Validity and Predictors of Medical Claims Coding for the Identification of Patients with Obesity.

BACKGROUND: The evolving obesity treatment landscape requires examining clinical and economic outcomes, but ICD-10-CM coding inconsistencies hinder patient identification in claims databases. OBJECTIVES: This study aims to validate obesity diagnosis codes and identify factors influencing their utilization. METHODS: This retrospective cohort study utilized claims and clinical data from the Healthcare Integrated Research Database (HIRD®) to identify adults with a BMI of 30 kg/m² or higher, spanning the period from June 2021 to October 2022. The study examined the use of ICD-10-CM diagnosis codes for obesity across two distinct timeframes: a 6-month period before and after the index date for prevalence and modeling purposes, and a 60-day window around the index date for validity calculations. For validation, the sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) of the ICD-10-CM codes were measured, classifying patients as true or false positives or negatives based on their BMI and coding presence. Multivariable logistic regression was used to identify factors predicting obesity coding, guided by clinical reasoning, statistical results, and refined through backward elimination regression. RESULTS: Out of 2 218 115 patients with available BMI data during the relevant period, 966 427 (44%) had a BMI of 30 kg/m² or above. The study identified 380,606 true positives, 17 422 false positives, 1 234 266 true negatives, and 585 821 false negatives, resulting in a PPV of 95.6%, a NPV of 67.8%, a specificity of 98.6%, and a sensitivity of 39.4%. The prevalence of obesity coding was 54%. Increased likelihood of coding was associated with being female, over the age of 24, non-White, insured by Medicare Advantage, having lower socioeconomic status, higher BMI, and receiving weight management interventions. CONCLUSION: ICD-10-CM codes demonstrate high specificity and PPV but low sensitivity, indicating that while coded cases are likely true obesity, many patients with obesity are not captured in claims.