Mortality by Socioeconomic Category in the United States
Author
Magali Barbieri, Ph.D., University of California-Berkeley
Table of Contents
Summary
Report
Podcast
Excel Data Files
Data Visualizations
Acknowledgements
Summary
This SOA-sponsored research report presents mortality analysis and rate estimates for the United States by year from 1982 through 2019, separately by socioeconomic quintile and decile. Details on the development of the estimates are summarized in the report. The life tables by socioeconomic category are available for download, and some of the results are presented in online graphs and maps, and a data summary report.
The results were produced by combining data from three sources:
- county-level mortality data from processed National Center for Health Statistics data
- socioeconomic data extracted from the United States Decennial Census
- socioeconomic data extracted from the American Community Survey
Separately for each year of data, a Socioeconomic Index Score was computed for each county. The Socioeconomic Index Scores, in turn, were used to group counties into deciles, with each decile holding 10% of the total U.S. population. Mortality rates were then estimated for each decile. In addition, a parallel analysis was performed using quintiles rather than deciles.
This web page, the associated report and the online data visualizations were initially released by the SOA in November 2020, using mortality data from the CDC-Wonder database. To address constraints in the CDC-Wonder data, the socioeconomic mortality analysis was repeated in December 2020, using a restricted dataset from the National Centers for Health Statistics with data back to 1982 and covering ages up to 110+.
This report updates the December 2020 version and relies on a different way to allocate all U.S. counties into deciles/quintiles. Changes to the methodology from the December 2020 report are:
- One more year of mortality data (2019) was incorporated in the analysis.
- The variables used to determine the Socioeconomic Index Scores were modified as follows:
- Instead of the percentage of the population aged 25 and over with at least a high school education, the percentage of the population aged 25 and over with at least 4 years of college education was used to account for the rise in education over the past forty years.
- Instead of the raw median household income, the median household income in each county was adjusted by the median housing cost at the state level to account for variations in standards of living across the country.
- Instead of recalculating the county Socioeconomic Index Scores for each year when data are available as in the previous report, the score is fixed to the year 2000, keeping the grouping of counties into the socioeconomic deciles/quintiles the same over the whole study period (1982-2019).
Rather than presenting the new results alongside the December 2020 results (both the report, Excel data files and the data visualizations) were removed from the website, and replaced with an updated report and visualizations that reflect the new data.
Report
Mortality by Socioeconomic Category in the United States
Videos
Insights on Mortality by Socioeconomic Category
Mortality and Socioeconomics
Podcast
Excel Data Files
County-Level Socioeconomic Data
Mortality Rate Estimates by Age, Sex and Socioeconomic Index Quintile
Mortality Rate Estimates by Age, Sex and Socioeconomic Index Decile
Data Visualizations
These interactive dashboards are visualizations of key metrics found in the report. They provide filtering, drill-down, and other interactive capabilities that allow you to focus on specific subsets of the data.
|
Map of County-Level Socioeconomic Index Quintiles This map shows the socioeconomic index quintile of each county across time. Quintile "1" captures the bottom 20% of the population, while quintile "5" captures the upper 20%. The rankings for years 1990 and 2000 were determined using data from the Decennial Census, while the rankings from 2007 through 2016 were determined using data from the American Community Survey (ACS). Note that 5-year samples were used from the ACS. Thus, what is labeled as "2007" on the map is based on an ACS dataset that runs from 2005 through 2009. Similarly, "2008" is based on an ACS dataset that runs from 2006 through 2010. In other words, the data label is the mid-point of each 5-year ACS sample. Use the year control on the right side of the exhibit to advance across time. The upper portion of the year control enables a user to manually step forwards or backwards through time, while the lower portion of the control activates a loop which automatically cycles forwards or backwards through time, rather like playing a video. Map of County-Level Socioeconomic Index Deciles This map shows the socioeconomic index decile of each county across time. Decile "1" captures the bottom 10% of the population, while decile "10" captures the upper 10%. The rankings for years 1990 and 2000 were determined using data from the