A Tale of Three Cities: Racial and Ethnic Disparities in Premature Mortality in the District of Columbia, 2005

October 22, 2008
Annette B. Ramirez de Arellano, Dr.PH., Sidney Wolfe, M.D., Kate Resnevic, Alejandro Necochea, M.D., M.P.H.

Full Report (pdf)
Introduction and Methods
Overall Years of Potential Life Lost
Premature Mortality by Sex
Premature Mortality by Race/Ethnicity - Part 1
Premature Mortality by Race/Ethnicity - Part 2
Sex Differentials within Subgroups - Part 1
Sex Differentials within Subgroups - Part 2
Discussion and Implications for Policy


When mortality data reveal racial or ethnic disparities, they are particularly poignant because how and when we die reflects the endpoint of a longer process, namely how we live. Death has thus been called “the ultimate sentinel event” because it can alert decision-makers that something is wrong with the body politic. In the health field, mortality data can express specific vulnerabilities; reflect socio-economic, environmental and other risk factors; and indicate service gaps. Collectively, mortality rates have been called “the quantification of a population’s collective tragedy.”[1]

Many nations, states, counties and cities use crude mortality rates to rank their major health issues, often allocating resources accordingly. But there are other indicators that better capture the relative priority of health problems, one of these being years of potential life lost (YPLL), which measures premature mortality. YPLL reflects the number of useful years of life that are not available to a population due to early death.[2]

This indicator has a long history[3] and has been subject to some variations. It has been used to examine premature mortality in different countries (including the United States,[4] Italy,[5] Ireland,[6] Israel[7] and Canada,[8] among others) and states (e.g., North Carolina[9]). It has also been used in assessing the impact of deaths associated with specific causes or risk factors (e.g., smoking,[10] automobile accidents[11] and the consumption of alcohol[12]). This article focuses on the District of Columbia, which is of particular interest because of its political importance and its ethnic and racial diversity.


The indicator YPLL is defined here as the number of years of life lost by persons before reaching age 70. The calculation of YPLL for a particular cause involves subtracting each deceased person’s age from 70. These differences – the “years lost” – are added for all deceased people in that category. These data are then grouped or broken down by the demographic variables of interest (e.g., sex, race/ethnicity and jurisdiction).[13]

Some researchers exclude infant deaths in calculating YPLL. For example, Romeder and McWhinnie argue that such deaths are most often due to causes specific to this early period of life and often have a different etiology than later deaths. Moreover,

each infant death would account for almost 70 years, giving a weight double that of a death between ages 30 and 40. This appears to be an overestimation of the value accepted by society for such a loss in the light that a very early death is often replaced by another birth.[14]

The U.S. Centers for Disease Control and Prevention (CDC), however, includes these early deaths in their computations of YPLL. This is in keeping with their emphasis on prevention, which includes averting infant deaths. As causes of early deaths such as sudden infant death syndrome and automobile and other injuries have become more preventable, it is important that these deaths be duly included in indicators used in epidemiological surveillance. We have therefore included these early deaths in our analysis.

Another methodological issue concerns the appropriate cut-off point for defining premature death. Here, we have followed Romeder and McWhinnie, who use the cut-off age of 70. They argue that using an older age (e.g., 75 or 85, which are sometimes used) includes deaths in which the underlying cause is difficult to determine. At the same time, a younger cut-off age (e.g., 65) excludes a significant proportion of the population that is still productive.[15] Because the cut-off age can vary and is basically arbitrary, current CDC databases provide YPLL using different cut-off points ranging between 65 and 85.[16] Our choice of 70 is therefore well within the parameters established in the literature. Deaths before this age are therefore considered premature for our purposes.

The CDC’s Web-based Injury Statistics Query and Reporting System (better known as WISQARS) provides data broken down by jurisdiction, year, cause and race/ethnicity. This database also allows users to choose from a menu of indicators, and we have used age-adjusted YPLL rates. Certain causes of death are more prevalent among particular age groups, and age adjustment allows us to compare rates without concern that differences in those rates are caused by variations in the age distributions of the populations being compared. We examined the District of Columbia data and data for the United States as a whole for the year 2005.  In focusing on D.C., we looked at both overall data as well as the YPLL for both sexes and three population subgroups: whites, blacks and Hispanics. Because Hispanics can be of any race and we wanted to avoid any overlap among groups, the categories of “white” and “black” consistently refer to non-Hispanic whites and non-Hispanic blacks, respectively. While there are other ethnic groups living in D.C., these are too small to yield meaningful data and reliable rates.

[1] Paul H. Wise, Confronting Racial Disparities in Infant Mortality: Reconciling Science and Politics. American Journal of Preventive Medicine Supplement to vol. 9 (6) November/December 1993: 7.

[2] United Health Foundation, Health Rankings --- 2007 Edition. A Call to Action for People and Their Communities: 20.

[3] William Petty (1623-1687), a pioneer in vital statistics and economics, conceived of such a measure in his Political Arithmetic (1687). John Last, Ed. A Dictionary of Epidemiology (New York: Oxford University Press, 1983): 81.

[4] Centers for Disease Control, Years of Potential Life Lost before Age 65 – United States, 1990 and 1991. Morbidity and Mortality Weekly Report 42 (1993): 251-253.

[5] Massimo Arca et al. Years of Potential Life Lost (YPLL) before age 65 in Italy. American J of Public Health 78 (9). 1988: 1202-1205.

[6] E. O’Shea, Social gradients in years of potential life lost in Ireland. European J of Public Health 13 (4). December 2003: 327-33.

[7] ED Richter, Potential-Years-of-Life- Lost from Motor Vehicle Crashes in Israel: An Epidemiological Analysis. International Journal of Epidemiology 1979; 8: 383-388.

[8] JM Romeder and JR McWhinnie, Potential Years of Life Lost Between the Ages of 1 and 70: An Indicator of Premature Mortality for Health Planning. International Journal of Epidemiology 6 (1977): 143-151.

[9] Daniel Rosenberg, SCHS Studies 130 (January 2002). www.schs.state.nc.us/SCHS/pubs.

[10] Annual Smoking-Attributable Mortality, Years of Potential Life Lost, and Productivity Losses – United States, 1997-2001. Morbidity and Mortality Weekly Reports. June 30, 2005. http://ncadi.samhsa.gov/govpubs/mmwr/vol54/mm5424aI.aspx

[11] Richter, op. cit.

[12] Alcohol-Attributable Deaths and Years of Potential Life Lost – United States, 2001. Morbidity and Mortality Weekly Report 2004; 53: 866-870.

[13] While most analyses use five-year age groups and the midpoint for each group to compute YPLL, the CDC uses individual deaths, subtracting the actual age at death from whatever cut-off point is used. See http://www.cdc.gov/ncipc/wisqars/fatal/help/helpfile.htm

[14] Romeder and McWhinnie, op. cit.: 148.

[15] Romeder and McWhinnie, op. cit.: 147.

[16] Centers for Disease and Prevention, National Center for Injury Prevention and Control. WISQARS YPLL Reports, 1999-2005.