Underdocumentation of Obesity in the National Inpatient Sample

This study aimed to characterize the underdocumentation of obesity by comparing the rates of obesity in databases based on measured BMI, BMI measured while inpatient, and obesity ICD-10 codes input by health providers for inpatients in the United States.

We compared three datasets to test the prevalence of obesity in the United States general population and the inpatient hospital in the United States. The 2017-2018 National Health and Nutrition Examination (NHANES) dataset was used to estimate the prevalence in the general population. For the inpatient hospitals, we used data from the 2017 National Inpatient Sample (NIS) and the 2017-2018 American College of Surgeons-National Surgical Quality Improvement Program (NSQIP) datasets.

There were 5434 participants in the NHANES, 2,006,047 participants in the NSQIP, and 6,084,184 participants in the NIS databases for analysis. The overall prevalence/diagnosis of obesity was 41.9% for NHANES (nationwide survey), 44.5% for NSQIP (in-hospital measurements), and 15.4% for NIS (ICD codes retrieved from medical charts).

The prevalence of obesity in NIS (based on recorded obesity in medical charts) is significantly lower than NHANES and NSQIP (based on measured BMI) suggesting underrecognition or underdocumentation of obesity in the inpatient setting.