ICD-10 Code Demonstrates High Level of Accuracy for Identifying Vitiligo

The level of accuracy of the ICD-10 code for vitiligo (L80) has been validated, according to new findings, with its accuracy being the strongest when codes are coupled with treatment data.1

These findings resulted from a multi-institutional chart review conducted to provide validation for the ICD-10 code for this autoimmune skin disorder. The study's authors, led by such investigators as Debby Cheng, BA, from Massachusetts General Hospital, highlight that in research relying on electronic medical records, the ICD-10 code L80 is often used to implemented. Nevertheless, prior to their analysis, the code's diagnostic accuracy in US populations had not yet been fully validated.2

“The ICD-10 code for vitiligo (L80) is commonly used in electronic medical record–based studies, yet its diagnostic accuracy in US populations remains unvalidated,” Cheng and coauthors wrote.1,2 “We conducted a multi-institutional medical record review to assess the positive predictive value (PPV) of L80 across several diagnostic algorithms.”

Study Design Details

Using the Mass General Brigham Research Patient Data Registry, including patients from Newton-Wellesley Hospital, Massachusetts General Hospital, Brigham and Women’s Hospital, and affiliated centers who had at least a single primary diagnosis entry of L80 between January 2017 - January 2025, data was gathered by the investigative team.

There were 4 approaches evaluated for case evaluations in this cross-sectional: (1) at least a single L80 code, (2) at least 2 codes recorded within 6 months, (3) at least 2 codes within 12 months, and (4) at least a single code plus documentation of a standard treatment for vitiligo. Cheng and colleagues noted that confirmation of vitiligo required clinician documentation naming this disease as the preferred diagnosis. They graded confidence in clinicians' diagnosis as on the higher end (≥3 secondary criteria), moderate (2 criteria), or the lower end (0–1 criterion) using predefined secondary features.

All of Cheng et al's reviewers prior to this analysis demonstrated ≥90% concordance on both primary and secondary criteria in a 10-record test set. Sensitivity analyses were performed to exclude cases shown to have a lower level of diagnostic confidence. They also sought to explore records coded by clinicians that are not dermatologists, for the purposes of testing nonspecialist setting generalizability. There were 300 eligible records among 3454 selected by the investigators for their detailed assessment.

Among these, confirmation of vitiligo was observed in 285 subjects. Among these individuals, there was a mean age of 48.2 years. 52.6% of these individuals were female and 47.4% male. Cheng and coauthors highlighted that the PPV for a single L80 code was 0.95 (95% CI, 0.92–0.97). Requiring a pair of codes within either 6 months or 12 months was found to have yielded a PPV of 0.98 (95% CI, 0.94–1.00) in both of these time frames. The combination of at least a single code plus treatment attained the highest PPV at 0.99 (95% CI, 0.97–1.00).

The investigative team highlighted that diagnostic confidence levels were distributed as high in 73.3% of cases, moderate in 20.4%, and 6.3%. The team noted that, when low-confidence cases were excluded, PPVs continued to be robust across all of the algorithms: 0.95 (95% CI, 0.93–0.98) for 1 code, 0.98 (95% CI, 0.96–1.00) for 2 within 6 months, 0.98 (95% CI, 0.96–1.00) for 2 within 12 months, and 0.99 (95% CI, 0.98–1.00) for 1 code plus treatment.

There was an independent sample of 100 patients assigned an L80 code by non-dermatologists that Cheng and colleagues reviewed. Among this cohort, with 95 having been confirmed to have vitiligo, the PPV was 0.95 (95% CI, 0.89–0.98) for a single code, 1.00 (95% CI, 0.92–1.00) for 2 codes within 6 months, 1.00 (95% CI, 0.94–1.00) for 2 codes within 12 months, and 1.00 (95% CI, 0.95–1.00) for at least a single code plus treatment.

“Compared with an ICD-9–based study, which yielded a PPV of 33.7% for vitiligo and 85.5% when combined with treatment data, our ICD-10–based algorithms demonstrated consistently higher PPVs,” the investigative team concluded.1 “This likely reflects the improved specificity of ICD-10 coding for vitiligo, which uses a dedicated diagnostic code (L80) rather than the broader dyschromia category used in ICD-9.”

References

1. Cheng D, Sanchez K, Biba U, et al. Validating the Use of ICD-10 Codes for Identifying Vitiligo. JAMA Dermatol. Published online October 01, 2025. doi:10.1001/jamadermatol.2025.3512.

2. Bell M, Lui H, Kalia S, et al. Validation of medical service insurance claims as a surrogate for ascertaining vitiligo cases. Arch Dermatol Res. 2023 Apr;315(3):541-550. doi: 10.1007/s00403-022-02383-7. Epub 2022 Sep 29. PMID: 36173455.