Predicting Penicillin Allergy: A United States Multicenter Retrospective Study.

BACKGROUND: Using the reaction history in logistic regression and machine learning (ML) models to predict penicillin allergy has been reported based on non-US data. OBJECTIVE: We developed ML positive penicillin allergy testing prediction models from multisite US data. METHODS: Retrospective data from 4 US-based hospitals were grouped into 4 datasets: enriched training (1:3 case-control matched cohort), enriched testing, nonenriched internal testing, and nonenriched external testing. ML algorithms were used for model development. We determined area under the curve (AUC) and applied the Shapley Additive exPlanations (SHAP) framework to interpret risk drivers. RESULTS: Of 4777 patients (mean age 60 [standard deviation: 17] years; 68% women, 91% White, and 86% non-Hispanic) evaluated for penicillin allergy labels, 513 (11%) had positive penicillin allergy testing. Model input variables were frequently missing: immediate or delayed onset (71%), signs or symptoms (13%), and treatment (31%). The gradient-boosted model was the strongest model with an AUC of 0.67 (95% confidence interval [CI]: 0.57-0.77), which improved to 0.87 (95% CI: 0.73-1) when only cases with complete data were used. Top SHAP drivers for positive testing were reactions within the last year and reactions requiring medical attention; female sex and reaction of hives/urticaria were also positive drivers. CONCLUSIONS: An ML prediction model for positive penicillin allergy skin testing using US-based retrospective data did not achieve performance strong enough for acceptance and adoption. The optimal ML prediction model for positive penicillin allergy testing was driven by time since reaction, seek medical attention, female sex, and hives/urticaria.

Utility and futility of skin testing to address concerns surrounding messenger RNA coronavirus disease 2019 vaccine reactions.

BACKGROUND: The mechanism of coronavirus disease 2019 (COVID-19) vaccine hypersensitivity reactions is unknown. COVID-19 vaccine excipient skin testing has been used in evaluation of these reactions, but its utility in predicting subsequent COVID-19 vaccine tolerance is also unknown. OBJECTIVE: To evaluate the utility of COVID-19 vaccine and vaccine excipient skin testing in both patients with an allergic reaction to their first messenger RNA COVID-19 vaccine dose and patients with a history of polyethylene glycol allergy who have not yet received a COVID-19 vaccine dose. METHODS: In this multicenter, retrospective review, COVID-19 vaccine and vaccine excipient skin testing was performed in patients referred to 1 of 3 large tertiary academic institutions. Patient medical records were reviewed after skin testing to determine subsequent COVID-19 vaccine tolerance. RESULTS: A total of 129 patients underwent skin testing, in whom 12 patients (9.3%) had positive results. There were 101 patients who received a COVID-19 vaccine after the skin testing, which was tolerated in 90 patients (89.1%) with no allergic symptoms, including 5 of 6 patients with positive skin testing results who received a COVID-19 vaccine after the skin testing. The remaining 11 patients experienced minor allergic symptoms after COVID-19 vaccination, none of whom required treatment beyond antihistamines. CONCLUSION: The low positivity rate of COVID-19 vaccine excipient skin testing and high rate of subsequent COVID-19 vaccine tolerance suggest a low utility of this method in evaluation of COVID-19 vaccine hypersensitivity reactions. Focus should shift to the use of existing vaccine allergy practice parameters, with consideration of graded dosing when necessary. On the basis of these results, strict avoidance of subsequent COVID-19 vaccination should be discouraged.

Predictors of recurrent hypoglycemia following a severe hypoglycemic event among hospitalized patients.

OBJECTIVES: Severe hypoglycemia is associated with poor hospital outcomes, but variables contributing to the adequacy of treatment have not been described. The objective of this study was to determine predictors of recurrent hypoglycemia among hospitalized patients with a severe hypoglycemic event. METHODS: Patients with severe hypoglycemia (glucose <40 mg/dl) with a concomitant insulin order were identified using the study institution's Information Warehouse. The primary outcome was the prevalence of recurrent hypoglycemia (defined as <70 mg/dl within 24 hours) and to identify independent predictors of recurrent hypoglycemia. Secondary outcomes included time to blood glucose recheck, time to blood glucose ≥ 70 mg/dl, and rebound hyperglycemia (defined as glucose >300 mg/dl within 24 hours). Multivariable linear and logistic regression models were performed. RESULTS: A total of 129 patients with severe hypoglycemia were identified. The median time to repeat glucose measurement was 29 (IQR 15-61) minutes, while the time to resolution of hypoglycemia was 49 (IQR 26-103) minutes. Recurrent hypoglycemia occurred in 49% of patients, while 19% of patients experienced rebound hyperglycemia. Independent predictors of recurrent hypoglycemia included lower repeat glucose (p = 0.025), low glomerular filtration rate (p = 0.033), and lack of insulin adjustment (p = 0.012). Independent predictors of maximum glucose post-event were type 1 diabetes (p = 0.0003), history of any diabetes (p = 0.013), and total bolus dose of insulin (p < 0.0001). Overnight timing of events was the only predictor of shorter time to hypoglycemia resolution (p < 0.0001). CONCLUSIONS: Recurrent hypoglycemia following severe hypoglycemia is common in the hospital, suggesting the need for enhanced monitoring in such patients. Further research is needed to identify methods to reduce the incidence of recurrent hypoglycemia.