Death Before Old Age: Untangling Premature Mortality in the U.S.
, by DCEG Staff
by Cora A. Hersh
Decades of improvements in cancer prevention and treatment have resulted in declining cancer mortality rates—about 1.5 percent decline per year since 1999 (NCI Annual Report to the Nation, 2018). Despite this progress, overall mortality rates in U.S. adults have leveled off, or in some cases, even increased. These trends stand in stark contrast to those of nearly every other high-income nation, including Canada and the U.K., where premature mortality rates have steadily fallen. Driven by the pressing need to understand the causes of this disconcerting trend and to inform public health interventions, DCEG scientists are spearheading the Premature Mortality Project—an interdisciplinary, multi-institutional effort to characterize these trends. In the process, the team has uncovered distinct differences by race, ethnicity, age, and region, and provided crucial information about the ongoing, well-publicized, drug overdose epidemic.
Focus on Early Deaths
Amy Berrington de González, D.Phil., Chief of the Radiation Epidemiology Branch and one of the lead investigators in the group, was mentored by the late epidemiologist Sir Richard Doll, who famously remarked, “Death in old age is inevitable, but death before old age is not.” While many epidemiologic studies measure mortality rates overall, the team decided to focus specifically on premature mortality rates—conducting sophisticated descriptive analyses that focus on deaths occurring before age 64. Joining them in this effort are scientists hailing from other NCI divisions, the National Institute of Minority Health and Health Disparities, the National Institute on Drug Abuse (NIDA), and the University of New Mexico College of Nursing. Neal Freedman, Ph.D., M.P.H., senior investigator in the Metabolic Epidemiology Branch, said, “being in the National Institutes of Health (NIH) Intramural Research Program facilitates collaborations with other institutes. Our collaborators’ diverse backgrounds and expertise help us decide what data is most useful to pursue for public health applications, and give necessary context to our findings.” He added, “premature mortality is a snapshot of U.S. population health. As the ultimate goal of the NIH is to improve health, this project gives us the opportunity to work together toward that larger objective.”
Experts in Intricate Analyses
The Premature Mortality Project team uses publicly available data, such as from death certificates, but their unique combination of skills and expertise allows them to draw new insights. In DCEG, descriptive epidemiology and cancer surveillance have been a core piece of the research portfolio for decades. Throughout their groundbreaking efforts to track diseases like lung and breast cancer, Division staff have carefully honed the tools to describe disease and death trends in detail.
One key tool DCEG epidemiologists bring to the table is the age-period-cohort model. Although the method has existed for a long time, DCEG scientists have more recently played an important role in making it accessible for a variety of applications. Philip Rosenberg, Ph.D., senior investigator in the Biostatistics Branch (BB), explained, “this complex statistical technique allows us to distinguish the natural effects of aging—after all, older people die at a higher rate than younger people—from any potential increases in mortality from one generation to the next. In addition, statistical models provide a way to quantify the uncertainty involved in separating these two effects.”
Using national mortality data also allows the team to conduct work that is as “real-time” as it gets in epidemiology; they can respond relatively quickly to questions their own work uncovers, as well as to the national discussion of these issues. Although the most recently-available death certificate data tends to be two or three years old, this remains a faster turnaround than that of other complex projects, since the team does not have to collect participant surveys or clinical evaluations. An agile research program is especially critical when investigating a problem evolving as rapidly as premature mortality trends in the U.S.
Mortality Rate Overview
The project’s flagship paper, published at the beginning of 2017, was a comprehensive look at U.S. trends in premature mortality by sex, race/ethnicity, age, and cause of death (Shiels et al., 2017a). “The overall premature death rate is an average of increases and decreases in death rates across different parts of the population,” said lead author Meredith Shiels, Ph.D., investigator in the Infections and Immunoepidemiology Branch. “We need to identify which segments of the population have premature death rates that are worsening, staying the same, or improving, and which causes of death are responsible for these changes over time.”
