Assessment of the validity of international classification of disease tenth revision codes for atopic dermatitis

Little is known about the validity of ICD-10-CM codes for atopic dermatitis (AD) in healthcare claims databases. We assessed the validity of ICD-10-CM codes for identifying adult patients with AD. The healthcare claims database from a large metropolitan tertiary care medical center was queried for diagnostic codes of L20.x and L30.9. Medical records were reviewed for demographics, comorbidities, Hanifin & Rajka (H-R), and United Kingdom Working Party (UKWP) criteria. Sensitivity, specificity, and positive predictive values (PPVs) were calculated. Overall, 833 patients were identified with ≥ 1 occurrence of the aforementioned ICD-10 codes for AD. Using H-R and UKWP criteria as the gold-standard definitions of AD, the PPV of any aforementioned L20.x ICD-10 codes was 39.1% and 51.4%, with sensitivity of 98.1% and 97.4%, respectively. The PPV was 25% and 50% for L20.82; 66.7% and 100% for L20.84; 37.5% and 62.5% for L20.89; 60.1% and 51.4% for L20.9, respectively. PPV generally increased when case definitions combined ICD-10 codes with coexistent ICD-10 codes for asthma or hay fever, food allergy, and allergic rhinitis, and to a lesser extent coding of the ICD-10 code for AD by a dermatologist. Among patients with all these criteria, PPV increased to 100% and 80% based on H-R and UKWP criteria, but sensitivity decreased to 83.3% and 80.0%, respectively. In conclusion, ICD-10-CM codes for AD alone had poor PPV for identifying AD in a healthcare claims database. Combined ICD-10 codes for AD and comorbid atopic disease improved PPV and specificity of classification at the expense of sensitivity.