Comparison of Persuasive and Restrictive Strategies for Antibiotic Stewardship for Hospital Inpatients: Meta-Analysis of 51 Interrupted Time Series Studies

Background:

Persuasive strategies for changing practice include dissemination of educational materials, audit and feedback or educational outreach. In contrast restrictive strategies impose constraints on prescribers with strategies such as compulsory authorisation of prescriptions or automatic stop orders. Conventional statistical analysis of interrupted time series (ITS) quantifies immediate and sustained effects of the intervention as change in level and slope of the regression line respectively. However, change in slope has little clinical meaning.

Objective:

To develop a method for meta-analysis that compares the impact of persuasive and restrictive interventions as % change in outcome at specified time intervals.

Methods:

Systematic review of interventions published from January 1980 to December 2007 using the Cochrane Effective Practice and Organisation of Care methods. We identified ITS studies with data that could be used to estimate effect size as % change in outcome at 1, 6, 12 or 24 months after the intervention. We extrapolated the pre-intervention regression line and subtracted the post-intervention value for each time point. We only used studies with actual data at each interval. All differences were expressed as: (persuasive – restrictive).

Results:

55 studies were identified, of which 51 included data for the meta-analysis. The other four studies used appropriate statistical models but did not include estimates of variance for the effect size at any of our time points and these could not be recalculated from the raw data in the papers. The outcomes for the remaining 51 studies were drug (n=37), microbial (n=14) and cost (n=4); 4 studies had more than one outcome. For drug outcomes the restrictive interventions had greater effect at 1 month (+32.0%, 27 studies, 95% CI +2.5% to +61.4%) but at 6 months the difference had diminished to +10.1% (15 studies, 95% CI -47.5% to +66.0%) and at 12 or 24 months the persuasive interventions had greater effect. Difference at 12 months was -24.6% (18 studies, 95%CI -71.9% to +22.6%) and difference at 24 months was -12.3% (11 studies, 95%CI -60.2% to +35.5%). For microbial outcomes the restrictive interventions had greater effect at 6 months but not at 12 or 24 months. For cost outcomes there were too few studies to compare effects.

Conclusions:

The results suggest that, in comparison with restrictive interventions persuasive interventions have lesser short term impact on prescribing but may have greater long term impact. Expression of outcomes as % change at different time points facilitates comparison of the effectiveness of interventions. Future interrupted time series studies of antibiotic stewardship should include estimates of effect size with variance at increasing intervals after the intervention.