Infectious Disease · Medical Literature
Eliminating Inappropriate Postoperative Antibiotics: A Stepped-Wedge Cluster Randomized Trial.
Abstract
Unnecessary postoperative antibiotic use is common. This study evaluated whether modifying electronic health record order sets plus participation in a facilitation workshop by the antimicrobial stewardship team reduced inappropriate postoperative antibiotic duration more than modifying order sets alone. A stepped-wedge cluster randomized trial compared order set modification (active control) vs order set modification plus facilitation (experimental) from November 1, 2020, to October 31, 2023, at 9 US children's hospitals. The primary outcome was the rate of inappropriate postoperative antibiotic duration of greater than 24 hours after incision closure. Secondary outcomes included inappropriate postoperative antibiotics, defined as any antibiotic use and the rate of surgical site infections (SSI) or Clostridioides difficile infections (CDI). Generalized estimating equations (GEEs) accounting for clustering at the step and hospital level were utilized to compare 2 groups. A total of 18 267 patients (5910 control, 12 357 experimental) were included, with a median age of 9.2 years (IQR 2-14), 50.6% of the female gender, and 58% with public insurance/self-pay. The primary outcome was lower in the experimental group (10.05%, (95% CI: 9.53-10.59) than the control (11.02%, 95% CI: 10.23-11.84, P = .05), but this difference was not apparent in the GEE analysis (P = .74). The secondary outcomes demonstrated no difference between control and experimental groups for any postoperative antibiotic use (45.1% vs 44.7%, P = .98), SSI (1.96% vs 2.03%, P = .70), and CDI (0.15% vs 0.06%, P = .85). Facilitation training did not reduce inappropriate postoperative antibiotic duration compared with order set modification; better understanding of how hospital context informs change strategies is needed.
Clinical relevance
Does adding an antimicrobial stewardship team facilitation workshop to EHR order set modification reduce the rate of inappropriate postoperative antibiotic duration (>24 hours after incision closure) more than order set modification alone in pediatric surgical patients at US children's hospitals? __SNAPSHOT__ {"studyDesign":"Stepped-wedge cluster randomized trial (OPerAtiC Trial; ClinicalTrials.gov NCT04366440)","setting":"9 US children's hospitals participating in the SHARPS Collaborative and ACS NSQIP-Pediatric; November 1, 2020 – October 31, 2023","sampleSize":"n = 18,267 (control [order set modification only]: n = 5,910; experimental [order set modification + facilitation]: n = 12,357)","population":"Pediatric surgical patients (clean and clean-contaminated cases) at 9 US children's hospitals; median age 9.2 years (IQR 2–14), 50.6% female, 58% with public insurance or self-pay. Surgical specialties included orthopedics (~31%), general surgery (~21%), and neurosurgery (~17%) based on prior conference data.","intervention":"EHR order set modification targeting removal of unnecessary postoperative antibiotics PLUS a virtual antimicrobial stewardship team facilitation workshop (informed by the i-PARIHS framework; emphasizing context analysis, evidence application, and recipient engagement)","comparator":"EHR order set modification alone (active control; all hospitals began in this arm before sequential crossover to the experimental condition)","primaryOutcome":"Rate of inappropriate postoperative antibiotic duration defined as antibiotic use >24 hours after incision closure","followUp":"3 years total trial duration (November 2020 – October 2023); 30-day postoperative clinical outcome surveillance per NSQIP-P protocol","fundingSource":"Government (Agency for Healthcare Research and Quality [AHRQ], grant R01HS026742); principal investigator also holds industry funding from Pfizer and Moderna for separate trials (disclosed)"}
Key findings
Primary outcome: The crude rate of inappropriate postoperative antibiotic duration (>24 hours) was numerically lower in the experimental group (10.05%; 95% CI: 9.53–10.59%) than in the control group (11.02%; 95% CI: 10.23–11.84%; unadjusted P = .05; absolute risk difference 0.97 percentage points [calculated]). However, after accounting for clustering at the step and hospital level using generalized estimating equations (GEE) — the pre-specified primary analysis — this difference was not statistically significant (P = .74). The NNT cannot be meaningfully calculated given the non-significant cluster-adjusted result. Secondary outcomes: No statistically significant difference was observed between groups for any postoperative antibiotic use (control 45.1% vs. experimental 44.7%; P = .98), surgical site infection (SSI; 1.96% vs. 2.03%; P = .70), or Clostridioides difficile infection (CDI; 0.15% vs. 0.06%; P = .85). A prior conference abstract (OFID 2025) using a slightly smaller dataset (n = 17,202) reported that the GEE model showed a statistically significant reduction in inappropriate antibiotic use across steps when controlling for time and hospital (P < 0.001); this discrepancy with the final published analysis may reflect differences in analytic datasets, model specification, or follow-up time, and the published Pediatrics abstract result should be considered definitive. No safety signals (increased SSI or CDI) were associated with either strategy.
