Investigators compare ICD and CPT codes for estimating the incidence of skin cancer.
The use of ICD codes is a poor proxy measure for the number of skin cancers per patient, according to a recent study.
The findings supported the use of CPT codes for estimating incidence of skin cancer and as a base for using electronic health record (EHR) data to estimate numbers of skin cancers per patient.
Nimay Anand, B.A., and a team of investigators assessed whether standard diagnostic and procedural codes presented in EHRs at a single healthcare system were a valid proxy for estimating the number of overall skin cancers. They conducted a retrospective cohort study of patients seen between July 2008 and June 2018. Patients were included if they had at least one EHR-based diagnostic or procedural code for any skin cancer and at least one pathology report of a skin cancer.
The team separated their sample into three groups based on the presence of CPT codes and pathology records. The first group was composed of internal patients who had more than one CPT code and more than one pathology report of skin cancer. The second group comprised outside patients whose pathology was read at Vanderbilt University Medical Center (VUMC) and who would have more than one pathology report, more than one ICD code, but zero CPT codes relating to skin cancer. Group three contained outside patients referred to VUMC for Mohs surgery or subspecialty oncology management with more than one ICD code and more than one CPT code. Investigators focused on group one.
There were 5,688 patients in the first group with at least one CPT code and at least one confirmed skin cancer. The investigators found a strong linear correlation between the number of pathology records and the number of CPT codes (r = .87). Conversely, the total number of ICD codes was poorly correlated with the records (r = .22). For patients with records but no CPT codes, there was a very poor correlation between ICD codes and the pathology records (r = .06).
CPT codes accounted for more than 75% of the variability in the number of skin cancers and were a better measure for the total number of skin cancers per patient.
A subset of patients had a marginally stronger correlation between CPT codes and confirmed skin cancers. There was a mean difference between the number of CPT codes and the number of pathology records of -1.06.
Among 100 patients with CPT codes and no reports selected randomly for EHR review, the correlation between the number of histologically verified skin cancers from outside pathology and the number of CPT codes for unique lesions was near perfect.
Findings suggested the method was an adequate proxy for the number of skin cancer per patient.
The study, “Validity of Using Billing Codes From Electronic Health Records to Estimate Skin Cancer Counts,” was published online in JAMA Dermatology.