis of RCTs (PROSPERO-registered, PRISMA, GRADE, all trials low risk of bias): 11 trials, 85,373 participants — 67,769 with type 2 diabetes across ten CVOTs, plus the 17,604 non-diabetic participants of SELECT (overweight/obese with established cardiovascular disease, added post hoc as the only completed GLP-1 CVOT without diabetes). Eligible trials needed >=500 T2D participants, a placebo comparator, >=12 months follow-up, and a primary clinical kidney or cardiovascular outcome — trials with a surrogate primary outcome, and dual GIP/GLP-1 agonists (tirzepatide), were excluded.
- Agents: semaglutide (four trials, incl. FLOW and SELECT), exenatide (two), and albiglutide, dulaglutide, efpeglenatide, liraglutide, lixisenatide (one each).
- Horizon — the studied range: median follow-up 25.2 months (range 15.9 to 64.8). The class effects below are characterized over ~2 years, not a lifetime; do not read them as lifetime effects.
- Kidney-risk context (bounds the absolute kidney benefit): mean baseline eGFR 77.2 mL/min/1.73m2, only 22.7% with eGFR <60 and 9.5% with ACR >300 mg/g — apart from FLOW, no trial enrolled high kidney risk, so kidney event rates were low.
Cardiovascular outcomes — a robust class effect, high-certainty
In participants with type 2 diabetes, versus placebo (all high-certainty evidence; with-SELECT figures in brackets, p-heterogeneity by diabetes status >0.05 throughout):
| Outcome | HR (95% CI), T2D | With SELECT added | NNT / notes |
|---|---|---|---|
| MACE (CV death, nonfatal MI, nonfatal stroke) | 0.87 (0.81-0.93), I2=49.75% | 0.86 (0.80-0.92) | NNT 74 over 25.1 mo |
| Cardiovascular death | 0.86 (0.80-0.92) | — | component |
| Nonfatal myocardial infarction | 0.90 (0.82-0.99) | — | p-het by diabetes 0.02, both arms benefit |
| Nonfatal stroke | 0.87 (0.79-0.96) | — | component |
| Hospitalisation for heart failure | reduced 13% | — | p-het 0.48 |
| All-cause death | 0.88 (0.83-0.93), I2=0% | 0.87 (0.82-0.91) | NNT 101 (T2D) / 104 (all) |
Leave-one-out analysis found no single trial drove MACE or all-cause death, and effects were consistent across all pre-specified subgroups (age, sex, BMI, CVD, eGFR, agent type, dosing frequency, follow-up duration). This is the decision-relevant fact the single-agent pages could not establish: the CV/mortality benefit is a class property spanning the T2D CVOT programme, not a semaglutide-only or obesity-only finding, and it holds regardless of diabetes status.
Kidney outcomes — the class moves a HARD outcome, not just a surrogate (the key refinement)
This is the beyond-the-predecessor move. The prior class meta-analysis (Sattar 2021, not held) reported a 21% reduction in a composite kidney outcome, but — in Badve’s own words — «this composite kidney disease outcome was mainly driven by new-onset macroalbuminuria, a surrogate outcome not validated for clinical kidney disease outcomes». Albuminuria is a risk marker, not a validated kidney-failure surrogate, especially at low albuminuria and near-normal function (Surrogate Outcomes). Badve re-runs the question on a hard composite that excludes new-onset macroalbuminuria:
| Kidney outcome (T2D) | HR (95% CI) | Certainty | NNT |
|---|---|---|---|
| Composite (kidney failure, or sustained >=50% eGFR fall, or kidney death) | 0.82 (0.73-0.93), I2=26% | high | 164 over 25.1 mo |
| Kidney failure (KRT, persistent eGFR <15, or kidney death) | 0.84 (0.72-0.99), I2=0% | moderate | — |
| Worsening kidney function (sustained >=50% eGFR fall) | 0.79 (0.68-0.92) | high | — |
The kidney failure reduction is demonstrated «for the first time». It matters because GLP-1-driven weight loss could artificially flatter creatinine-based eGFR — «GLP-1 receptor agonists cause weight loss, which may in turn reduce creatinine generation due to reduced muscle mass and therefore artificially improve creatinine-based eGFR measurements. The separately significant reduction in the risk of kidney failure provides important reassurance that the kidney benefits of GLP-1 receptor agonists are real and clinically important». Kidney failure (dialysis/transplant) is not creatinine- confoundable in that way, so its reduction anchors the composite against the muscle-loss artifact. FLOW carried 49% of the kidney-failure weight, but heterogeneity was 0%, so the direction holds across the kidney-risk range.
Absolute kidney benefit is modest. Because most trials enrolled low kidney risk, «the observed absolute risk reductions were modest with relatively higher NNT than that reported in the dedicated kidney disease outcomes trials of SGLT2 inhibitors» (kidney NNT 164 vs MACE NNT 74). The relative effect is real; the absolute payoff scales with baseline kidney risk (route (a), Baseline Risk and the Relative-Absolute Split) and is largest in the FLOW-type high-risk stratum.
The cost side — net of the drug’s own harms
The class benefit must be read net of its costs (the substitution principle: a lever’s marginal value is net of the alternative’s own harms).
- Serious adverse events: no difference — 31.7% vs 32.9%, RR 0.95 (0.90-1.01) — but I2=88.5%, so the source itself says to interpret the pooled safety numbers «carefully».
