The Risks of Rurality: Novel Trends in Non-Hispanic White Mortality in Pennsylvania
The adverse economic and mortality trends affecting middle-aged non-Hispanic whites have now been widely reported (Centers for Disease Control and Prevention, 2012). In 2015, Case and Deaton revealed this jarring reversal in the long-term mortality decline of middle-aged non-Hispanic whites (henceforth whites) across the United States between 1999 and 2013 (Case & Deaton, 2015). Absent from their analysis was a comparison between rural and urban white populations, which exhibit significant differences in cause-specific mortality (Snyder, 2016). This rural-urban gradient extends notably to suicide mortality rates that rise with increasing levels of rurality (Singh & Siahpush, 2002). In this report, I provide evidence for Case and Deaton’s conclusions on the reverse transition occurring in white populations at the micro-level in the state of Pennsylvania. I then delve into the finer geographical and racial mortality gradients at infancy, childhood ages 1-9, and midlife ages 45-54 to show profound differences in mortality and courses of death between rural and urban Pennsylvania communities.
Data & Methods
Mortality Data. Data was gathered on all-cause and cause-specific mortality from the Center for Disease Control and Prevention (CDC) Wonder online database. The 2013 CDC Urbanization scheme was used to classify rural and urban areas of Pennsylvania (Centers for Disease Control and Prevention, 2014). In this report, Large Central Metro and Large Fringe Metro were considered urban areas while Small Metro, Micropolitan (non-metro), and NonCore (non-metro) were designated as rural areas. These 2013 CDC classifications were based on the 2010 Office of Management and Budget (OMB) standards for defining metropolitan statistical areas (MSAs) and on data from the 2010 US census. The urban categorization used in this report included the metropolitan categorizations with one million or more people and the rural category included MSAs with populations less than 250,000. Counties in the Medium Metro category were not included in rural-urban comparisons. In certain cases, mortality data was unavailable or insufficient. The Medium Metro category, which includes counties with populations of 250,000 to 999,999, was included in these instances and are denoted as suburban. Mortality trends of Asians and Hispanics are not consistently presented in the figures of this report because of the unavailability or unreliability of data on these groups.
Course of death classifications were based on the International Statistical Classification of Diseases and Related Health Problems 10th Revision (ICD-10) codes. The top 15 leading causes of death identified in this report for infants, children, and midlife white Pennsylvania residents were based on designations defined by the National Center for Health Statistics (NCHS) (Centers for Disease Control and Prevention, 2015). The causes of infant death investigated were sudden infant death syndrome (R95) and congenital defects that encompassed congenital malformations, deformations, and chromosomal abnormalities (Q00-Q99). In children, causes of death included cancer that is termed by the CDC as malignant neoplasms (C00-C97); congenital malformations, deformations, and chromosomal abnormalities (Q00-Q99); influenza and pneumonia (J09-J18); motor vehicle traffic accidents as defined by the NCHS; and assault (homicide) (U01-U02, X85-Y09, Y87.1). The cause-specific mortality trends in adults ages 45-54 were traced for external causes of morbidity and mortality (V01-Y89); poisonings as defined by the NCHS; lung cancer that encompasses the malignant neoplasms of the trachea, bronchus, and lung (C33-34); intentional self-harm (suicide) (U03, X60-X84, Y87.0); diabetes mellitus (E10-E14); and liver diseases that are broadly termed chronic liver disease and cirrhosis (K70, K73-K74).
In one instance, mortality data outside of Pennsylvania was included. Because the state’s vital data was insufficient, the age-standardized death rates by cause for white children ages 1-9 in the United States were calculated.
Life Expectancy Data. Life tables were constructed using data from the CDC Wonder online database for Pennsylvania residents in 2015. The average years lived in the age interval (nax) by persons who die in this interval was assumed to be half the age interval for those ages 5 and above. For those below age 5 the values of nax for males as suggested by Preston et al. (2000) were used (Preston, Heuveline, & Guillot, 2000). That is, if 1m0 ≥ 0.107, then 0.330 was used for 1a0 and if 1m0 < 0.107, then the value of 0.045 + 2.684 * 1m0 was used for 1a0. Likewise, if 1m0 ≥ 0.107, then 1.352 was used for 4a1 and if 1m0 < 0.107, the value of 1.651 - 2.816 * 1m0 was used for 1a0.
