Increasing Cefazolin Use for Perioperative Antibiotic Prophylaxis in Penicillin-Allergic Children

This article conforms to the SQUIRE 2.0 guidelines.¹⁹  The project setting was the operating rooms (ORs) of a large quaternary care pediatric institution, including the main university hospital and 4 stand-alone ambulatory surgery centers. The project involved all children over a 9-month period with a nonsevere penicillin-class (Table 1) allergy recorded in the EHR who had a procedure with general anesthesia requiring perioperative antibiotics for SSI prophylaxis.

Patients were excluded if they had severe allergy to a penicillin-class antibiotic reported in the EHR, predefined as anaphylaxis, edema, respiratory distress, chest tightness, cardiac arrest, skin breakdown, Stevens-Johnson syndrome, and erythema multiforme. All other documented reactions were included. Despite the limited cross-reactivity between penicillin and cefazolin, we excluded patients with potentially severe reactions to ensure safety and improve buy-in for this quality improvement (QI) work. Patients excluded from the study also included those with an allergy listed to a cephalosporin or a carbapenem, those undergoing a common procedure in which a perioperative antibiotic other than cefazolin was indicated (ie, laparoscopic appendectomy), and those taking antibiotics preoperatively. Our goal in these exclusions was to focus on the group of patients with a reported penicillin allergy who most likely should receive cefazolin for perioperative prophylaxis. Finally, we excluded cases in which a pediatric cardiac anesthesiologist was providing care because this is a separate group of providers, not targeted by the planned interventions.

The project was initiated in December 2019 with our first interdisciplinary team meeting and extended through September 2020. Data collection is ongoing with data analysis completed through March 2021. The QI team consisted of 3 anesthesiologists, an infectious disease physician, two allergy and immunology physicians, two members of the Antimicrobial Stewardship Program (ASP), a pharmacist, and a surgeon. The team was supported by a QI project advisor and a data analyst. The team planned to use iterative plan-do-study-act (PDSA) cycles²⁰  to achieve their specific, measurable, achievable, relevant, and time-bound²¹  aim of increasing the percentage of patients with nonsevere penicillin-class allergies who receive cefazolin as perioperative antibiotic prophylaxis to 85%. The QI team developed a key driver diagram to identify possible interventions and then created an impact-effort matrix to determine the most efficient approach. Process mapping revealed that antibiotic selection most often occurs on the day of the procedure. Any necessary clarification of allergies and reactions happens between the anesthesiologist and the family as part of the preoperative discussion. The antibiotic is then chosen on the basis of an evidence-based reference guide for antibiotic prophylaxis by a surgical procedure created by our institution’s ASP.

The first intervention was to provide formal education to the general anesthesia division, including attending physicians, pediatric anesthesiology fellows, and certified registered nurse anesthetists. This was because, in our practice, anesthesiology teams are responsible for the administration of the antibiotics. The education intervention comprised of an anesthesiology Grand Rounds lecture delivered by two allergy physicians, after an informal survey of the anesthesia providers on antibiotic selection and reasoning. They presented data on the over labeling of penicillin allergies and the associated implications as well as on the low rates of cross-reactivity between penicillins and cefazolin. Furthermore, they specified what was considered a severe penicillin allergy for the purposes of this project and specifically clarified that a history of urticaria alone was not because this was the most commonly held misconception within the anesthesia group.

The formal education intervention was followed by an e-mail to the surgical division leaders informing them of the QI project. We elicited their feedback and asked that they share this information with their respective departments.

Finally, the allergy physicians provided similar educational content formally to the OR nursing staff. OR nursing staff are often the individuals who physically acquire the prediluted antibiotics from the Pyxis machine (for clindamycin) or antibiotic refrigerator (for cefazolin) and deliver them to the anesthesiology team. Broadly, we wanted them to know that a penicillin allergy on the patient’s allergy list did not by default require an alternative to cefazolin.

An evidence-based reference guide for antibiotic prophylaxis by surgical procedure, created by our institution’s ASP, is linked within the EHR interface to assist in antibiotic selection at the point of care. During our first PDSA cycle, we discovered that the wording of this guide for antibiotic selections suggested alternative to cefazolin for all cephalosporin and penicillin allergies but did not differentiate between severe and nonsevere penicillin allergies. To the treating anesthesiologist, the oversimplification of this language could mislead to an automatic decision to choose an alternative antibiotic incorrectly. We modified this document to reflect that alternative antibiotics be used only if the reaction to penicillin was severe.