Practice implications
What the study established: EHR order set modification alone is associated with a low overall crude rate of inappropriate postoperative antibiotic duration (~11%) in clean and clean-contaminated pediatric surgical cases, consistent with prior stewardship progress. Adding a facilitation workshop did not produce an additional statistically significant reduction in this rate in the cluster-adjusted primary analysis. What the study did not establish: The trial did not demonstrate superiority of facilitation over order set modification alone using the pre-specified GEE analysis. The study cannot confirm whether either active intervention is superior to no intervention (no true no-treatment control arm was included). The trial was conducted exclusively at children's hospitals in the SHARPS Collaborative and NSQIP-P networks, which represent relatively resource-rich, academically affiliated pediatric centers; findings may not generalize to community hospitals, adult surgical populations, or resource-limited settings. Critically, the study population had only a ~10–11% baseline rate of inappropriate postoperative antibiotics, substantially lower than the 33% rate assumed in the original power calculation, raising the possibility of a floor effect limiting the detectable benefit of facilitation. The pre-specified primary endpoint was not met. Practice implications: EHR order set modification alone remains a sound and evidenced stewardship tool for reducing inappropriate postoperative antibiotics in pediatric surgery. Routine addition of a formal facilitation workshop does not appear to confer incremental measurable benefit on this endpoint in hospitals that have already engaged in order set reform. Subgroup patterns (e.g., high-burden spine cases) and contextual factors that may identify facilities most likely to benefit from facilitation require further investigation. NOTE: This summary is based solely on the published abstract; full-text data including subgroup analyses, order set penetration metrics, and implementation outcomes are not yet available for independent evaluation. __LIMITATIONS__ Only 9 hospitals included; small cluster number substantially limits statistical power and generalizability, despite large patient-level n,Stepped-wedge design introduces risk of time-effect confounding (secular trends in stewardship practice during the 3-year COVID-era study period may have narrowed the evidence-practice gap independently of the intervention),The observed inappropriate antibiotic rate (~10–11%) was far below the 33% assumed in the original power calculation, suggesting the trial was effectively underpowered for the observed effect size and representing a likely floor effect,Active control design (both arms received order set modification): the incremental benefit attributable to facilitation alone is assessed, not the benefit of either intervention over no intervention; absolute stewardship impact of either arm is unquantifiable from this design alone,Imbalance in group sizes (5,910 control vs. 12,357 experimental) is inherent to the stepped-wedge design and may affect GEE model stability with few clusters,Data collected via NSQIP-P (30-day case sampling, ~35 cases per 8-day cycle): sampling methodology may not capture all cases and introduces potential selection bias within each hospital,CDI rates were extremely low (0.06–0.15%), leaving the trial profoundly underpowered to detect differences in this clinically important safety outcome,Discordance between conference abstract (significant GEE result, P 24 hours) beyond order set modification alone in the cluster-adjusted primary analysis, and secondary outcomes including SSI and CDI rates were also unchanged. The low observed baseline rate of inappropriate antibiotic use (~10–11%) suggests meaningful prior progress from order set reform alone, possibly creating a floor effect that limited detectable benefit from facilitation. Clinicians and stewardship programs should continue to prioritize EHR order set optimization for postoperative antibiotic de-implementation; facilitation may add value in contexts with high baseline inappropriate use or low order set penetration, but this has not been established in the current trial.