Decennial Census, while the rankings from 2007 through 2016 were determined using data from the American Community Survey (ACS). Note that 5-year samples were used from the ACS. Thus, what is labeled as "2007" on the map is based on an ACS dataset that runs from 2005 through 2009. Similarly, "2008" is based on an ACS dataset that runs from 2006 through 2010. In other words, the data label is the mid-point of each 5-year ACS sample. Use the year control on the right side of the exhibit to advance across time. The upper portion of the year control enables a user to manually step forwards or backwards through time, while the lower portion of the control activates a loop which automatically cycles forwards or backwards through time, rather like playing a video. |
||||||
|
Expected Lifetimes by Year, Sex, Age and Socioeconomic Index Quintile Using this graph's controls, you may select which sex, age group(s), socioeconomic index quintile(s) and years you wish to display. Note that the year labels on the graph's horizontal axis are determined automatically by Tableau, and are typically intermediate time points rather than end points. For example, if you select the year range from 1999 to 2018 to plot, the axis labels might be 2003, 2009 and 2015. While 1999 and 2018 are not shown as axis labels, the full user-specified time period is indeed plotted on the graph. Expected Lifetimes by Year, Sex, Age and Socioeconomic Index Decile Using this graph's controls, you may select which sex, age group(s), socioeconomic index decile(s) and years you wish to display. Note that the year labels on the graph's horizontal axis are determined automatically by Tableau, and are typically intermediate time points rather than end points. For example, if you select the year range from 1999 to 2018 to plot, the axis labels might be 2003, 2009 and 2015. While 1999 and 2018 are not shown as axis labels, the full user-specified time period is indeed plotted on the graph. |
||||||
|
Mortality Rate Estimates by Year, Sex, Age Group and Socioeconomic Index Quintile This exhibit contains 2 graphs, each displaying mortality rates by socioeconomic index quintile. Each rate is the probability that an individual who is alive at the beginning of the age interval will die before the end of that age interval. For example, the rate for ages 10 through 19 is the probability that an individual who is alive at age 10 will die before reaching the age of 20. Age zero is placed in a group by itself. Graph 1 shows mortality rates on a standard scale, while Graph 2 shows the same rates on a logarithmic scale. On a logarithmic scale, the values 0.1%, 1%, 10% and 100% are equally spaced. You may select which sex, age group(s), socioeconomic index quintile(s) and years you wish to display. Note that the year labels on the graph's horizontal axis are determined automatically by Tableau, and are typically intermediate time points rather than end points. For example, if you select the year range from 1999 to 2018 to plot, the axis labels might be 2003, 2009 and 2015. While 1999 and 2018 are not shown as axis labels, the full user-specified time period is indeed plotted on the graph. Mortality Rate Estimates by Year, Sex, Age Group and Socioeconomic Index Decile This exhibit contains 2 graphs, each displaying mortality rates by socioeconomic index decile. Each rate is the probability that an individual who is alive at the beginning of the age interval will die before the end of that age interval. For example, the rate for ages 10 through 19 is the probability that an individual who is alive at age 10 will die before reaching the age of 20. Age zero is placed in a group by itself. Graph 1 shows mortality rates on a standard scale, while Graph 2 shows the same rates on a logarithmic scale. On a logarithmic scale, the values 0.1%, 1%, 10% and 100% are equally spaced. You may select which sex, age group(s), socioeconomic index decile(s) and years you wish to display. Note that the year labels on the graph's horizontal axis are determined automatically by Tableau, and are typically intermediate time points rather than end points. For example, if you select the year range from 1999 to 2018 to plot, the axis labels might be 2003, 2009 and 2015. While 1999 and 2018 are not shown as axis labels, the full user-specified time period is indeed plotted on the graph. |
||||||
|
