A stepped-wedge cluster randomised trial across Lao hospitals compared antimicrobial prescribing guidelines delivered through a smartphone application, with stewardship training, against the same guidelines on paper. Adherence meant the correct antimicrobial at the correct dose, assessed by repeated point-prevalence surveys across 3,561 inpatient prescriptions.
Raw adherence was 17.0% with paper and 25.6% with the app. After adjusting for time, clustering, and confounders, the difference vanished: an adjusted odds ratio of 1.26 with an interval from 0.8 to 1.9. Outpatient prescribing showed the same, at 0.91.
The adjustment is the finding. A stepped-wedge design rolls the intervention out in sequence, so anything improving over time — familiarity, the Hawthorne effect, the training itself — is confounded with the intervention unless time is modelled. Here the raw gap looked like a nine-point improvement and was mostly time. Note also what neither arm achieved: three-quarters of prescriptions did not follow the guideline in either group.
- The unadjusted 17% versus 25.6% gap is what a before-and-after study would have reported
- Stepped-wedge designs require time-adjustment; without it the result reverses
- Both arms left about three-quarters of prescriptions off-guideline
- Prescriptions could not be linked to individual prescribers, so behaviour change is unmeasured
- The medium was not the barrier — worth asking what is, before buying an app
The statistics, in plain English
This is a clean demonstration of why adjustment matters. The crude comparison — 25.6% against 17.0% — looks convincing. The adjusted odds ratio of 1.26 with an interval from 0.8 to 1.9 includes 1.0, meaning that once you account for adherence improving over the study period regardless of arm, the app explains none of it.
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 infectious diseases, written by the desk, every claim tied to its paper. Six minutes.
Get the app — free