We set up an automated e-mail to be sent to the anesthesia team (including the attending and pediatric anesthesiology fellow or certified registered nurse anesthetist) and the surgery attending physician whenever clindamycin was given to a patient in our cohort. Automated e-mail delivery was triggered if the EHR listed a nonsevere reaction to a penicillin class antibiotic and clindamycin was administered because clindamycin was by far the most common alternative to cefazolin in our cohort and is infrequently recommended for nonallergy indications. E-mail notifications were sent on the Monday after the procedure. The intent of the e-mail notifications was to raise awareness and provide an educational opportunity as well as inform the improvement process through discussion.

Of note, beginning on March 24, 2020, all elective cases were postponed because of the coronavirus disease 2019 (COVID-19) pandemic. Because of this decreased OR volume and because teammates were redeployed to work on pandemic efforts, we paused any interventions until May 4, 2020, when elective cases resumed.

The primary outcome measure was defined as the percentage of surgical encounters among patients with nonsevere penicillin-class allergies undergoing a procedure requiring perioperative antibiotic prophylaxis who received cefazolin. Our secondary outcome measure was the incidence of SSIs within this cohort, as measured by using an electronic clinical surveillance system, TheraDoc (Premier Inc, Salt Lake City, UT).

As a process measure, we also followed the percentage of surgical encounters in the same patient population who received clindamycin as their perioperative antibiotic prophylaxis because clindamycin was the most commonly given alternative to cefazolin in this patient population and infrequently recommended for nonallergy indications. Measuring clindamycin use would therefore eliminate some of the nuances in antibiotic selection that we were unable to anticipate in the primary outcome measure.

Our balancing measure was the incidence of severe allergic reactions within the perioperative period and incidence of a new allergy to cefazolin added to the patient’s EHR within 48 hours of the surgical encounter.

Patient surgical encounters with reported nonsevere penicillin-class allergies were identified by using an enterprise data warehouse containing data from the EHR (Epic Systems, Verona, WI). A clinical application containing patient, surgical, and medication variables was developed and displayed on a data visualization platform, RStudio Connect (RStudio, Boston, MA). For the primary outcome and process measures, data were aggregated on a monthly basis. Statistical process control (SPC) charts were used to measure the impact of the interventions and differentiate special cause variation from common cause variation. Special cause variation was defined when there were 8 consecutive points above or below the mean.²²  A new baseline was established when the data met criteria for a shift. Balancing measures were extracted from the anesthesia department’s QI database and EHR and evaluated quarterly.

This project was undertaken as a QI initiative and did not constitute human subjects research by our institution. Safety considerations were addressed by implementing policies consistent with published evidence and guidelines^5,12,23–25  and through the balancing measures.

Key interventions were conducted over 3 PDSA cycles. Approximately 400 patients were included. The SPC chart revealed the percentage of patients in our cohort who received cefazolin for perioperative antibiotic prophylaxis (Fig 1A). Baseline data are depicted over a 3-year period, starting in January 2017. Within this baseline period, there was a shift in the center line from 41% of eligible patients receiving cefazolin to 60%. With the start of this QI project, there was another shift in the center line to 81%, which was maintained throughout the rest of the project. In September 2020, 90% of eligible patients received cefazolin, surpassing our project goal of 85%. This was the first time this was achieved outside of the time period when elective surgery was paused secondary to COVID-19. The data points after the project’s end reveal sustained improvement. The rate of SSIs in our cohort did not change from the baseline of one per year (0.2%).

The process measure reflecting the percentage of patients receiving clindamycin decreased throughout the initiative, corresponding to the increase in cefazolin (Fig 1B).

There were no changes in the balancing metrics: severe allergic reactions in the operative room remained at zero for this population and the incidence of cases with a new cefazolin allergy documented within 48 hours of the surgical encounter did not change, with two before project initiation, during the baseline phase, and two during the project duration.

We successfully increased the percentage of patient surgical encounters with reported nonsevere penicillin-class allergies who received cefazolin for perioperative antibiotic prophylaxis using 3 strategic improvement interventions. In September 2020, ∼6 months from our first intervention, we achieved our goal of having at least 85% of eligible patients receiving cefazolin.

Our results suggest that, despite studies to the contrary,^26,27  education, with multidisciplinary collaboration, was critical for improvement. We considered an education-based intervention a high-value opportunity to provide maximum impact with relatively low cost and risk. In our preparations for this QI project, through an informal survey, we found a knowledge deficiency among those responsible for antibiotic administration. Interdisciplinary education from experts in allergies helped to close this knowledge gap and resulted in measurable improvement. The automated e-mails provided reminders to specific practitioners who may have forgotten or missed this initial education, directing refreshers to where they were needed.