- Treatment discontinuation for adverse events is higher on drug — 12.7% vs 9.2%, RR 1.51 (1.18-1.94), I2=96.3%. This is the GI tolerability tax, and it is an adherence cost: an intervention not taken has no effect. The class page for the fuller non-cardiometabolic ledger is GLP-1 Non-Cardiometabolic Effects and Safety.
- No signal for the feared harms: acute pancreatitis, medullary thyroid cancer, pancreatic cancer, all cancers, severe hypoglycaemia, retinopathy — all similar to placebo.
- Not covered by the drug: lifelong dependency (benefit decays on stopping — see the withdrawal evidence on Semaglutide for Cardiovascular Risk in Obesity), cost, and the fact that a drug manages markers without fixing the upstream driver (structural leverage stays with the lifestyle lever).
Synthesis — Layer-1 sizing: what this shrinks, and what it does NOT
The class summary. GLP-1 receptor agonists are, on this evidence, «the first and only class of medications with proven benefits on composite kidney and cardiovascular outcomes, all individual components of composite kidney and cardiovascular outcomes, death due to any cause, and hospitalisation for heart failure across a range of cardiovascular risk and chronic kidney disease severity in people with and without diabetes» (Badve et al., 2025). For the T2D / high-CV-risk stratum this is a mature, effective, low-harm drug that captures CV-event, mortality, heart-failure and (in high kidney risk) hard-kidney benefit.
What it shrinks (Layer 1). A mature effective drug for an outcome shrinks the marginal rank of a lifestyle lever for the outcomes it covers (Layer 1 - Ranking Interventions for a Stratum). For the T2D stratum, the availability of a drug that robustly cuts MACE, CV death, all-cause death and hard kidney events reduces how much additional CV/renal/mortality benefit a lifestyle weight-loss lever must be relied on to deliver at the margin.
What it does NOT shrink — two guards.
- Guard (i): substitution is outcome-specific. The lifestyle weight-loss lever retains its other-channel value that the GLP-1 drug does not substitute — T2D remission (Total Diet Replacement and Type 2 Diabetes Remission), the all-cause-mortality benefit of intentional weight loss (Does Weight Loss Reduce Cardiovascular Events, Ma 2017), MASLD regression, function. The drug shrinks the lever’s rank only for CV events, mortality and kidney outcomes, not for these.
- The CV benefit here is largely weight-INDEPENDENT. These are glucose-lowering CVOTs with modest weight loss, and GLP-1 CV benefit separates early, before much weight is lost — it is a drug pleiotropic effect, not a weight-loss effect. So this evidence does not show that losing weight cuts CV events; it reinforces the Does Weight Loss Reduce Cardiovascular Events thesis that the route matters. (inferred from Badve et al., 2025)
The choice stays the person’s (Layer 3). Layer 1 only sizes the marginal rock; whether a person takes the drug, the lifestyle lever, or both is elicited against adherence, cost, GI tolerability and reversibility.
Independence — this is refinement (F), NOT independent backing (E)
Badve is not an independent (type-E) route to the GLP-1 CV/kidney findings held elsewhere, and no
[E-independent] claim is made anywhere on this page or in the cluster.
- Author overlap defeats it outright. Badve shares Colhoun with SELECT (Semaglutide for Cardiovascular Risk in Obesity) and shares Perkovic, Tuttle, Rossing and Mahaffey with FLOW (Semaglutide and Kidney Outcomes in Chronic Kidney Disease). A shared author defeats independence before any citation-chase.
- Trial-constituency overlap defeats it again. Badve pools SELECT and FLOW as constituent trials — a primary trial pooled into a meta-analysis is not independent of that trial.
- It is the successor to the un-held predecessor. Badve updates the 2021 GLP-1 CVOT meta-analysis (Sattar 2021, a Badve co-author) that Semaglutide for Cardiovascular Risk in Obesity flagged as an echo rather than independent backing. Badve now holds the class-level evidence that page’s caveat said was missing — this is claim-refinement across the cluster (type F), the composite beating any single-agent page alone.
Decision relevance
- T2D + high CV or kidney risk: the GLP-1 class is a strong, high-certainty lever for MACE, CV death, all-cause death, heart failure, and (at high kidney risk) hard kidney outcomes — NNT 74 for MACE and 101 for death over ~2 years. This sizes down the marginal CV/renal/mortality rank of the lifestyle weight-loss lever for this stratum, net of a GI-driven discontinuation cost (RR 1.51) and lifelong dependency.
- Non-diabetic: the class effect appears consistent regardless of diabetes status, but SELECT is the only non-diabetic trial in the pool, so the non-diabetic generalization rests on one trial — hold it as high-certainty for T2D, weaker for primary-prevention non-diabetic strata (the standing primary-prevention gap on Semaglutide for Cardiovascular Risk in Obesity).
- Kidney: the class prevents hard kidney events, not just macroalbuminuria — but the absolute benefit is modest outside high kidney risk (NNT 164), so it is decision-moving mainly in the FLOW-type stratum.
(inferred from Badve et al., 2025)
Gaps
- G (needs aggregation across a missing exposure): no dedicated GLP-1 primary-prevention CVOT in non-diabetic low-risk adults; the non-diabetic arm is SELECT alone.
- G: dual GIP/GLP-1 agonists (tirzepatide) were excluded by design, so this class effect does not transport to them — the tirzepatide CV-outcome trial is still awaited (Comparing Obesity Drugs, Semaglutide for Cardiovascular Risk in Obesity).
- G: no individual-participant data, so no baseline-characteristic subgroup analysis (e.g. heart failure phenotype).