The immediate vs the important.

Abstract

WHEN HIPPOCRATES OBSERVED THAT “PROTECTing and developing health must rank even above that of restoring it when it is impaired,” he set a standard that is difficult to meet. One of the most difficult challenges is to ensure that the urgent does not crowd out the important. In health, this challenge is especially difficult because urgent matters can be so riveting. At the personal level, the presence of illness or injury often overpowers all other concerns, and the search for effective treatment often dominates all other pursuits. At the policy level, with 15% of the US gross domestic product devoted to health care, medical care expenditures often drive decisions in which cost cutting is aimed first at discretionary investments, such as those in prevention and public health that offer the greatest prospects for overall health improvement. Hence, tools are needed to facilitate the gathering, analyzing, and reporting of data in a fashion that enables taking action not merely on the urgent but on issues most important to the health of a population. In health, the most basic measure of importance is found in mortality tables—the registry of leading killers drawn from death certificate data. Despite the vital statistics system relying substantially on unverified physician reports that may be subject to various sources of bias, the results over time seem to be reasonably reflective of the dominant diagnoses at the time of death. For example, while the occasional suicide may be reported as an accidental death, a coronary heart disease death is likely to be accurately recorded, with reasonable indication of related conditions. Every death has a definable history that usually can be traced back for decades and sometimes even for generations. Reporting of deaths, diseases, and disabilities in traditional diagnostic categories tends to obscure the importance of factors that often play determinant antecedent roles in the occurrence of the reported conditions. When it comes to ranking health problems and committing resources, attention seems more naturally drawn to the conditions most proximate to serious illness or death. For instance, a diagnosis of lung cancer draws interest and immediate concern because it is lifethreatening and the certainty of its threat is unquestioned. Yet, nearly 9 times out of 10 lung cancer is merely the natural pathologic consequence of exposure to tobacco use, the single most prominent actual cause of the lung cancer. Ability to make progress on many of the key health challenges will remain constrained until focus and resources are directed to the root causes of these conditions. Fortunately, much has been learned from research of the past generation about the factors leading to disease and injury and the magnitude of their contributions. The National Center for Health Statistics reports that the 10 leading causes of death in the United States in 2000 were heart disease, cancer, stroke, chronic lower respiratory tract disease, unintentional injury, diabetes, influenza and pneumonia, Alzheimer disease, nephritis, and septicemia. The article by Mokdad and colleagues in this issue of THE JOURNAL assesses these mortality data against current knowledge about the contributors to those conditions. The findings indicate that the leading actual causes of death for 2000 are tobacco, poor diet and physical inactivity, alcohol consumption, microbial agents, toxic agents, motor vehicles, firearms, sexual behaviors, and illicit use of drugs. Together, these causes accounted for approximately half of all deaths in 2000, with nearly two fifths attributable to the top 3: tobacco, poor diet and physical inactivity, and alcohol use. This analysis is an update of our earlier report, which estimated the contributions of the actual causes of death in 1990. The order of the ranking for 2000 is similar to that for 1990, with an especially notable increase in those deaths attributable to poor diets and physical inactivity and a decline in deaths due to sexual behavior. However, there are some differences in the approaches used in these 2 studies. First, the 1990 estimate for the impact of poor diet and physical inactivity was drawn from the range of estimates of the proportion of deaths from conditions related to these factors. In contrast, Mokdad et al used hazard ratios in an ambitious primary computation of annual deaths from the specific issue of obesity, rather than the full range of conditions related to diet and activity patterns. They added a correction of 15000 for the non–obesityrelated deaths but noted that this may be conservative. This estimate seems low, given the substantial number of cardiovascular and cancer deaths that may be attributable to the dietary and inactivity patterns of those who are not overweight or obese. Second, the 1990 assessment parsed the alcohol-related portion of motor vehicle fatalities and assigned these deaths to the alcohol category, which explains the higher 2000 number for motor vehicle deaths and lower number for alcohol deaths estimated by Mokdad et