Optimal methods to detect DRESS (drug reaction with eosinophilia and systemic symptoms) syndrome by electronic medical records

ABSTRACT

PURPOSE: Since drug reaction with eosinophilia and systemic symptom (DRESS) syndrome is very rare and difficult to diagnose, its exact epidemiology is still unknown. If screening tools based on laboratory results or electronic medical records are available, the occurrence of DRESS syndrome can be monitored in real time. METHODS: To screen cases with DRESS syndrome, all the results of both eosinophil and alanine transaminase (ALT) level from July 2014 to June 2015 were analyzed by 36 searching conditions for the signal detection of 7 definite DRESS cases among 199,924 patients during the study period. Those searching conditions were diverse combinations of different cutoff levels of eosinophil and ALT with or without nursing records presenting skin symptoms. Sensitivity, specificity, positive predictive value (PPV) and negative predictive value were calculated for individual searching conditions. RESULTS: As cutoff levels of eosinophil and ALT for screening DRESS increased from 3% to 5% and 40 U/L to 300 U/L, respectively, the sensitivity decreased from 100% to 42.9% and the PPV increased from 0.06% to 13.0%. A combination of eosinophil >10% and ALT >300 U/L which had the highest PPV among 36 search conditions could detect DRESS syndrome by sensitivity 42.9% and PPV 13.0%. When nursing records for skin symptoms were added, PPV was augmented to 21.4%. CONCLUSION: A combination of eosinophil and ALT levels is a useful search condition for the screening of DRESS syndrome. Nursing records can provide an additional increment in PPV.