Improving the Documentation of Penicillin Allergy Labels Among Pediatric Inpatients Free

The project was conducted January 2020 through June 2022 at a large quaternary-care academic medical center serving adult and pediatric patients across the Southeastern United States. The medical center includes a 190-bed pediatric hospital with 6500 annual discharges. One hundred inpatient pediatric beds are dedicated to PHM, surgical, and medical subspecialty admissions. The medical center uses the electronic health record (EHR), EPIC, (Epic Systems Corporation, Verona, WI).

Starting in January 2021, patients admitted to PHM services were the initial target population as the PHM team was the primary project stakeholder. The 3 PHM services include 2 general teams and 1 complex care team. During the weekdays, the 2 general PHM rounding teams are composed of an attending hospitalist, a senior resident, 2 intern physicians, 2 medical students, a clinical pharmacist, a pharmacy student, a case manager, and a dietician. The complex care PHM rounding team is composed of an attending hospitalist, a resident physician, a clinical pharmacist, a dietician, and a case manager. During weekends and holidays, the 3 PHM rounding teams are divided among 2 hospitalist attending physicians, 2 senior residents, 2 intern physicians, and medical students. One pharmacist covers all the pediatric floor patients on the weekends and is available by phone.

We introduced a stepwise spread of the improvement interventions to other pediatric populations following 9 months of sustained improvement within the PHM service. In September 2021, our population expanded to include all pediatric patients admitted to acute care medical and surgical units involving surgical and subspecialty teams. In December 2021, we again expanded to all pediatric patients admitted, excluding only intensive care units. This resulted in an expansion of inpatients to include cardiology, hematology and oncology, and step-down unit patients; however, the PHM patient cohort remained the same as that population was identified by hospitalist attending assignment.

Notably, the pediatric inpatient units relocated to a new bed tower in mid-December 2021, the 12th month of the intervention period. The move to the new location resulted in a significant redistribution of nurses and pharmacy staffing as patients were distributed over 5 units instead of 4.

For patients admitted to the hospital with a reported PCN allergy, our goal is for documentation to be complete and those with a low or moderate risk of the reaction being an IgE mediated reaction are given an opportunity for delabeling directly or after allergy testing or oral challenge in the allergy clinic. In this project, PCN allergy was defined as a documented allergy to penicillin, amoxicillin, amoxicillin-clavulanate, ampicillin, ampicillin-sulbactam, and piperacillin-tazobactam. Completeness of the allergy label was defined as having the reaction, reaction type, and severity documented in the allergy label within the EHR.

As a part of an institutionally sponsored improvement science course, we assembled an interprofessional team including physicians (hospitalist attending, fellow, and residents), a clinical pharmacist, and a management engineer. We also sought multidisciplinary input from infectious disease, allergy, hospital medicine, and antimicrobial stewardship stakeholders in the creation of the clinical pathway to guide documentation and recommendations based upon risk assessment of reaction history.

We conducted a Qualtrics survey (Qualtrics, Provo, UT) of nurses and pediatric residents (90 respondents, 45% response rate) to identify barriers to drug allergy documentation. The survey was distributed to the relevant pediatric nursing unit and resident listservs and accessed by e-mail link. Anonymous responses using multiple-choice and free text formats were used to answer: “If you do not always document a medication reaction, type, and severity, what are the reasons? (Choose as many as apply)”. Respondents indicated barriers to complete allergy documentation were: reaction details unknown by patient or family (35%), other – free text with themes of another team member’s responsibility, not knowing the correct choice (16%), and not relevant to current admission (6%) (Supplemental Fig 6). Participant perceptions, organized using a Pareto chart based on category frequency, informed our initial key drivers: accountability and comfort of updating allergy label, knowledge of reaction detail, and relevance to admission.

Organized around the identified key drivers, the 3 categories of interventions to address our aims were (1) education across the pediatric department and key stakeholders within the inpatient units, (2) clinical pathway development and implementation, and (3) EHR-based standard documentation and workflow. Each is described in detail below (Fig 1).

Education to trainees, faculty, pharmacists, and nurses were interventions targeting the following key drivers: comfort updating allergy label and relevance to admission. As our study population initially included inpatients on the 3 PHM teams, we began interventions in January 2021 with education for pediatric resident physicians. Education interventions included 2 didactic lectures attended by most residents as a part of their standard weekly conference. An additional 2 conferences were held in July 2021 and March 2022.

