The Washington Post’s methodology

https://www.washingtonpost.com/national/the-washington-posts-methodology/2016/04/09/e4f95ffe-fdc9-11e5-886f-a037dba38301_story.html

Version 0 of 1.

The Washington Post analyzed the Multiple Cause of Death data collection from the Centers for Disease Control and Prevention.

The analysis included a record for every person who is known to have died in the country from 1990 through 2014. The data is compiled from death certificates and supplemented with additional reporting by the CDC. Mortality rates were created by combining the deaths with matching population estimates from the CDC Bridged-Race Population Estimates that has breakdowns by year for each age, gender, race and ethnicity. Rates are expressed as the number of deaths for every 100,000 people in a group. All calculations for whites and blacks are for non-Hispanic people. Calculations for Hispanics include white Hispanic, black Hispanic and other Hispanics.

The Post also used rates produced by the CDC’s online Wonder tabulation tool.

Death rates are calculated in age ranges because adults die at significantly greater rates as they age. The back side of the baby-boom population bubble is causing population in the adult age groups to be shrinking, and to be increasingly skewed toward the older end of the age ranges over time. To avoid distortions, The Post used five-year age groups for national comparisons.

For smaller geographies, The Post used 10-year age groups. For mapping small areas, The Post used 20-year age groups with age-adjusted mortality rates. The Post used single-year data except for small geographies and analysis of single causes of death. To avoid distorted calculations from one-year data variations, The Post combined 1999 through 2001 for a starting value and 2012 through 2014 as an endpoint for comparisons of change in maps and for particular causes of death.

To study levels of urbanization, The Post used the National Center for Health Statistics classification codes for each county. That scheme has six levels. The most urban levels are cities and suburbs in metropolitan areas with at least 1 million people. The less urban levels are for smaller metropolitan areas and for nonmetropolitan areas. Based on patterns in analysis, The Post grouped the data into two groups: the largest cities with their suburbs, and all other areas.