Diagnosis of peanut allergy using continuous likelihood ratios

Area of Science:

• Allergy and Immunology
• Diagnostic Methods
• Biostatistics

Background:

• Peanut allergy (PA) diagnosis relies on peanut specific immunoglobulin E (sIgE) tests, often categorized using a 0.35 kU/L cutoff.
• This categorization presents a significant economic and psychological burden on patients and families.
• Bayes' theorem offers a continuous interpretation of sIgE values, incorporating PA prevalence and ingestion history.

Purpose of the Study:

• To evaluate a novel method for estimating the likelihood ratio (LR) for each sIgE value.
• To utilize oral food challenge results in a pilot investigation to refine diagnostic accuracy.
• To establish a basis for estimating the probability of PA using continuous sIgE data.

Main Methods:

• Collected oral food challenge outcomes and serum sIgE values from 117 children.
• Employed polynomial regression of the receiver operating characteristic (ROC) curve to determine sIgE-specific LRs.
• Utilized linear regression for estimating LRs across the range of sIgE values.

Main Results:

• sIgE values ranged from <0.1 kU/L to 35 kU/L.
• Bayes' theorem and ROC curve analysis yielded LRs for each peanut sIgE value.
• An sIgE value of 0.22 kU/L corresponded to an LR of 1, comparable to the traditional 0.35 kU/L cutoff.

Conclusions:

• The developed method allows for a continuous interpretation of sIgE, moving beyond simple positive/negative categorization.
• Integration with pretest probability estimates can lead to computerized algorithms for PA probability assessment.
• This approach promises more accurate and personalized diagnostic decision-making for peanut allergy.