Editorial 

April 10, 2025

Electronic Stewardship Prompts—Exploring the Why Behind the What

Shinya Hasegawa, Daniel J. Livorsi

JAMA Netw Open. 2025;8(4):e2512634. doi:10.1001/jamanetworkopen.2025.12634

Increasing antibiotic resistance is an urgent public health threat, and antibiotic stewardship is an important strategy for mitigating this threat.¹ Antibiotic stewardship programs strive to improve patient safety by optimizing the use of antibiotics that are indicated and reducing unnecessary antibiotic use. One of the key targets for antibiotic stewardship is extended-spectrum antibiotics; these agents exert more selective pressure than narrow-spectrum antibiotics and, in turn, can contribute to the spread of antibiotic resistance.²

The Joint Commission and the Centers for Medicare and Medicaid Services require hospitals in the US to have an antibiotic stewardship program.^3,4 However, stewardship programs generally have limited funding and are unlikely to see an increase in financial support any time soon. Therefore, a major challenge for stewardship programs is prioritizing their activities given these limited resources and the seemingly never-ending opportunities to improve antibiotic use.

To improve antibiotic use, stewardship programs frequently use a strategy called prospective audit and feedback. This involves the stewardship team routinely reviewing antibiotics being prescribed to hospitalized patients and providing real-time feedback to inpatient clinicians with the goal of optimizing current antibiotic therapy.⁵ Although effective, this strategy requires high levels of personnel effort for manual medical record reviews and communication.^6-8 Furthermore, it is challenging to influence decision-making at the moment antibiotics are first ordered and difficult to reach clinicians who initiate antibiotics during off-hours.

In this context, computerized provider order entry (CPOE) prompts can be a powerful tool that more efficiently provides real-time feedback to clinicians ordering antibiotics. Gohil and colleagues^9,10 previously reported on the effectiveness of a CPOE bundle in reducing empirical extended-spectrum antibiotic use for pneumonia and urinary tract infections in the INSPIRE trials. These were both cluster-randomized clinical trials across 59 community hospitals.^9,10

In the current issues of JAMA Internal Medicine and JAMA Surgery, Gohil and colleagues^11,12 report results from 2 additional clustered-randomized clinical trials investigating the use of CPOE prompts for skin and soft tissue infections (SSTIs) or intra-abdominal infections (IAIs). These 2 new INSPIRE trials, conducted across 92 community hospitals, compared routine stewardship with a CPOE bundle that incorporated clinician education, clinician progress reports, and electronic prompts recommending standard-spectrum antibiotics for patients with a low absolute risk (<10%) of multidrug-resistant organism (MDRO) infection. The risk of MDRO infection was estimated by a predictive model that was derived from a retrospective dataset of over 190 000 patients in 151 hospitals.

The CPOE algorithm, which was built within the MEDITECH electronic health record (EHR), activated within 72 hours of admission when extended-spectrum antibiotics were ordered for SSTI or IAI in nonintensive care unit (ICU) inpatients and emergency department patients. Eligible patients were identified when the clinician chose SSTI or IAI as the indication for their antibiotics; the indication was correctly chosen in roughly three-quarters of all SSTI cases and half of IAI cases. The prompt encouraged clinicians to consider replacing orders for extended-spectrum antibiotics with guideline-recommended standard-spectrum antibiotics for patients with a low risk of MDRO infection. In both trials, over 95% of patients were deemed at low risk of having a MDRO by the predictive model.

The INSPIRE SSTI trial included 118 562 patients (57 837 during the baseline period and 60 725 during the intervention period). Hospitals assigned to the CPOE bundle intervention reported a 27.5% reduction in empirical extended-spectrum days of therapy (DOT) (rate ratio, 0.72; 95% CI, 0.67-0.79; P < .001) compared with the routine stewardship group. Notably, 657 of 6886 clinicians (9.5%) changed their orders from extended- to standard-spectrum antibiotic therapy after receiving the real-time prompt during antibiotic ordering. Similarly, the INSPIRE IAI trial included 198 480 patients (93 476 during the baseline period and 105 004 during the intervention period). Hospitals using the CPOE bundle intervention saw a 35% reduction in empirical extended-spectrum DOT (rate ratio, 0.65; 95% CI, 0.60-0.71; P < .001) compared with the routine stewardship group. Among 10 256 clinicians, 1255 (12.2%) changed their orders in response to the CPOE prompt. Safety outcomes, including ICU transfer rates, days to antibiotic escalation, and hospital length of stay, were unchanged during the intervention period. Sensitivity analyses, considering ICU transfers on the first admission day, competing risks of death, and alternative definitions of antibiotic exposure (ie, doses-per-patient instead of DOT), showed findings consistent with the primary analysis, reinforcing the intervention’s safety and efficacy.