Methods. Rates of mortality are shown as deaths per 100,000. The age pattern of mortality by race and urbanization was a graph of the natural logarithm of age-specific mortality of 5-year age intervals across the lifespan plotted against age. Life expectancies were calculated using conventional life table methods and by using the values of nax as described previously. Infant mortality in Pennsylvania by race, urbanization, and cause from 1999 to 2015 were calculated using crude death rates.
Death rates by cause for white children ages 1-9 were age-standardized between rural and urban populations by taking the mean of the proportionate age distributions (nCx) across the two urbanization categories by age intervals. Age-specific mortality rates were multiplied by this composite value of nCx and graphed. All-cause mortality by race and urbanization was age-standardized in the same way for Pennsylvania residents ages 45-54 by taking the mean of the nCxurban and nCxrural within each race. The age standardization procedure was not done among races. Non-age-standardized depictions of all-cause mortality by sex, urbanization, and race were also created for midlife Pennsylvania residents using age-specific crude death rates. The same was done for cause-specific mortality trends among white Pennsylvania residents from 1999 to 2015 among different age groups.
In one figure, differences in age-specific mortality rates by year are depicted. Urban age specific crude death rates in the age interval 45 to 54 were subtracted from the rural age-specific crude death rates in each given year for each cause of mortality (i.e. for each year, nMxrural - nMxurban).
Figure 1 shows the age pattern of mortality by urbanization for whites, non-Hispanic blacks (henceforth blacks), and Hispanics. In 2015, rural whites had age-specific mortality rates that were near consistently higher than that of their urban counterparts. For the most part, blacks and Hispanics had age-specific mortality rates that were higher and lower, respectively, than whites in either urbanization category. Rural Hispanics had the lowest mortality rates of any group across ages 45 to 80. Child mortality ages 1 to 10 was considerably higher for rural whites than urban whites.
Across the lifespan, life expectancy for rural whites residing in Pennsylvania is consistently lower than their urban white counterparts (see Figure 2). Based on 2015 mortality rates, the life expectancy of rural whites at birth is 76.25 years compared to urban whites who can expect to live nearly 3 years longer at 79.15 years. While this difference is large, urban and rural blacks can expect to live only 71.43 years, nearly 5 years less than rural whites. Racial gradients in mortality, though important, is beyond the scope of this report and serve only as comparison. Between rural and urban whites, there is also an appreciable divergence in later-life mortality at age 85+ (see Appendix tables (i) and (ii)). Differences in life expectancy were investigated further through an analysis of cause-specific mortality at infancy, childhood ages 1-9, and midlife ages 45-54.
Table 1 presents the top fifteen leading causes of infant mortality by urbanization for white Pennsylvanians from 1999 to 2015. Among the top five causes, three common causes of infant mortality are at higher levels in rural areas. These causes of mortality include congenital deformities, sudden infant death syndrome, and newborns affected by complications of placenta, cord, and membranes. As Figure 3 shows, infant mortality for sudden infant death syndrome and congenital defects have been at higher levels in rural counties since 1999 although the trends do suggest improvement. Noticeably, urban mortality rates from these conditions are mostly stagnating with an increase in congenital defects from 2001 to 2009.
All-cause infant mortality since 1999 has also demonstrated a divergence between rural and urban mortality rates among Pennsylvanian whites. Figure 4 presents a striking trend since 2010 in which rural white infants have higher mortality rates than their urban counterparts. Such consistency is not apparent in any other race except for black infants between 2003 and 2013.
Differences in child mortality ages 1-9 between urban and rural counties of Pennsylvania were not substantial except for death by accidents, especially motor vehicle traffic accidents. Table 2 depicts accident mortality to be nearly three times as high for rural white children than their urban counterparts. However, Figure 5 shows that mortality by motor vehicle traffic has declined across the United States since 1999. Mortality rates of other courses of death including cancer, congenital issues, assault/homicide, and influenza/pneumonia have remained relatively stable over this time frame. Notably, influenza and pneumonia rates have also consistently been higher among rural white children. Pennsylvania data for these rates were not available.
