![[ Visit Client Website ]](images/banner.jpg)
BACKGROUND: Obesity is an extremely prevalent epidemic, associated with many adverse outcomes. With numerous clinical- and community-based obesity reduction efforts, effective monitoring tools to assess change over time are highly desirable. Self-assessed reporting of height and weight (i.e., Behavioral Risk Factor Surveillance System [BRFSS]) are typical for surveys. An alternative and novel source is electronic health records (EHR), where routine height and weight measures are objectively collected. This study sought to assess agreement between two methods of adult BMI monitoring.
METHODS: Denver’s rates of adult (<18 years) obesity was measured in two data sources: 1) BRFSS surveys and 2) EHR data. Each data source has some respondent-associated geographic and demographic characteristics. 2011-2012 BRFSS telephone surveys (land and wireless users) were administered and individuals were asked their height and weight. Using 2009-2013 EHR data, the last height and weight data were included for individuals making clinic visits at 3 large healthcare providers in the region. BMI was calculated and biologically implausible values excluded. Rates of obesity (BMI >30) were calculated by data source, demographic group and geographically represented (e.g., neighborhood). US Census data provided denominator estimates and measures of social determinants.
RESULTS: Denver’s population of adults was 474,106. Inhabitants surveyed or for whom BMI measures were available was 0.4% by BRFSS and 33% by EHR. Penetration by census tract ranged from 2% to 63%. Estimates of obesity in Denver County varied from 393 of 1977 surveyed (20%) using BRFSS to 48,770 of 158,036 (31%) using EHR. Census tract maps of BMI rates showed variable levels of obesity (5%-44%). Sub analysis of data showed that obesity rates by BRFSS were 20% in men and 20% in women. Rates by gender using EHR were 28% and 32%, respectively. Blacks had rates of 34%, Hispanics 28%, and whites 16% using BRFSS. Those rates were 38%, 38% and 29%, respectively for blacks, Hispanics and whites, using EHR data.
CONCLUSIONS: EHR BMI data is more comprehensive and available for monitoring obesity trends. Compared with BRFSS, EHR data were more ample and showed higher obesity rates generally, and by gender and race. Objective BMI measures are presumably more accurate than self-assessment. BRFSS may underestimate prevalence; restricted sample size limits BRFSS small area analyses, compared with EHR. Selection biases (more obese people visit healthcare providers) may skew these EHR results. Rate discrepancies among complementary data sources need to be resolved for a consistent message.