Gohil and colleagues^11,12 should be applauded for conducting these 2 well-designed studies within a large multicenter health care system across the US. Their findings provide valuable insights into the effectiveness of the CPOE bundle, which included electronic prompts informed by individualized patient-level risk assessments. In both trials, the effect of the CPOE bundle was fairly immediate. That is, a new baseline of improved antibiotic prescribing was achieved within only a few months at the intervention sites.

So how can antibiotic stewardship programs make use of these findings? A key question that readers might ask is how these CPOE prompts can be deployed in their own health care systems. Depending on local resources, building and sustaining the MDRO prediction models and electronic prompts with a local EHR system could be challenging. Given these potential difficulties, we would argue that it is important to understand how, or why, the bundle worked. In other words, which elements of the bundle—clinician education, clinician progress reports, or electronic prompts—were the primary drivers of its success?

While the electronic prompts were a key feature of the CPOE bundle, the percentage of prescribers who changed from an extended- to standard-spectrum antibiotic in response to the real-time prompt was fairly small (ie, 9.5% in the SSTI trial and 12.2% in the IAI trial). This suggests that only a small proportion of the observed reductions in extended-spectrum antibiotic use can be directly attributed to the alerts. We suspect that other aspects of the bundle, or the process of bundle implementation itself, may have also contributed to changes in antibiotic prescribing behavior.

At the clinician level, several factors may have influenced changes in prescribing behavior. First, it is possible that the very act of explaining to clinicians how MDRO risk is estimated and the rationale for the alerts addressed knowledge deficits, thereby promoting behavior change. Second, clinicians may have learned to adjust their prescribing practices after receiving the electronic alert only once or twice. It would be interesting to plot the frequency at which the electronic prompt was triggered over time. If clinicians quickly learned how to change their antibiotic prescribing, the frequency of alerts would have been initially high but then drastically decreased as the new behavior was adopted. Finally, feedback from the local stewardship team may have contributed to behavior change. The intervention hospitals received clinician progress reports to give individualized feedback on extended-spectrum antibiotic prescribing, but the frequency at which these reports were used is never described.

Although not directly measured, we suspect that several factors at the hospital level also contributed to behavior change. First, hospitals that volunteered to participate may have had an organizational culture that was more receptive to changing antibiotic-prescribing behavior. A total of 41 of 137 hospitals (approximately 30%) declined participation in either trial, and it would be interesting to know how these hospitals differed from sites that were involved. Second, the very decision to deploy the CPOE bundle was a demonstration of how antibiotic stewardship was a priority to hospital leadership. Leadership support is often critical to achieving buy-in among clinicians and facilitating effective communication. Third, as antibiotic-prescribing habits began to change in response to the CPOE bundle, the shared expectations around how to treat SSTIs or IAIs likely also began to change. Shifting norms around antibiotic prescribing may have helped clinicians feel more comfortable with avoiding extended-spectrum therapy. Finally, it appears that the same hospitals implemented the CPOE bundle for both SSTIs and IAIs during the same time period. Implementing alerts for both infections simultaneously may have collectively had a larger impact on changing practice, especially if certain elements of the intervention benefited from focusing on both infections concomitantly.

Qualitative research methods could help explore which of the previously metnioned factors were influential.^13,14 If feasible, the study team may consider using semistructured interviews, focus groups and electronic surveys of clinicians, antibiotic stewards, and other key partners to explore these mechanisms of change while also exploring other important implementation questions. Specifically, it would be useful to ask clinicians about their experience receiving these alerts and to inquire about why they chose to change their behavior. Additional key questions for the stewardship team could include how the antibiotic stewardship teams leveraged the educational tools and the clinician progress reports to motivate behavior change, and how much time and effort was required by local information technology and stewardship personnel to build and sustain the CPOE bundle.

The INSPIRE trials have demonstrated that a CPOE bundle can safely decrease extended-spectrum antibiotic use for the 4 most common infections in noncritically ill hospitalized patients: pneumonia, urinary tract infections, SSTIs, and IAIs. These trials represent a major step forward for the field of antibiotic stewardship. By next exploring the why behind the bundle’s success, strategies to replicate INSPIRE can be developed and tailored to a variety of hospital settings.