Buy this Article for $10.95

Have a coupon or promotional code? Enter it here:

When you buy this you'll get access to the ePub version, a downloadable PDF, and the ability to print the full article.

Authors

  1. Monaghan, Alyssa MPH, CPH
  2. Jones, Lynda MPH
  3. Brink, LuAnn PhD
  4. Hacker, Karen MD, MPH

Abstract

Objectives: To compare city and census tract-level diabetes and hypertension prevalence using 500 Cities Project modeled estimates from the Centers for Disease Control and Prevention (CDC) and insurance claims data.

 

Methods: Insurance claims by census tract were collected from 3 local health plans for the city of Pittsburgh, Pennsylvania, for 2015-2016; conditions were defined using International Classification of Diseases, Ninth Revision (ICD-9) and Tenth Revision (ICD-10) codes. Crude prevalence estimates with 95% confidence intervals were downloaded from the CDC 500 Cities Web site to obtain modeled estimates by census tract. Confidence intervals were calculated for claims and compared with modeled estimates; nonoverlapping intervals were considered significant. Pearson correlation coefficients were generated for census tract-level comparison.

 

Results: City-level model-based and claims estimates were 9% versus 10% for diabetes and 31% versus 21% for hypertension. At the census tract level, model-based and insurance claims estimates were more concordant for diabetes (r = 0.366) than for hypertension (r = 0.220). Modeled estimates were significantly higher than claims estimates for 89% of census tracts for hypertension and 35% for diabetes.

 

Conclusions: Modeled estimates from the 500 Cites Project were significantly higher than insurance claims estimates for hypertension but were more consistent for diabetes. Utilization of multiple data sources to understand local-level chronic disease burden requires consideration of the strengths and limitations of each.