Validation of body mass index (BMI)-related ICD-9-CM and ICD-10-CM administrative diagnosis codes recorded in US claims data.

Abstract

To quantify the sensitivity and positive predictive value (PPV) of body mass index (BMI)-related ICD-9-CM and ICD-10-CM diagnosis codes in claims data. De-identified electronic health record (EHR) and claims data were obtained from the Optum Integrated Claims-Clinical Database for cross-sections of commercial and Medicare Advantage health plan members age≥20years in 2013, 2014, and 2016. In each calendar year, health plan members' BMI as coded in the insurance claims data (error-prone measure) was compared with their BMI as recorded in the EHR (gold standard) to estimate the sensitivity and PPV of BMI-related ICD-9-CM and ICD-10-CM diagnosis codes. The unit of analysis was the person-year. The study sample included 746763 distinct health plan members who contributed 1116283 eligible person-years (median age 56years; 57% female; 65% commercially insured and 35% with Medicare Advantage). BMI-related diagnoses were coded for 14.6%. The sensitivity of BMI-related diagnoses codes for the detection of underweight, normal weight, overweight, and obesity was 10.1%, 3.7%, 6.0%, and 25.2%, and the PPV was 49.0% for underweight, 89.6% for normal weight, 73.4% for overweight, and 92.4% for obesity, respectively. The sensitivity of BMI-related diagnosis codes was higher in the ICD-10-CM era relative to the ICD-9-CM era. The PPV of BMI-related diagnosis codes for normal weight, overweight, and obesity was high (>70%) but the sensitivity was low (<30%). BMI-related diagnoses were more likely to be coded in patients with class II or III obesity (BMI ≥35kg/m2 ), and in 2016 relative to 2013 or 2014.