Author + information
- Jack V. Tu, MD, PhD∗ ()
- Institute for Clinical Evaluative Sciences, Sunnybrook Schulich Heart Centre, and the University of Toronto, Toronto, Ontario, Canada
- ↵∗Reprint requests and correspondence:
Dr. Jack V. Tu, Institute for Clinical Evaluative Sciences (ICES), G250-2075 Bayview Avenue, Toronto, Ontario M4N 3M5, Canada.
The burden of deaths from diseases such as ischemic heart diseases (IHD) and acute myocardial infarction (AMI) are assessed using a variety of outcomes measures. At the community level, age-adjusted death rates from IHD can be tracked as a cross-sectional measure of the burden of disease. At a clinical level, 30-day AMI case-fatality rates often are used as a preferred metric to assess the quality and outcomes of acute cardiac care. Although both of these measures are readily understandable to clinicians and policymakers, they have important limitations. Overall population death rates weight each death equally regardless of the age of occurrence, and tend to be driven by diseases of the elderly, where the vast majority of deaths occur (a death at age 90 years is given the same weight as a death at age 30 years). Improved AMI case-fatality rates from medical interventions are important, but do not consider the impact on life expectancy, which may be a more relevant measure to patients and their families. Some research has suggested that clinicians’ bedside estimates of life expectancy may explain the “treatment-risk” paradox, whereby patients who appear to have high short-term mortality risks are less likely to receive clinically indicated therapies (1).
Because the prevention of premature mortality is considered a worthwhile public health objective, alternative measures of disease burden have been developed. A commonly used summary measure of premature mortality in population epidemiology is years of potential life lost (YPLL) (2). The U.S. Centers for Disease Control and Prevention (CDC) has reported YPLL since 1982, and it is also used by many other public health organizations (3). Typically, YPLL can be calculated by subtracting the age at which an individual dies from the projected life expectancy, then summing these values across a population and dividing it by the population at risk (2,4). Deaths at the youngest ages are given the greatest weight. The age of 75 years is often used as a reference life expectancy for both men and women, with deaths before this age considered premature, although other age cutoffs, such as age 65 years, also have frequently been used. Alternative methods for calculating YPLL use life tables to calculate an age-specific life expectancy for deaths at different ages rather than a fixed life expectancy (4). Cardiovascular diseases, cancer, and accidents are the leading causes of YPLL in the United States when using YPLL before age 75 years as a measure of the burden of premature deaths (5).
Using data derived from the CDC, Figure 1 from the National Center for Health Statistics, shows U.S. temporal trends between 1980 and 2010 in YPLL before the age of 75 years from IHD (5). Striking declines in the YPLL in both sexes and by race are observed over time, with a substantial narrowing of differences between males and females. These trends highlight the gains that have been achieved in preventing premature IHD mortality in the United States and are likely a reflection of several factors including: 1) a decrease in the incidence of new ischemic events because of better risk factor detection and control; and 2) improved case-fatality rates because of the availability of newer treatments (6). By this measure of YPLL, males have a greater burden of premature mortality than females, and blacks have a greater burden than whites.
In this issue of the Journal, Bucholz et al. (7) use an alternative cohort-based method to estimate the YPLL that are due to AMI for men and women, and blacks and whites. The authors used data from the Cardiovascular Cooperative Project, a large nationwide prospective cohort study of 146,743 elderly Medicare patients hospitalized across the United Sates from 1994 to 1995, with follow-up conducted through linkage to administrative databases. Because approximately 10% of patients had not died during the follow-up period, it was necessary to use statistical methods to extrapolate survival curves beyond the 17 years of follow-up that were available. From the areas under the survival curves, they were able to estimate the average age-specific life expectancy for AMI patients across a variety of age spans, ranging from 65 to 97 years. For benchmarking purposes, they also used data from the Medicare administrative databases to determine the age-specific life expectancy for whites and blacks, and men and women in the general population. The authors conclude that at the age of 65, white women lose 4.9 years more of life after an AMI than white men, and black women lose 5 years more of life than black men. The differences between their conclusions and the CDC data are due in part to taking into account the fact the women live longer in the general population than men, which highlights the sensitivity of YPLL to the calculation methods utilized. Their analyses reinforce the need to improve the clinical presentation and treatment of all AMI patients but especially those of black race. They also demonstrate that the sex differences in YPLL are primarily due to differences in life expectancy in the general population rather than in the post-AMI population.
The estimation of life expectancy and YPLL from large clinical trial or observational cohort databases offers advantages over the life table methods traditionally used in epidemiology. These include not requiring everyone to have passed away during the follow-up period, and the ability to use regression modeling techniques to statistically adjust for the impact of patient characteristics and treatments on projected life expectancy. However, the major disadvantage of these methods is that they require extrapolation of the survival function beyond the follow-up period, and various statistical assumptions need to be made about the nature of the survival curve. Some empirical research has been conducted using clinical databases to validate survival projections using alternative statistical methods, but more research is required to determine the optimal methods for doing so (8,9).
The development of advanced methods to estimate life expectancy using large clinical databases also has other potential applications. For example, the projected impact on life expectancy and the cost effectiveness of new drug therapies or device therapies can be calculated from clinical trial data (10). Accurate life expectancy estimates would also prove invaluable to clinicians in making decisions about use of life-extending therapies in high-risk patients, such as use of implantable cardioverter defibrillators in chronic heart failure patients or transcatheter aortic valve replacement procedures in elderly aortic stenosis patients (9,11,12). Life expectancy and YPLL are infrequently used outcome measures in the field of clinical cardiovascular medicine. They will become more widely used in the future as the population ages, and the need for accurate estimates of life expectancy increases.
↵∗ Editorials published in the Journal of the American College of Cardiology reflect the views of the authors and do not necessarily represent the views of JACC or the American College of Cardiology.
Dr. Tu is supported by a Tier 1 Canada Research Chair in Health Services Research and a Career Investigator Award from the Heart and Stroke Foundation of Ontario.
- American College of Cardiology Foundation
- Ko D.T.,
- Austin P.C.,
- Tu J.V.,
- et al.
- National Center for Health Statistics
- Bucholz E.M.,
- Normand S.-L.T.,
- Wang Y.,
- Ma S.,
- Lin H.,
- Krumholz H.M.
- Mark D.B.,
- Nelson C.L.,
- Anstrom K.J.,
- et al.
- Holmes D.R.,
- Mack M.J.,
- Kaul S.,
- et al.