- Design
- post-hoc analysis of individual participant data from two randomised, double-blind, placebo-controlled trials; Cox and Andersen-Gill models
- Population
- 8,614 participants with type 2 diabetes in canagliflozin trials; abnormal leukocyte esterase in 10.7%, nitrite in 3.5%
- Primary outcome
- time to first urinary tract infection; total infection events as secondary
- Effect
- canagliflozin HR 1.07 (95% CI 0.93-1.22) overall; 0.94 (0.73-1.21) with abnormal leukocyte esterase, interaction P = .10
Fear of urinary tract infection restrains prescribing of sodium-glucose cotransporter 2 inhibitors, particularly in patients whose urine already looks abnormal — and urologists are often the ones asked whether the drug should be stopped. This post-hoc analysis of individual participant data from two randomised placebo-controlled canagliflozin trials in type 2 diabetes, covering 8,614 participants, tested both halves of that worry.
The first half held: the dipstick predicts infection. Abnormal leukocyte esterase was present in 10.7% and nitrite in 3.5%, and the risk of a first urinary tract infection rose with the degree of leukocyte esterase positivity (hazard ratios 1.72, 1.92 and 2.81 for 1+, 2+ and 3+ against normal). Abnormal nitrite carried hazard ratios of 2.28 and 1.66.
The second half did not. Canagliflozin did not raise the risk of a first infection (HR 1.07, 95% CI 0.93-1.22) or of total events (1.01, 0.86-1.17), and — the question that matters — the dipstick findings did not modify that. In participants with abnormal leukocyte esterase the hazard ratio for canagliflozin was 0.94 (0.73-1.21) against 1.17 (1.00-1.38) in those with normal dipsticks, with no significant interaction.
So the two facts have to be held separately. An abnormal dipstick identifies a patient at higher baseline risk of urinary tract infection, which is worth knowing and worth acting on in its own right. It does not identify a patient in whom this drug adds to that risk, and withholding a medication with established cardiorenal benefit on that basis removes a real benefit to avoid a risk these data do not support. That is a clear recommendation to give a referring physician.
- Do not advise withholding or stopping an SGLT2 inhibitor because of a dipstick showing leukocyte esterase or nitrite
- Record that an abnormal dipstick marks higher baseline urinary tract infection risk, independent of the drug
- Give the referring physician a clear answer — this question usually arrives as a referral rather than a decision you make
- Continue to assess symptomatic infection on its merits and treat it as you would otherwise
- Note that these data concern canagliflozin in type 2 diabetes; the class effect is presumed rather than directly tested here
Why it matters
A drug with established cardiorenal benefit is being withheld from the patients most likely to need it, on a test result that does not carry that information.
Don't overread it
This is a post-hoc analysis of one drug in type 2 diabetes; the subgroup interaction tests have limited power and the class effect is inferred.
The statistics, in plain English
The overall hazard ratio of 1.07 with an interval from 0.93 to 1.22 crosses 1.0, so no increase in infection risk was shown; with 8,614 participants this is a reasonably well-powered null rather than a failure to look. The interaction P values of .10 and .45 test whether the drug behaves differently by dipstick result, and neither reaches significance — though an interaction test is always less powerful than the main comparison, so a modest difference between subgroups cannot be excluded. This is a post-hoc analysis, meaning the question was asked after the trials were designed for something else.
Read the rest in the app
You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.
Scan to keep reading on your phone. No account needed to start.
Tomorrow morning, before your first patient
One edition a day for urology, written by the desk, every claim tied to its paper. Six minutes.
Get the app — free