Inter-rater variability in patch test readings and final interpretation using store-forward teledermatology

Background: Data regarding teledermatology for patch testing are limited. Objectives: Compare patch test readings and final interpretation by two in-person dermatologists (IPDs) with eight teledermatologists (TDs). Methods: Patch tested patients had photographs taken of 70 screening series of allergens at 48 hours and second readings. Eight TDs reviewed photos and graded reactions (negative, irritant, doubtful, +, ++, +++) at 48 hours and second readings; in addition, they coded a final interpretation (allergic, indeterminant, irritant, negative) for each reaction. TDs rated overall image quality and confidence level for each patient and patch test reaction, respectively. Percentage of TD–IPD agreement based on clinical significance (success, indeterminate, and failure) was calculated. Primary outcome was agreement at the second reading. Results: Data were available for 99, 101, and 66 participants at 48 hours, second reading, and final interpretation, respectively. Pooled failure (+/++/+++ vs negative) at second reading was 13.6% (range 7.9%-20.4%). Pooled failure at 48 hours and final interpretation was 5.4% (range 2.9%-6.8%) and 24.6% (range 10.2%-36.8%), respectively. Confidence in readings was statistically correlated with quality of images and disagreement. Conclusion: For patch testing, teledermatology has significant limitations including clinically significant pooled failure percentages of 13.6% for second readings and 24.6% for final interpretation.