In April and May 2021, we presented the improvement initiative to the academic faculty, pharmacists, and nurses through presentations at staff and committee meetings. Presentations focused on education about the current state, dissemination of our project aims, and receiving feedback to inform subsequent interventions. These presentations outlined the importance of complete allergy label documentation, rarity of true PCN hypersensitivity, and our proposal of a clinical pathway to standardize the process of labeling (and delabeling) PCN allergies.

In June 2021 and November 2021, announcements and education sessions were conducted at nursing team meetings on the units. In September 2021 and March 2022, e-mail reminders were sent to all pediatric faculty, resident physicians, pharmacists, and nursing unit listservs about the risk stratification process and included educational materials.

The widespread education included each member of the interprofessional team to emphasize their roles in identifying PCN allergies, inquiring about reaction history, documentation, and risk stratification. The intervention directly targeted both the documentation completeness and delabeling outcome measures, in addition to the process measures of allergy referral and clinical pathway use.

The Penicillin Allergy Risk Stratification Clinical Pathway targeted the following key drivers: standardization of documentation and assessment of the allergy label and provided guidelines to increase comfort updating allergy label. Development of the clinical pathway began shortly after the team was assembled in fall 2020 as we anticipated this would be a key intervention requiring multidisciplinary input. In April 2021, we held a stakeholder meeting discussing the proposal for the clinical pathway with hospital medicine, allergy, infectious disease, pharmacy, and antibiotic stewardship stakeholders. Feedback and adjustments were incorporated, and the pathway was approved by the institution’s Pediatric Quality Program in May 2021. Starting in June 2021, the clinical pathway was implemented on the PHM teams. In September 2021, the clinical pathway was spread to all pediatric patients admitted to the 2 acute care medical and surgical units. In December 2021, we again spread the clinical pathway to all pediatric inpatients, excluding only intensive care units.

The clinical pathway is an evidenced based intervention to guide the documentation and risk stratification of PCN allergies based upon patient and family reported reaction history (Fig 2). The pathway provides a guide for pharmacists, nurses, and clinicians for adequate documentation of reaction history to risk stratify patients’ PCN allergy into low, moderate, or high-risk categories (Supplemental Table 1). PCN allergies were risk stratified on weekdays by the rounding inpatient pharmacist, who then would notify the primary team of the pathway recommendation (delabel, place allergy clinic referral, label to remain in chart). Reported reactions deemed low risk (eg, family history of PCN allergy, isolated gastrointestinal symptoms) are unlikely to be true IgE mediated allergies, and the allergy label can be safely deleted without further evaluation. This was intended to be the primary source of delabeling. Families were given the choices: (a) remove the allergy label, or (b) keep it in the EHR, completely documented and were provided with education regarding the label.

Reactions deemed moderate risk (eg, rash, anaphylaxis) may be suggestive of an IgE mediated reaction, and these patients are recommended to have a pediatric allergy consultation, only available outpatient for our population. A referral to the pediatric allergy clinic was placed by the primary team after discussion with the patient and family. High risk reactions (eg, drug rash eosinophilia and systemic symptoms, interstitial nephritis) are suggestive of antibody mediated, immune complex mediated, or cytotoxic cell mediated hypersensitivities and remained in the chart with adequate documentation.

The clinical pathway provided a standardized method for the determination and documentation of the severity of the reaction. Before the start of the project, low, moderate, and high designations of allergy reactions were at the discretion of the person documenting in the chart. The clinical pathway also provided guidance on which patients should have further allergy testing and those whose allergy should be delabeled. The pathway was intended to improve and provide explicit guidance on allergy label documentation, delabeling, and allergy referrals.

The documentation templates for allergy labels and risk stratification, defined workflow and team roles of allergy reconciliation and risk stratification, and a prompt in the admission history and physical (H&P) targeted the following key drivers: standardization, knowledge of reaction detail, medical record functionality, accountability, and relevance to admission. In March 2021, we added a predefined list to the allergy section of the admission H&P. The list involved 2 options: “type, reaction, severity of each allergy updated or verified” or “unable to verify,” with options for reasons including caregiver not present, allergy details unknown, or emergency. In April 2021, an allergy column was added to PHM shared patient lists in the EHR visible to both pharmacists and clinicians for ease of identifying patients with PCN allergies.