Table 4 shows the top fifteen causes of mortality in midlife whites ages 45-54 between rural and urban whites between 1999 and 2015. Both groups share similar causes of death including cancer (neoplasms), heart disease, accidents, suicide, diabetes, and liver diseases. Based on this data alone, the similarity in crude death rates for each course of mortality also seems to suggest striking similarities rather than differences between rural and urban whites. However, Figure 6 and Figure 7 vividly shows that, in comparison with other racial groups across urbanization, midlife mortality for whites in urban and rural areas for both men and women are stagnating while the mortality rates for other racial groups are declining quite substantially. This holds true even when the rates between urbanization classifications are age standardized (see Figure 7). According to the age-standardized mortality trends seen in Figure 7, there has also been a slight rise in rural white midlife mortality since 2010 while a noticeable decrease in urban white midlife mortality has occurred across the same period. Both differences between races and between urbanization categories were explored.
Compared to blacks and Hispanics from 1999 to 2015, Pennsylvanian whites in midlife ages 45-54 show substantial increases in all-cause mortality, death by external causes, and diseases of the liver (see Table 5). Blacks and Hispanics show negative trends for these causes of mortality except for death by external causes. However, the increase in death by external causes for whites (46.7 per 100,000) across this sixteen year period is nearly double that of blacks (17.2 per 100,000. Figure 8 also shows a considerable increase in the suicide rate of whites in this age interval compared to that of blacks.
Comparing by urbanization among Pennsylvanian whites, the age-standardized crude death for urban whites in 2015 was 706 per 100,000 people while that of rural whites was 764 per 100,000 people (see Table 6). Figure 9 presents the difference between rural and urban age-specific mortality rates by year for five leading causes of death in midlife white Pennsylvanians ages 45-54. In contrast to their urban counterparts, rural midlife whites have experienced increased mortality from cancers generally and lung cancer specifically. This analysis also shows that mortality by external causes are more prevalent in urban than rural areas since most recently in 2010. From 1999 onwards, death from liver diseases has also disproportionately affected midlife whites in urban areas. Death by suicide across urbanization was found to be inconsistent.
Delving further into the specific causes of death, the increase in mortality from external causes outpaced any other cause of mortality for midlife whites ages 45-54 in Pennsylvania (see Figure 10). Poisonings have also increased considerably since 2010. Mortality from diabetes and liver disease have only slowly increased since 1999. As expected, death from lung cancer has decreased gradually. Figure 11 shows a drastic increase in mortality from external causes since 1999 in Pennsylvanian whites ages 30-64. A closer study of external causes by five-year age intervals reveals that older whites (e.g. those aged 45-64) have higher rates of mortality from external causes than their younger counterparts, except in 2015 where there was sudden rise in mortality for those ages 30-34.
The seminal paper by Case and Deaton (2015) on the reverse transition occurring in whites focused on a slim demographic slice—midlife whites ages 45-54. This report sought to confirm their conclusions on midlife mortality at the micro-level but also investigate mortality trends by urbanization and extend the analysis to other age groups, particularly infants and children ages 1-9.
Mortality trends of whites in Pennsylvania mostly confirm the findings of Case and Deaton (2015). Mortality from all external causes, poisonings, suicide, and liver diseases show surprising increases when compared to that of blacks and Hispanics. These trends coupled with rising suicide rates among white Pennsylvanians could suggest diminished economic advancements and sociopolitical empowerment (Masters, Tilstra, & Simon, 2017). Further study by socioeconomic class and education on white residents of Pennsylvania will be needed to identify precise ties between economic conditions and health outcomes.
Across all major mortality measures, rural whites have higher mortality rates than their urban counterparts across their lifespan. Perhaps most strikingly, newborn rural whites can expect to live nearly three full years less than whites living in urban areas. Particularly for white infants, despite significant improvements, those residing in rural areas are disproportionately inflicted by sudden infant death syndrome and congenital defects, suggesting inequitable access to healthcare services (Bailey & Timpe, 2016). Large mortality improvements do, however, suggest that advances in medical care are improving infant health outcomes. Since 2010, rural white infant mortality has diverged considerably from that of their urban counterparts, paralleling trends in midlife ages 45-54. Midlife whites in urban areas have better health outcomes than in rural areas, especially evident beginning in 2010. The global financial crisis of 2008 may have taken its toll on mortality levels of white residents of Pennsylvania beginning in 2010. To my knowledge, the mortality divergence occurring in 2010 for either infants or midlife whites by urbanization in Pennsylvania has not been reported in the literature. The cause for this parallel requires further study and could be due to the growing feelings of disempowerment and neglect (Inglehart & Norris, 2016). However, the impact of economic hardship on mental health and health more broadly has been widely reported, especially for midlife whites (Bailey & Timpe, 2016; Phillips, 2016; Young, 2013).
