Validity of Lichen Planus and Lichen Planopilaris Case Identification Using Diagnostic Codes from a Clinical Database

Background: Case identification strategies to conduct population-based studies have not been developed for lichen planus (LP) or lichen planopilaris (LPP). Objectives: To assess the validity of using diagnostic codes to establish an overall LP cohort, a cutaneous (non-oral) LP cohort, and an LPP cohort from a large clinical database. Methods: A retrospective chart review was performed to determine whether patients with ICD-9 or ICD-10 codes for LP and ICD-10 codes for LPP are confirmed cases of LP and LPP. Validation samples were used to estimate the positive predictive value (PPV) of three case definitions any LP, non-oral LP, and LPP defined as: at least one code by any physician, at least two codes by any physician, and at least one code by a dermatologist. Results: Among the 199 reviewed LP charts, 166 and 123 were confirmed cases of any LP and non-oral LP, respectively. The PPVs for any LP were: 83.4% (166/199) for one code by any physician, 84.6% (77/91) for two codes by any physician, and 95.1% (97/102) for one code by a dermatologist. The PPVs for non-oral LP were: 61.8% (123/199) for one code by any physician, 70.3% (64/91) for two diagnoses by any physician, and 86.3% (88/102) for one diagnosis by a dermatologist. Of the 139 patients with at least one code for LPP, 122 were confirmed cases of LPP. The case definition for one LPP code applied by any physician had a PPV of 87.8% (122/139) to identify a true case of LPP, whereas two diagnoses by any physician had a PPV of 96.2% (76/79) and a diagnosis by a dermatologist had a PPV of 93% (107/115). Conclusions: LP and LPP diagnosis codes applied by any physician may generate roust and specific case cohorts for any form of LP and LPP, respectively. The use of LP codes applied by a dermatologist may be necessary to accurately identify non-oral LP cases.

Automated summary

The study aimed to validate the use of diagnostic codes to establish LP and LPP cohorts. The results showed that LP and LPP diagnosis codes applied by any physician can generate reliable and specific cohorts, while the use of LP codes applied by a dermatologist is necessary to accurately identify non-oral LP cases.