Completing the documentation of PCN allergies in patient charts required a multidisciplinary team workflow. Clinicians, nurses, and pharmacists worked together to complete allergy labels in the EHR. Upon admission, nurses and resident physicians verify and enter all allergy reaction history in the EHR. Once a patient was identified as having a qualifying PCN allergy label, the pharmacist confirms the reaction history and composes a templated note in the EHR to assign a risk category with a suggested action (Supplemental Fig 7). The pharmacist ensures the PCN allergy label has the 3 components of reaction, reaction type, and severity for documentation completeness. In the comment section of the allergy label, the pharmacist writes “Risk stratified on [date] to Low/Mod/High. Recommend Delabeling/Allergy Referral/Allergy to remain in chart.” This comment is viewable to all clinical staff. Additionally, to prevent future relabeling, the note template includes an element that triggers an alert to future users if a patient had an allergy replaced any time after deletion. Once the risk stratification process is completed in the EHR, the pharmacist notifies the primary team of the recommendation. The primary team then discusses delabeling or referral for allergy testing upon discharge if clinically appropriate and considering family preference.

Clinician and patient education handouts were created and hyperlinked within the risk stratification documentation template. Process meetings with the improvement team and pharmacists occurred monthly or every other month between May 2021 and March 2022 to ensure continued optimization of the process.

The EHR prompts and standardized workflow links the complete allergy label documentation directly to the risk stratification clinical pathway.

Our primary outcome measure was the proportion of pediatric inpatients with reported PCN allergy who had complete allergy documentation. This measure was tracked monthly for patients admitted to the PHM service and the non-PHM inpatient population. Our goal percentage of completeness was 70% for PHM patients and 50% for non-PHM patients within 12 months.

The secondary outcome measure was increasing the percentage of pediatric inpatient allergies delabeled per month to 20% within 12 months.

The primary outcome measure of documentation completeness was driven by the workflow standardization of the clinical pathway and associated risk stratification. Documentation completeness includes the reaction, reaction type, and severity. After determining a reaction, only then can a type and severity be assigned. The severity assigned dictates the process of delabeling, allergy referral, or label to remain in chart completely documented. In this way, the documentation completeness and risk stratification (thus delabeling) are closely linked.

Process measures included the number of clinical pathway documentation template uses per month as a surrogate marker for number of patients who underwent complete risk stratification and the number of allergy referrals placed at discharge for patients with PCN allergies. The process measures are linked directly to the use of the guidance provided in the clinical pathway, influencing both the documentation and delabeling outcome measures.

Our balancing measure was the number of adverse drug reactions to PCN among pediatric inpatients reported to the institutional safety reporting system. This balancing measure was chosen to monitor adverse outcomes related to introducing changes in allergy label documentation and performance monitoring.

A project database was established using Clarity tables (Clarity Ventures, Inc. Austin, TX) and included inpatient encounters for patients aged 18 years and younger with a documented PCN allergy in the EHR admitted January 1, 2020 through June 30, 2022. Of note, the same patient could have multiple encounters if the patient had multiple admissions, which are each a unique opportunity to verify allergies.

Project measures were reported monthly in a secure Tableau dashboard (Tableau Software LLC, Seattle, WA). Baseline data were obtained retrospectively for 1 year before the first intervention. PHM patients were tracked as a distinct cohort over time.

Outcome measures were tracked monthly using statistical process control charts (p-charts). Special cause variation was determined using established definitions.²⁶ 

The study was deemed exempt by the University’s Institutional Review Board with a waiver of informed consent.

During the study period, there were a total of 712 (257 PHM, 455 non-PHM) unique patients with a PCN allergy accounting for 735 (264 PHM, 471 non-PHM) total admission encounters. Of the encounters with a PCN allergy label, 288 (105 PHM, 183 non-PHM) were in the baseline period and 447 (159 PHM, 288 non-PHM) were during the intervention period.