This study, which used death certificate data from the National Centers for Health Statistics, revealed major differences in premature mortality trends. Death rates for Hispanics, blacks, and Asian/Pacific Islanders (A/PI) have declined, driven by reduced deaths from cancer, heart disease, and HIV/AIDS—reflecting public health efforts to reduce tobacco use, as well as advances in diagnosis and treatment. However, rates of premature mortality have risen among certain age groups of whites and American Indians/Alaska Natives (AI/AN). Most alarmingly, overall mortality rates for white and AI/ANs individuals aged 20-30 increased by two to five percent per year over the past decade, a rate virtually unheard of in high-income nations. Suicide and liver disease contributed to the increases in these groups, but the primary contributor was accidental deaths, mainly drug overdoses. This finding lent additional weight to previous, widely-publicized reports of a growing opioid epidemic in the U.S.
To put the U.S. trends in context, the investigators compared these rates to those in England, Wales, and Canada. They found that premature deaths in those latter countries are on the decline in nearly all age groups, further highlighting the uniqueness of the U.S. mortality increases.
Identifying Public Health Targets
Informed by the broad picture that the initial paper painted, the team decided to drill down on areas of particular public health concern. In December 2017, Drs. Shiels, Freedman, and Berrington de González collaborated with David Thomas, Ph.D., of NIDA, to publish an analysis of drug overdose deaths (Shiels et al., 2017b). They disaggregated overdose deaths by race/ethnicity (non-Hispanic white, non-Hispanic black, and Hispanic) and by type of drug. “Rising opioid deaths among non-Hispanic white Americans have received a great deal of national media attention. Recently, however, reports have indicated increases in drug overdose deaths for non-Hispanic black and Hispanic Americans as well. We wanted to better characterize these patterns to help inform targets for intervention,” said Dr. Berrington de González.
In fact, while the study confirmed strong increases over time for opioid-related overdose deaths in nearly all age groups of non-Hispanic whites, opioid deaths have also risen in both non-Hispanic black and Hispanic populations. Furthermore, the authors identified an epidemic of cocaine overdose deaths, primarily among non-Hispanic black men and women, that rivals deaths from opioids. Between 2012 and 2015, cocaine overdose deaths were almost as common among non-Hispanic black men as prescription opioid deaths among non-Hispanic white men, and slightly more common among non-Hispanic black women than deaths from heroin overdose among non-Hispanic white women. “Our findings stress the necessity of strategies to combat the U.S. prescription opioid, heroin, and fentanyl epidemics for all racial and ethnic groups,” said Dr. Shiels. “Additionally, this analysis suggests the need for additional public health efforts to prevent cocaine-related deaths, which disproportionally affect older, non-Hispanic black persons.”
Several more analyses are in the works to expand on issues highlighted by these recent studies. For example, one important line of investigation is using age-period-cohort modeling to produce projections of premature mortality in the U.S. through 2030. Ana Best, Ph.D., a postdoctoral fellow in BB who has been contributing to the Premature Mortality Project, has a particularly significant role in this effort. Another analysis compares rates of drug poisoning deaths by county, socioeconomic status, and rurality, to explore whether the prevailing narrative that the opioid epidemic principally affects lower-income, rural areas, stands up to statistical rigor. The team is also comparing drug poisoning deaths between the U.S. and other countries, as well as tracking trends in very premature (under 25 years old) death rates.
"Our findings stress the necessity of strategies to combat the U.S. prescription opioid, heroin, and fentanyl epidemics for all racial and ethnic groups."
Studies like the ones produced by the Premature Mortality Project have chronicled the epidemic of drug overdose, particularly opioids, in the U.S. These data have sparked a concerted effort to find solutions to this multifaceted public health emergency. For example, in 2017, the Department of Health and Human Services awarded over $800 million in grants to support opioid addiction treatment, prevention, and recovery. The President’s proposed 2019 budget adds another bolus of funding to increase the availability of overdose-reversing drugs. In addition to helping direct public health interventions to those who need it most, future descriptive studies will be essential to evaluate the impact of those interventions.
Cronin KA, Lake, AJ, Scott S, et al. Annual Report to the Nation on the Status of Cancer, part I: National cancer statistics. Cancer 2018 May 22; Epub ahead of print.
Shiels MS, Chernyavskiy P, Anderson WF, et al. Trends in premature mortality in the USA by sex, race, and ethnicity from 1999 to 2014: An analysis of death certificate data. The Lancet 2017.
Shiels MS, Freedman ND, Thomas D, et al. Trends in U.S. Drug Overdose Deaths in Non-Hispanic Black, Hispanic, and Non-Hispanic White Persons, 2000–2015. Ann Intern Med 2018; Epub 2017 Dec 5.