Two diabetes drug classes, two different stone risks. Korean national data put the gap at about half

Very few people choose a diabetes drug with their kidneys' stone-forming chemistry in mind, and no guideline asks them to. Korean national data now suggest the choice carries a urological consequence large enough to be worth naming out loud, particularly for the patient who has already passed one stone.

What was compared, and how#

A study in Diabetes Care used the Korea National Health Insurance Service database from 2010 to 2021 to emulate a trial that has not been run: patients with type 2 diabetes starting a sodium-glucose cotransporter 2 inhibitor, against patients starting a dipeptidyl peptidase 4 inhibitor. The two groups were matched one to one on propensity score, separately in people who had never formed a stone and in people who had.

• Stone never-formers: 105,378 matched pairs.
• Previous stone formers: 11,628 matched pairs.
• Total: 117,006 matched pairs, followed for a mean of 654 days.

The comparison is deliberately drug-against-drug rather than drug-against-nothing. Comparing a treated group with an untreated one in claims data mostly measures who gets treated. Comparing two drug classes prescribed to the same kind of patient for the same indication removes much of that, and the study added a check most do not: osteoarthritis encounters as a negative control outcome, an event neither drug should influence. Those associations were near-null, which is what a working design looks like.

The result#

Pooled across both strata the hazard ratio was 0.54, with a confidence interval from 0.50 to 0.57. The subgroup analyses by sex, age, thiazide co-use and baseline cardiovascular risk were consistent.

The two numbers that point in opposite directions#

This study is a clean example of why a relative risk on its own can mislead a clinical decision, and it is worth reading slowly.

The relative reduction is much larger in people who have never had a stone — a 57 percent lower hazard against 36 percent. The absolute reduction is seven times larger in people who have. Both statements are true of the same data, and only the second one tells a urologist which patient the drug choice actually matters for.

Where this sits in Korean stone risk#

Diabetes is not a bystander in Korean stone disease. The national cohort analysis that estimated Korea's lifetime prevalence of urolithiasis at 11.5 percent named diabetes among the contributing factors, alongside age over 60, income level, body mass index, hypertension and a cancer history.

An older Korean study makes the metabolic link directly. Following 40,687 people attending a health promotion centre between 1995 and 2009, with kidney ultrasonography as well as anthropometric and biochemical measurement, it found nephrolithiasis in 1.5 percent — 1.9 percent of men against 1.0 percent of women — and identified metabolic syndrome and each of its component traits as risk factors. In men, high body mass index, high blood pressure and abnormal glucose metabolism were each significant; in women, high body mass index and abnormal glucose metabolism.

What a claims study cannot deliver#

This is an emulation, not a randomised trial, and propensity-score matching on measured variables cannot balance what the database never recorded — fluid intake, occupational heat exposure, diet, or urine chemistry, all of which bear on stone formation and none of which appear in a claim. The negative control result argues that gross confounding by health-seeking behaviour is unlikely; it cannot exclude confounding specific to stone risk.

The follow-up is also short for a disease measured in decades. A mean of 654 days is under two years, and Korea's own national data put recurrence at 21.3 percent within five. Whether the separation between the two drug classes widens, holds or closes over a stone patient's actual horizon is not answered here.

What the study does establish is a difference large enough, in a cohort large enough, to belong in the conversation. For a Korean patient with type 2 diabetes who has already formed a urinary stone, which glucose-lowering class is started is not a urologically neutral decision.