For the completeness outcome measures, the baseline was 20% for PHM and non-PHM inpatients. After widespread education and an EHR allergy column added to shared patient lists, special cause variation was noted for PHM cohort in May 2021 (8 points above the centerline), corresponding to 1 month before implementation of the clinical pathway. For the non-PHM cohort, special cause variation was noted beginning September 2021 (1 point above the UCL; 10 of 11 points above the original centerline from February 2021 to December 2021; 4 of 5 points above 1 σ from September 2021 to December 2021), which corresponded with project interventions spread to include non-PHM services. The centerline shifted to 64% for PHM patients (Fig 3A) and 45% for non-PHM pediatric inpatients (Fig 3B).

The average monthly rate of delabeled PCN allergies for all inpatients (number of allergies delabeled divided by total number of PCN allergies) was 7% during the baseline period and remained unchanged for the duration of the study period. No special cause variation was noted during the study period (Fig 4).

Over the 13 months following the implementation of the clinical pathway, 98 unique patients (109 total PCN allergies) were risk stratified, resulting in 34 outpatient allergy referrals at discharge for further evaluation and allergy testing (Fig 5). Of the 109 allergies: 10 were low risk, 82 were moderate, 12 were high, and 5 were unspecified. Upon retrospective review, 6 patients were risk stratified while inpatient to moderate risk and completed the outpatient allergy appointment. All 6 patients had rash or hives listed as the reaction to PCN. Two of the 6 patients were delabeled after negative skin testing or direct oral challenge. The other 4 patients were not tested on the day of the appointment because of age or current antihistamine use but will be scheduled for testing in the future.

No adverse drug reactions to PCN were reported during the baseline or intervention time periods.

Using the Model for Improvement Methodology and Creation and Implementation of a Clinical Pathway with associated documentation protocols, this project provided an institutional standard for the documentation and risk stratification of PCN allergies. We observed a significant increase in the completeness of PCN allergy documentation both in the PHM cohort and the broader non-PHM population, although the increase did not meet our initial goals. We attribute the improvement in allergy documentation to establishing an institutional standard for PCN allergy documentation for admitting physician teams and nurses in addition to risk stratification via the clinical pathway driven by a reliable team of inpatient pharmacists.

The substantial increase in the percent of patients with complete PCN allergy documentation represents the multidisciplinary effort of pharmacists, physician teams, and nurses. For PHM patients, the observed shift occurring 1 month before implementation of the clinical pathway likely represents a response to the extensive departmental education in preparation for the launch. With educational efforts emphasizing the individual and public health consequences of carrying a PCN allergy label and the rarity of true PCN allergy, we were able to establish acceptance of the added work of risk stratification on the PHM services. The observed shift in the non-PHM cohort corresponded to the spread of the clinical pathway to the initial group of non-PHM patients in September 2021 and was sustained with additional spread in December 2021, although notably declined in January 2022, likely related to the move to the new bed tower.

Accountability of allergy documentation was placed on admitting physician teams and nurses, with the clinical pathway as a guide for linking reaction history and severity. Documentation was checked for accuracy and completeness during the pharmacists’ risk stratification workflow. The success of increasing the allergy completeness particularly on the PHM teams was because of the primary project leaders and stakeholders being PHM faculty and trainees, in addition to the consistency of pharmacists who round on the weekdays directly with the physician teams. There were 3 primary pharmacists who became familiar with the risk stratification workflow and met regularly with project leaders. Notably, 70% (69 of 98) of risk stratifications were completed by the 2 inpatient clinical pharmacists who round primarily with the 3 PHM teams.

There are several limitations in our project. First, the project was conducted at a single institution and may not be widely generalizable. Additionally, it proved challenging to elicit buy-in from subspecialty and surgical service pharmacists, nurses, advance practice providers, and physician teams particularly for patients who did not require antimicrobial agents during admission. Because the subspecialty and surgical services comprise about 10 different rounding teams with frequently rotating members, it was difficult to target education and reminders to use the clinical pathway. As a result of factors including the severe acute respiratory syndrome coronavirus 2 pandemic, move of the pediatric units to a new bed tower with staff redistribution and intermittent staffing shortages, our patient population was no longer confined to a particular unit.