Though decreasing, mortality from vehicle traffic accidents in white children ages 1-9 in rural areas is significantly greater than in urban areas. This is likely due to the amount of time spent on the road in rural areas. Across the nation, rural white children in this age group also have higher rates of mortality by influenza and pneumonia, suggesting barriers to healthcare access.
White midlife mortality by urbanization in Pennsylvania reveals striking differences. Contrary to popular sentiment on the discontent and subsequent mortality regression occurring in the American rural white working class, I found mortality from external causes to affect predominantly whites living in urban areas (Inglehart & Norris, 2016). Mortality from chronic diseases such as cancer is driving the difference between rural and urban whites in Pennsylvania. While this report did not account for socioeconomic status (SES) or education, the trends I uncovered do challenge the “economic insecurity perspective” posed by Case and Deaton (2015) and provides support for the “cultural backlash thesis” proposed by Inglehart and Norris (2016). Specifically, the presumption that rural whites who are disproportionately less educated and of lower SES behave according to diminished economic prospects should be called into question (see Gimpel & Karnes, 2006; Young, 2013).
Future studies on whites should overlay cause-specific mortality with information not only on urbanization and SES, but also on cultural value metrics such as political or religious affiliation. Case and Deaton (2015) report on the substantial increases in depreciating mental health, rising alcohol use, and increased pain among whites. As my report suggests, these reversals could be due to factors besides diminished economic prospects such as discontent with a shrinking white population and subsequent shifts in cultural norms and values across the United States.
Download the pdf of this research article to view the Appendix.
Bailey, M. J., & Timpe, B. (2016). CHILD HEALTH. Hope for America's next generation. Science, 352(6286), 661–662. http://doi.org/10.1126/science.aaf7270
Centers for Disease Control and Prevention. (2012). CDC grand rounds: prescription drug overdoses-a US epidemic. MMWR. Morbidity and Mortality Weekly Report, 61(1), 10.
Centers for Disease Control and Prevention. (2014). 2013 NCHS Urban–Rural Classification Scheme for Counties.
Gimpel, J. G., & Karnes, K. A. (2006). The Rural Side of the Urban-Rural Gap. PS: Political Science & Politics, 39(3), 467–472.
Inglehart, R., & Norris, P. (2016). Trump, Brexit, and the Rise of Populism: Economic Have-Nots and Cultural Backlash, 1–53. Retrieved from https://ssrn.com/abstract=2818659
Masters, R. K., Tilstra, A. M., & Simon, D. H. (2017). Mortality from Suicide, Chronic Liver Disease, and Drug Poisonings among Middle-Aged U.S. White Men and Women, 1980–2013. Biodemography and Social Biology, 63(1), 31–37. http://doi.org/10.1080/19485565.2016.1248892
Phillips, J. A. (2016). Testing a Sociological Explanation for Rising Rates of White Mortality. Sociological Forum, 31(3), 716–719. http://doi.org/10.1111/socf.12280
Preston, S., Heuveline, P., & Guillot, M. (2000). Demography: measuring and modeling population processes. Wiley-Blackwell.
Singh, G. K., & Siahpush, M. (2002). Increasing Rural–Urban Gradients in US Suicide Mortality, 1970–1997. American Journal of Public Health, 92(7), 1161–1167. http://doi.org/10.2105/AJPH.92.7.1161
Young, F. W. (2013). “What's the Matter With Kansas?” A Sociological Answer. Sociological Forum, 28(4), 864–872. http://doi.org/10.1111/socf.12060
Centers for Disease Control and Prevention (2015) National Center for Health Statistics: Underlying cause of death 1999-2015 on CDC Wonder online database, released. Available at wonder.cdc.gov/ucd-icd10.html. Accessed April 10, 2015.