Mortality Improvement Rate Estimates by Year, Sex, Age Group and Socioeconomic Index Quintile This exhibit contains 3 graphs, comparing mortality rates at the beginning and end of a user-specified time interval. Use the "First Year" and "Last Year" parameters on the right side of the screen to specify this interval. Additional parameters are available to determine which sex and quintiles to graph. Each mortality rate is defined as the probability that an individual of age "x" dies before reaching age "x+10". For example, for age group 60-69, the graph shows the probability that an individual who is exactly 60 years old will die before reaching the age of 70. Graph 1 shows annualized rates of mortality improvement across the user-specified period. Consider the following example: the mortalilty rate in 2000 is 10%, and the mortality rate in 2010 is 8%. In this case, the annualized rate of mortality improvement would be computed as follows: 1 - (8% / 10%) ^ (1 / 10) = 2.2%. Graph 2 shows the mortality rate in the final year of the user-specified period, divided by the mortality rate in the initial year. A value of 100% indicates that the initial and final mortality rates are identical. A value greater than 100% indicates that mortality has increased, while a value less than 100% increases that mortality has decreased. Graph 3 compares mortality rates in the first and final years. Mortality Improvement Rate Estimates by Year, Sex, Age Group and Socioeconomic Index Decile This exhibit contains 3 graphs, comparing mortality rates at the beginning and end of a user-specified time interval. Use the "First Year" and "Last Year" parameters on the right side of the screen to specify this interval. Additional parameters are available to determine which sex and quintiles to graph. Each mortality rate is defined as the probability that an individual of age "x" dies before reaching age "x+10". For example, for age group 60-69, the graph shows the probability that an individual who is exactly 60 years old will die before reaching the age of 70. Graph 1 shows annualized rates of mortality improvement across the user-specified period. Consider the following example: the mortalilty rate in 2000 is 10%, and the mortality rate in 2010 is 8%. In this case, the annualized rate of mortality improvement would be computed as follows: 1 - (8% / 10%) ^ (1 / 10) = 2.2%. Graph 2 shows the mortality rate in the final year of the user-specified period, divided by the mortality rate in the initial year. A value of 100% indicates that the initial and final mortality rates are identical. A value greater than 100% indicates that mortality has increased, while a value less than 100% increases that mortality has decreased. Graph 3 compares mortality rates in the first and final years. |
||||||
|
Survival Probabilities by Year, Sex and Socioeconomic Index Quintile This exhibit contains 3 graphs, each of which shows the probability of surviving from an initial age to a subsequent age "X". In graphs 1, 2 and 3, the initial age is 60, 20 and 0, respectively. You may select which year(s), sex (or sexes) and socioeconomic index quintile(s) you wish to display. Survival Probabilities by Year, Sex and Socioeconomic Index Decile This exhibit contains 3 graphs, each of which shows the probability of surviving from an initial age to a subsequent age "X". In graphs 1, 2 and 3, the initial age is 60, 20 and 0, respectively. You may select which year(s), sex (or sexes) and socioeconomic index decile(s) you wish to display. |
||||||
|
National-Level Mortality Rate Estimates: Comparison with SSA Quintiles and deciles do not play a role in this exhibit. Rather, the entire dataset was used to estimate national-level mortality rates. These rates are labeled "NCHS-CDC" in the graph below, and are compared to rates developed by the Social Security Adminstration (SSA). Note that the SSA rates are smoothed, while the NCHS-CDC rates are unsmoothed. Also, the NCHS-CDC analysis based death rates on NCHS deaths and Census populations for all ages, while SSA bases death rates on NCHS deaths and Census populations for under 65, and also incorporates Medicare deaths and enrollments for ages 65 and older. Each mortality rate is the probability that an individual who is alive at age "x" dies before reaching age "x+1". Graph 1 shows mortality rates on a standard scale, while Graph 2 shows the same rates on a logarithmic scale. On a logarithmic scale, the values 0.1%, 1%, 10% and 100% are equally spaced. You may use the controls on the right-hand side of the graph to select which sex (or sexes), and ages to display. Use the year control to advance across time. The upper portion of the year control enables a user to manually step forwards or backwards through time, while the lower portion of the control activates a loop which automatically cycles forwards or backwards through time, rather like playing a video. |
Acknowledgements
Thank you to the following individuals who served on the Project Oversight Group:
Philip Adams, FSA, CERA, MAAA
Mary Bahna-Nolan, FSA, CERA, MAAA
Mark Bye, ASA
Jean-Marc Fix, FSA, MAAA
Sam Gutterman, FSA, CERA, MAAA, FCAS, FCA, HONFIA
Edward Hui, FSA
Al Klein, FSA, MAAA
Larry Pinzur, PhD
Marianne Purushotham, FSA, MAAA
Manny Santos, FSA, FCIA
Joel Sklar, ASA, MAAA
Larry Stern, FSA, MAAA
Dale Hall, FSA, CERA, MAAA, SOA Managing Director of Research
Jan Schuh, SOA Sr. Research Administrator
Ronora Stryker, ASA, MAAA, SOA Sr. Practice Research Actuary
Patrick Wiese, ASA, SOA Modeling Actuary
Questions or Comments
Give us your feedback! Take a short survey on this report. Take Survey
If you have comments or questions, please email research@soa.org.