Nobody has run the study that could answer that. The comparison everyone cites followed 295,907 adults with type 2 diabetes and found gout at 7.8 events per 1,000 person-years on a GLP-1 drug against 4.9 on an SGLT2 inhibitor [1]. Both arms were taking something, so the study establishes a difference between two treated rates and nothing about where either one sits against no treatment. Meanwhile the thing that produces gout moved the other way: across 2,539 adults in SURMOUNT-1, serum uric acid fell up to 0.95 mg/dL on tirzepatide against 0.18 on placebo [2].
A study comparing two drugs produces one number and two possible readings, and only one of them is supported. This is a clean example, and it is worth working through because the misreading circulates. A reader who has heard that these drugs raise gout risk has probably met this study wearing the wrong label, the same way a composite gets over-heard in what a composite endpoint hides.
What was compared
Researchers identified 295,907 adults with type 2 diabetes newly prescribed either an SGLT2 inhibitor or a GLP-1 receptor agonist. Gout occurred at 4.9 events per 1,000 person-years in the SGLT2 group and 7.8 in the GLP-1 group. That is a hazard ratio of 0.64 (95% CI 0.57 to 0.72) and a rate difference of 2.9 events per 1,000 person-years (95% CI 2.1 to 3.6).
The authors’ conclusion is that SGLT2 inhibitors may reduce the risk of gout, and they say further studies are needed to confirm it. SGLT2 inhibitors have a known mechanism for this: they increase urinary excretion of uric acid, which is what causes gout.
What happened to uric acid itself
The mechanism runs through urate, and that has been measured inside a randomized trial. A post hoc analysis of SURMOUNT-1 tracked serum uric acid in 2,539 adults with obesity or overweight across 72 weeks [2]. It fell by 0.69, 0.92 and 0.95 mg/dL on the 5, 10 and 15 mg tirzepatide doses, against 0.18 mg/dL on placebo, all at P < .001.
A mediation analysis attributed 72.7% of that reduction to weight loss, leaving roughly a quarter attributed to something else about the drug. The fall did not differ by baseline uric acid quartile (P = .610) or by baseline BMI (P = .362), which is unusual in this literature. The full reading, including why a mediation split is a model output rather than a measurement, is in the SURMOUNT-1 urate analysis.
The same design problem, in a second condition
Kidney stones were compared the same way and produced the same trap. A new-user study matched 358,203 pairs starting an SGLT2 inhibitor or a GLP-1 receptor agonist and followed them a median of 192 days [3]. Nephrolithiasis occurred at 14.9 events per 1,000 person-years in the SGLT2 arm against 21.3 in the GLP-1 arm, a hazard ratio of 0.69 (95% CI 0.67 to 0.72).
That is again a study about SGLT2 inhibitors, with the GLP-1 arm as the thing they were measured against. It cannot say whether a GLP-1 raises the stone rate above no treatment at all, because no untreated group was followed. The same reading is set out in the kidney stone comparison.
Why the misreading is easy
Because an active-comparator design is a good way to answer a clinical question — which of two drugs should this patient take — and a bad way to answer a safety question about either one alone. The output is a ratio, ratios have a numerator and a denominator, and a reader who cares about the denominator will read the ratio as being about it.
The same structure appears throughout this literature. A result reported against another drug, or against people who were prescribed nothing for reasons nobody recorded, is answering a narrower question than its sentence suggests. That thread runs through what the semaglutide trials are worth and the suicidal ideation cohorts.
What this does and does not change
If you have gout and type 2 diabetes and a choice between these two classes, the first study is directly relevant and points one way. If you have gout and are considering a GLP-1 drug for weight, it tells you almost nothing, because the population is different and the comparison you care about is absent. Your urate is likely to fall, which is the direction gout treatment aims for, and the trial that would count flares rather than measure blood has not been done.
That is worth knowing before a seller’s page cites a study at you. Most storefronts tracked here publish no evidence summary at all, and the ones that do rarely name the comparator. It is a gap in the same family as the ones counted in what sellers will not tell you and what happens outside a trial.