Risk-stratified Management to Remove Low-Risk Penicillin Allergy Labels in the ICU

To the Editor:

Between 8% and 15% of the U.S. population carries a penicillin allergy label, yet <5% of these can be verified by allergy testing (1). A false label has a negative impact on care, including use of broader-spectrum and second-line antibiotics, increased healthcare utilization, surgical site infections, and treatment failure for common infections, delayed antimicrobial therapy, and longer lengths of stay (1, 2). β-Lactam allergy labels also affect antimicrobial stewardship and are associated with increased infections with Clostridium difficile, methicillin-resistant Staphylococcus aureus, and vancomycin-resistant Enterococcus (3–6).

Penicillin allergies are commonly diagnosed in childhood and go unquestioned throughout life (7). However, these allergy diagnoses are largely inaccurate, explained instead by viral exanthems, drug–viral interactions, or nonallergic side effects. Even for penicillin allergies verified by skin testing, ≥10% lose reactivity every year without evidence of resensitization (8, 9).

Patients admitted to a medical ICU (MICU) often have chronic illnesses or altered immunity, increasing their need for immediate antibiotic use. We sought to determine whether MICU patients with low-risk penicillin allergy history could be challenged directly with amoxicillin to have their allergy label safely removed during an acute inpatient stay. Some results of this study have been reported in abstract form (10).

Methods

Penicillin allergy risk stratification tool

We developed a history-based risk-stratification tool to identify patients with a penicillin allergy label at low risk of having a true allergy (2). During routine clinical encounters in our outpatient drug allergy clinic, patients provide a history of their index reaction to penicillin and are appropriately skin tested using a standardized panel of reagents. After a negative skin test, patients proceed to an observed oral challenge with amoxicillin 250 mg. When oral challenge is asymptomatic after a 1.5-hour observation period and a 24-hour follow-up nursing phone call to assess delayed skin test reactions and symptoms, a patient’s penicillin allergy label is removed. Data from these visits were used to derive and validate our risk-stratification algorithm. This project was approved by the Vanderbilt institutional review board (#181180 and #181734).

Data from 318 consecutive patients seen from 2014 to 2018 were collected using standardized chart review and data collection and were categorized via historical assessment by a physician into one of three groups: 1) highest risk-delayed reactions (blistering, mucosal involvement, severe rash, and/or immune-mediated organ injury); 2) moderate to high risk, where patients reported symptoms consistent with standard anaphylaxis criteria; or 3) low risk, where patients reported symptoms that were inconsistent with either of the higher-risk criteria (Figure 1). The outcomes of allergy testing in the clinic were compared with history-based risk criteria to estimate the negative predictive value for a positive penicillin allergy skin test and oral challenge among low-risk patients.

Figure 1.

Risk stratification of a historical penicillin reaction. In patients who were already admitted to the ICU, historical reactions consistent with a low-risk penicillin allergy were offered direct challenge with 250 mg oral amoxicillin followed by a 1-hour observation period.

Penicillin allergy delabeling study

All patients admitted to the MICU between March 31, 2019, and October 31, 2019, who were both hemodynamically stable and could provide a history of their index reaction were screened. Those whose reported index reaction was consistent with a low-risk penicillin allergy were offered direct challenge with 250 mg oral amoxicillin followed by 1-hour observation. Consent for the clinical procedure was obtained using a structured conversation guide. Patients lacking evidence of reaction after observation had their allergy label removed from their record at the point of care and were provided with a letter and wallet card detailing the delabeling procedure’s implications.

Results

Validation of a penicillin allergy risk stratification tool

Of 318 drug allergy clinic patients, 195 (61%) were identified as low risk. The negative predictive value of low-risk categorization was 99% (95% confidence interval, 96–100%). Two low-risk patients had positive skin tests to penicillin G and ampicillin, respectively, and were not challenged. Of low-risk patients who agreed to undergo single-dose oral challenges, 184 of 184 (100%) were asymptomatic, demonstrating they no longer needed to be considered allergic to penicillin.

Penicillin allergy delabeling in the MICU

Over 7 months, there were 216 of 1,859 (11.6%) MICU patients admitted with a penicillin allergy label. Of these, 114 of 216 (53%) were eligible for evaluation during their stay, and all were evaluated. A total of 68 of 114 (60%) eligible patients had a low-risk history. A total of 54 of 68 (79%) low-risk patients agreed to observed amoxicillin oral challenge, of whom 54 of 54 (100%; 95% confidence interval, 93.4–100) had no immediate or delayed symptoms after direct challenge with amoxicillin 250 mg. These patients were counseled on the removal of their penicillin allergy and had their allergy removed from their chart (Table 1).