Although formal education initiatives were the basis of our interventions, it is interesting to note that our baseline data revealed special cause variation, with a shift in the center line that occurred in the year before our project start. This in part might be explained by the beginning of interdisciplinary discussions and informal education that also began in the year before the official start.

Furthermore, although the primary measure for this project was a process measure, the potential implications for improving clinical outcomes are far broader. For example, Blumenthal et al¹⁴  reported a 51% increased risk of SSI in patients with a reported penicillin allergy who received an alternative to cefazolin for preoperative antibiotic prophylaxis. Appropriate selection of SSI prophylaxis in patients labeled as penicillin allergic can potentially extend to a reduction in SSI. In reflection of our own practice, our baseline rates of SSI were relatively low and, therefore, would make it challenging to demonstrate a change because of our QI interventions. In addition, safe administration of cefazolin in the OR may lead other clinicians to prescribe cephalosporins for nonsurgical indications in which these agents are first-line, thus having potential positive down-stream impacts for patients.

A review of the cases with new documentation of cefazolin allergies revealed that in one case, the documented reaction of itching alone was unlikely a hypersensitivity reaction. This patient subsequently received vancomycin in a later surgical procedure at another institution. The allergy service was not consulted on any of the patients who received a new allergy label perioperatively.

These newly documented cefazolin allergies emphasized the need for allergy consultation and referral. Allergists can help to identify key details during a presumed drug reaction and better guide clinical management in the short- and long-term. Challenges to perioperative allergy consultation include time limitations and interruptions to existing workflow patterns; therefore, establishing criteria to identify patients who may require further outpatient allergy investigation is crucial for comprehensive patient care.²⁸ 

Our initial project time frame was extended because of the COVID-19 pandemic. Between the months of March 2020 and May 2020, our health system canceled all elective surgical procedures and shifted priorities to responding to the pandemic, as many other health systems did across the United States. Although we continued tracking our antibiotic administration practices during this time, we do not consider April to have met special cause variation due a >80% reduction in patient volume during this month, which resulted in a larger confidence interval. This highlights the importance of interpreting results and QI analysis in the larger context of practice and real-world results.

Understanding the unique practice and processes of each institution is also critical toward being successful in QI initiatives. Yet, the uniqueness also underscores the limitations of applying our specific interventions to a different practice and organization. For example, in our institution, the anesthesiologists select and administer the antibiotic prophylaxis after discussions with the surgeon and patient and/or family on the day of procedure. This is different from other pediatric practices in which antibiotics are preselected and ordered by the surgical team in advance of the day of surgery. Our specific interventions are unlikely to be as successful in this type of practice.

Our education methods were limited in that they did not directly engage anesthesia residents who rotate from outside institutions for their pediatric experience. Anesthesia residents are front-line providers who administer perioperative antibiotic prophylaxis, and focused education could improve the sustainability of this project.

We will continue to track our data to assure we can sustain our results and continue to improve perioperative antibiotic selection for patients, while monitoring for any severe allergic reactions. We plan to bring our experience and success to the cardiac anesthesia group to consider spread to their patients. We are exploring an EHR-based decision-support tool to help guide the health care provider ordering and administering perioperative antibiotics.²⁹ 

Beyond our project, a broader opportunity for improvement includes an organized effort to delabel patients who are not truly allergic to penicillin. This will require health care provider education and encouragement to remove false labels in conjunction with preoperative allergy consultation and penicillin allergy testing. The foundation for this effort at our institution was laid by two of our authors (JL and SS) in the article “Clinical Pathway for the Assessment of Children with a Penicillin Drug Allergy.”³⁰  However, patients and families are not likely to accept delabeling in the perioperative period on the basis of history³¹  and are high-risk for relabeling even after allergy evaluation.³²  Penicillin allergy delabeling is truly a multidisciplinary task,^33,34  and our team has laid the foundation for ongoing collaboration.

We safely increased the use of cefazolin for perioperative antibiotic prophylaxis in patients with a reported penicillin allergy. We found a knowledge gap and showed that, in this case, iterative interdisciplinary education was necessary for this improvement.

ASP

Antimicrobial Stewardship Program

COVID-19

coronavirus disease 2019

EHR

electronic health record

OR

operating room

PDSA

plan-do-study-act

QI

quality improvement

SPC

statistical process control

SSI

surgical site infection