Validity of a Claims-Based Diagnosis of Obesity Among Medicare Beneficiaries

Abstract

Population-level data on obesity are difficult to obtain. Claims-based data sets are useful for studying public health at a population level but lack physical measurements. The objective of this study was to determine the validity of a claims-based measure of obesity compared to obesity diagnosed with clinical data as well as the validity among older adults who suffer from chronic disease. This study used data from the National Health and Nutrition Examination Survey 1999-2004 for adults aged ≥ 65 successfully linked to 1999-2007 Medicare claims (N = 3,554). Sensitivity, specificity, positive and negative predictive values, κ statistics as well as logistic regression analyses were computed for the claims-based diagnosis of obesity versus obesity diagnosed with body mass index. The claims-based diagnosis of obesity underestimates the true prevalence in the older Medicare population with a low sensitivity (18.4%). However, this method has a high specificity (97.3%) and is accurate when it is present. Sensitivity was improved when comparing the claim-based diagnosis to Class II obesity (34.2%) and when used in combination with chronic conditions such as diabetes, congestive heart failure, chronic obstructive pulmonary disease, or depression. Understanding the validity of a claims-based obesity diagnosis could aid researchers in understanding the feasibility of conducting research on obesity using claims data.

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

“…Lloyd et al assessed the validity of obesity diagnoses recorded in Medicare claims data in 1999 to 2007 via patient‐level linkage to BMI data from the National Health and Nutrition Examination Survey (NHANES). In their cohort, the sensitivity and PPV of BMI‐related diagnoses for obesity were 18.4% and 73.6%, respectively . Consistent with our results, Lloyd et al found that individuals with more severe obesity were more likely to have a diagnosis coded in the claims data.…”

“…Unfortunately, the coding of overweight/obesity and BMI is inconsistent in administrative databases. Prior studies in the USA and Canada indicate that International Classification of Diseases (ICD) diagnosis codes for obesity recorded in administrative data have a high positive predictive value (PPV; 74%‐91%) but low sensitivity (18%‐19%) . There is limited evidence from a study in women with breast cancer that the sensitivity of administrative diagnosis codes for overweight/obesity may have improved over time in the 2000s .…”

“…Prior studies in the USA and Canada indicate that International Classification of Diseases (ICD) diagnosis codes for obesity recorded in administrative data have a high positive predictive value (PPV; 74%-91%) but low sensitivity (18%-19%). [4][5][6][7] There is limited evidence from a study in women with breast cancer that the sensitivity of administrative diagnosis codes for overweight/obesity may have improved over time in the 2000s. 6 However, this trend has not been confirmed in a broader population, and prior studies have relied on administrative data records from 2011 and earlier.…”

International Classification of Diseases (ICD)-coded obesity predicts risk of incident osteoporotic fracture

“…It is impracticable to measure body weight and height for each individual in the population. Therefore, a number of studies have explored the use of diagnosis information captured in population-based administrative healthcare data for ascertaining obesity [ 7 , 8 ]. Specifically, International Classification of Diseases (ICD) codes have been used to ascertain individuals who are obese.…”

“…Specifically, International Classification of Diseases (ICD) codes have been used to ascertain individuals who are obese. Studies have demonstrated ICD-coded obesity has low sensitivity but high specificity [ 7 , 8 ]. However, there has been limited research about the predictive value of ICD-coded obesity for major chronic conditions at the population level.…”