Across a stepped-wedge cluster randomized trial in six Lao hospitals, a smartphone version of national antimicrobial prescribing guidelines plus one stewardship training session did not significantly improve adherence over paper versions after adjustment for time, clustering, and confounders. Inpatient adherence was 25.6% versus 17.0% before adjustment, but the adjusted odds ratio was 1.26 (95% CI 0.8–1.9; p=0.276); outpatient results were also null. The finding argues for sustained, multifaceted stewardship rather than guideline access alone.
Key findings
- For inpatients, adherence was 25.6% during the app phase versus 17.0% during the paper phase, but the advantage disappeared after adjustment for time, clustering, and confounders (aOR 1.26; 95% CI 0.8–1.9; p=0.276). The trial therefore did not establish superiority of the app.
- For outpatients, adherence was similarly low at 23.0% with the app and 21.7% with paper; the adjusted comparison was null (aOR 0.91; 95% CI 0.7–1.1; p=0.406). Differences by rollout sequence point to contextual and temporal influences.
- Prophylaxis was a major gap: full adherence was only 4.2% for surgical prophylaxis and 15.7% for medical prophylaxis. Wrong drug selection and excessive duration, especially around obstetric and gynecological procedures, were prominent problems.
- The app was accessed 6,239 times over 22 months among 251 prescribers who reported using it—about once per user per month. This supports limited engagement, although access logs cannot show whether an individual prescription was informed by the app.
Why this matters globally
Antimicrobial resistance is a global threat that weighs especially heavily on resource-constrained health systems. This pragmatic randomized evidence shows that smartphone delivery of prescribing guidance is feasible in a lower-income setting, yet access alone did not clearly change prescribing. That null result is valuable because it cautions against stand-alone app investment and redirects attention toward repeated education, audit and feedback, leadership support, microbiology capacity, and department-specific interventions. It does not show that apps are useless; rather, digital tools should support a sustained, multifaceted stewardship program.
Thai researcher contribution
Four Thailand-affiliated authors were linked to Mahidol University and MORU. Sue J. Lee contributed data curation and formal analysis; Mavuto Mukaka contributed formal analysis; Nicholas P. J. Day contributed conceptualization, funding acquisition, methodology, and supervision; and Paul N. Newton contributed conceptualization, methodology, and supervision. Their roles connected Thailand-based tropical-medicine and analytical expertise with the Lao network. However, Vilada Chansamouth was the principal investigator and Lao institutions and hospitals led field implementation, so this should be described as Thai-affiliated collaboration rather than a wholly Thai-led study.
Limitations to consider
Only six hospitals participated and cluster sizes varied, limiting power for small effects and condition-specific analyses. COVID-19 disrupted timing and hospital operations despite statistical adjustment for time. The intervention bundled the app with one training session and was unblinded, so the effect of each component cannot be isolated. Point-prevalence surveys provide snapshots rather than continuous surveillance, and prescriptions could not be linked reliably to individual prescribers. The study did not directly measure patient outcomes or antimicrobial resistance, and findings from Laos should not be transferred to Thailand without examining health-system context.