Table 1.

Characteristics and Outcomes of Risk-stratified Amoxicillin Oral Challenges Leading to Allergy Label Removal in ICU Patients (N = 114)

Risk Group	Low Risk: Challenge Performed (n = 54)	Low Risk: Declined Challenge (n = 14)	Higher Risk (Both Groups) (n = 46)
Median age (IQR), yr	57 (46–66)	48 (41–59)	50 (41–65)
Sex, F	24 (44)	5 (36)	29 (63)
Race
White	51 (94)	12 (86)	41 (89)
Black	3 (6)	2 (14)	4 (9)
Asian	0 (0)	0 (0)	1 (2)
Non-Hispanic ethnicity	54 (100)	14 (100)	44 (96)
Immediate challenge outcome
Tolerated amoxicillin oral challenge, leading to allergy removal from the chart	54 (100)	N/A	N/A
Primary outcomes at 7 mo
Subsequent use of a penicillin treatment course	17 (31)	2 (14)	2 (4); 1 direct treatment, 1 after a desensitization
More than one subsequent penicillin treatment	4 (7)	1 (7)	0 (0)
No. of penicillin labels reentered into the chart during subsequent care	1 (2)	N/A	N/A
Secondary outcomes at 7 mo
Subsequent use of a cephalosporin treatment course	24 (44)	9 (64)	21 (46)
More than one subsequent cephalosporin treatment	9 (17)	2 (14)	8 (17)

Definition of abbreviations: IQR = interquartile range; N/A = not applicable.

Data are presented as n (%) unless otherwise noted.

Several patients had an indication for and were subsequently treated with multiple doses of a penicillin (17 of 54; 31%) or cephalosporin (24 of 54; 44%) during the same hospital stay or later health care encounters in our system, without report of allergic reaction. A total of 23 penicillin and 37 cephalosporin treatment courses in 30 delabeled patients were given after the removal of their penicillin allergy label. During the 7-month study period, only one patient, who experienced nausea and diarrhea associated with amoxicillin-clavulanic acid, was relabeled as penicillin allergic.

Discussion

Incorrect diagnoses of penicillin allergy are a problem that adversely affects patient care, public health, antimicrobial stewardship, and health care costs (1, 2). Although recent recommendations promote direct challenge for low-risk penicillin-allergic children, there has been limited evidence supporting direct oral challenge in adults, in whom the reported severity of the index reaction can vary (2). Tucker and colleagues demonstrated that in healthy Marine recruits, most patients reported a low-risk penicillin allergy, and therefore direct challenge only elicited symptoms in 5 of 328 (1.5%) patients (11). Our results demonstrate the safety of direct oral challenge among critically ill patients in a MICU with a low-risk penicillin allergy. When applying our results to the critical care setting, it is important to note that we only offered challenges to patients who were stabilized and able to participate. It is also important to note that our criteria for low risk includes some categories (urticaria >5 yr ago, cutaneous rash) that are deemed as moderate risk in recently proposed approaches (12).

Our study was conducted in a MICU, a highly controlled environment with staff experienced in managing allergic reactions. Oral challenge was safely tolerated in our low-risk group (total n = 238), both retrospectively and prospectively. Extrapolation of our results in adults beyond the ICU environment remains unknown, but we anticipate that our approach could be safely implemented in a wide variety of settings.

Skin testing, as an additional step in disproving a penicillin allergy, may be unnecessary in low-risk patients, adding expense and time or being refused by the patient. Penicillin skin testing is also subject to false negatives in the critical care setting (13). Other studies have used full- or split-dose challenges, often not to the penicillin drug for which the patient was labeled allergic but to structurally unrelated drugs to which they were unlikely to react (e.g., ceftriaxone) (2). Direct oral challenge is safer than full- or split-dose intravenous challenge to the labeled drug in critically ill patients where patient tolerance is unknown. Direct oral challenge is also a procedure that delabels the patient’s reported allergy, therefore creating a much larger number of therapeutic choices (2). The formalized ritual of an amoxicillin challenge providing a transition from “allergic” to “nonallergic” may be an important element of our intervention’s success, requiring further study.

Moreover, our results suggest that desensitization, the current approach to initiate a penicillin treatment in someone who reports anaphylactic-like symptoms, is unnecessary for low-risk patients. Systematic application of risk-stratified penicillin allergy management could reduce the prevalence of adverse outcomes associated with penicillin allergy labels.