Furthermore, weekend and holiday coverage and the added responsibilities of pharmacists and other health care team members played a role in inadequate documentation of allergies as pharmacists were only able to perform risk stratification on weekdays. Because of high patient census, acuity, and intermittent staffing shortages, the pharmacists were unable to dedicate time to daily PCN allergy risk stratification, and it proved challenging to ensure it continued to be a part of the admission workflow for admitting teams and nurses. Also, only PHM, Hematology and Oncology, and Cardiology teams have a dedicated pharmacist present on rounds during weekdays. Having dedicated staff, such as pharmacy technicians, documenting allergy histories and medication reconciliations or incorporating the risk stratification process into antimicrobial stewardship team workflows could improve the process. Documentation completeness of PCN allergies could likely be further improved with project team weekly reviews of admitted patients with a PCN allergy and personalized feedback to attending physicians, senior residents, charge nurses, and pharmacists.

Unfortunately, we did not observe a change in the frequency of delabeled allergies. There was a notable, but not significant or sustained, increase in delabeled allergies in the 7 months following the start of interventions in January 2021, which may be related to a Hawthorne effect associated with intervention monitoring. The 7% average rate of delabeling, unchanged with our interventions, may reflect the robust antimicrobial stewardship program across the pediatric and adult hospitals present before the start of our interventions. Notably, the adult hospital has a well-established PCN allergy testing and oral challenge protocol to drive delabeling for inpatients. Additionally, our PHM team has many medicine and pediatric colleagues who frequently work as hospitalists both on the adult and pediatric sides of the hospital, which may have contributed to a higher baseline awareness of PCN allergy inaccuracies.

To reach a consensus among stakeholders combined with the inability to perform inpatient allergy testing or direct oral challenge, the clinical pathway was intentionally conservative, considering that an unknown reaction or isolated rash are designated as moderate risk, thus indications for outpatient allergy referral. The number of patients requiring outpatient referral for further allergy testing is likely overestimated in our protocol. The percentage of patients who completed the allergy appointment was quite low at 18% (6 of 34), but 2 of the 6 patients were delabeled after testing in the allergy clinic. The scope of the current project is limited to the inpatient setting, but an area of future improvement should focus on connecting patients to outpatient allergy visits from referrals placed while inpatient.

Additionally, another limitation is, although the clinical pathway was conservative, many families and clinicians remained hesitant about removing an allergy label on a patient chart, even if a low-risk reaction was documented. Some reasons cited for these labels remaining were: a preference to keep it in the EHR to ensure monitoring if a PCN was prescribed in the future or severity of intolerance symptoms (ie, diarrhea). Although not currently an option at our institution while inpatient, direct amoxicillin oral challenge in certain patients could be considered as a future intervention to improve the level of confidence among clinicians and patients and families to delabel the allergy.²⁷  Another consideration to overcome hesitancy around allergy delabeling low risk reactions would be the addition of a patient and family advisory council member to the project leadership to guide future efforts. Even if individuals declined delabeling the allergy, the rounding team clinicians or pharmacists provided education and completed the allergy label documentation, noting the preference for the label to remain in the chart.

We hoped more patients could be directly delabeled based upon eliciting reaction history alone without further testing. A benefit of this risk stratification pathway is, unlike other studies which included point of care allergy testing,¹³  this protocol can be used in settings in which allergy testing and oral challenge are not available or not clinically appropriate and can be used by a wide variety of healthcare workers or even automated in the EHR. Although the clinical pathway did not contribute to an increased frequency of delabeling, it did provide guidance and standardization of allergy documentation at our institution.

Of note, another limitation is that our balancing measure does not capture the outpatient continuum of patient reactions after delabeling. We acknowledge the possibility that a patient whose allergy was delabeled inpatient may receive a PCN in the outpatient setting and thus an adverse reaction would not be captured.

Next steps include improvements to the referral process, ensuring eligible patients are risk stratified, allergy referrals placed, and patients follow up in allergy clinic upon discharge. Future directions will be focused on continuing to retrospectively study the results of allergy testing performed in the clinic to determine the frequency of outpatient delabeling as a result of our inpatient improvement efforts. Further next steps will measure the impact of complete allergy labels on β-lactam and broad-spectrum antibiotic prescribing. Improving allergy documentation in pediatric patients allows for more accurate assessments of allergy labels and may improve antibiotic selection.

Completely documented PCN allergy labels, although initially rare in our population, were significantly increased through education to key groups, introduction of a risk stratification clinical pathway, and a multidisciplinary standardized workflow. Documentation completeness and risk stratification did not change the frequency of delabeling. Barriers to delabeling should be considered in future efforts.