DailyDoctor Archive Specialties Get app
Back to the 2 September 2026 edition

Pearl · 05 of 05

Ask who was monitored more closely before believing a complication rate

For any complication defined by a laboratory value rather than an event, ask whether both groups were tested equally often — unequal monitoring inflates the rate in the more closely watched arm.

The thrombectomy analysis contains its own correction, and it is the most instructive part of the paper.

Acute kidney injury appeared twice as common with intensive blood pressure management. But intensive management means more frequent monitoring, and acute kidney injury is defined by a change in creatinine — an outcome you can only detect if you measure it. When the authors ran a sensitivity analysis restricted to patients with equal creatinine ascertainment, the 48-hour association fell from significant to an odds ratio of 4.20 with an interval crossing 1.0.

This is ascertainment bias, and it is endemic to any outcome defined by a test result rather than by an event. A patient who dies is counted whether or not anyone was looking; a patient whose creatinine rises 30 micromoles is counted only if someone drew the blood.

The habit: for any complication defined by a laboratory value, ask whether the groups were tested equally often. If they were not, the difference in rates is partly a difference in looking — and the direction of that bias always favours finding more disease in the more closely watched arm.

  • Outcomes defined by test results require equal testing to compare fairly
  • More intensive treatment usually means more intensive monitoring
  • The bias always inflates the complication rate in the watched group
  • Look for a sensitivity analysis restricted to equal ascertainment
  • Hard outcomes — death, dialysis — are far less vulnerable to this

The statistics, in plain English

Ascertainment bias differs from confounding: it is not that the groups differ in risk, but that they differ in the chance of a real event being recorded. Adjustment cannot fix it, because the missing events are absent from the data entirely. Only restricting the analysis to equally monitored patients, as done here, addresses it — and it usually shrinks the effect.

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.

QR code to install Daily Doctor
Get Daily Doctor — free

Scan to keep reading on your phone. No account needed to start.

akielectrolytestonesckddialysisglomerular

Tomorrow morning, before your first patient

One edition a day for nephrology, written by the desk, every claim tied to its paper. Six minutes.

Get the app — free
Daily Doctor All 27 specialties, every morning. Free.
Get the app