OBJECTIVES: Undercoding of obesity disease codes in administrative health claim data sets is a well-known problem; the absence of this documentation possibly results in reduced preventive, treatment, and care measures, as patients cannot be quickly identified. Predictive models based on obesity-related comorbidities can be used to identify patients with a high probability of living with obesity. As most obesity-related comorbidities are also associated with advanced age, the aim is to investigate whether the association between these comorbidities and obesity decreases with advanced age in a predictive model.

METHODS: Based on patients’ characteristics (as risk factors), a predictive model analyzed stratified statutory health insurance claims data in Germany (n=3,244,611) and UK biobank data (n=502,384) for verification.

Associations between obesity-related comorbidities and obesity were assessed using a logistic regression by comparing age-stratified odds ratios (OR) of the risk factors within a joint model in age groups 40–49, 50–59, and 60–69.

RESULTS: Multiple comorbidities were identified, with strongest associations between coded obesity and sleep apnea with an OR of 4.03, 3.33, and 2.96 for age groups 40–49, 50–59, and 60–69, respectively in German claims data (UK biobank: 3.11, 2.65, 2.32, respectively); type 2 diabetes with 4.13, 3.46, and 2.91 (UK biobank: 2.81, 2.50, 2.10, respectively); and hip/knee osteoarthritis with 2.31, 2.14, and 1.94 (UK biobank: 4.07, 3.42, 2.73, respectively). In both databases, decreasing ORs were observed with increasing age for almost all obesity-related comorbidities, suggesting a robust underlying correlation, despite varying documentation standards between the countries.

CONCLUSIONS: Models based on obesity-related comorbidities allow patients with a high probability of obesity to be identified. Since the association between these comorbidities varies by age, these models can be further improved by age-dependent assessment of these comorbidities.