The Public Health Paradigm

A famous article appeared in the 1990s in the Journal of the American Medical Association, by Drs. McGinnis and Foege. They noted the famous "10 leading causes of death" in the U.S., from death certificates. As of 1990 (and it hasn't changed much) these were heart disease, cancer, strokes, accidents, chronic obstructive pumonary disease (AKA emphsyema), pneumonia and influenza, diabetes, suicide, chronic liver disease and cirrhosis, and HIV disease.

Now, the items on this list are not of an entirely consistent character. Some are quite specific -- emphysema, AIDS -- others are broad categories containing many different diseases, such as cancer. But most of them are actual diagnoses -- biomedically defined conditions of disease -- present at the time of death, though a couple of them are causes of injury -- accidents, suicide. The physiological condition resulting in death in those cases would presumably be some sort of trauma, or you might argue that "suicide" as a cause of death really means depression. But the list for the most part focuses on diseases, that is on conditions of the patient which physicians are trained to diagnose and treat.

McGinnis and Foege point out that these conditions can all be traced to root causes farther back in time, called the etiology of a disease. They went through the published literature to extract what is known about the etiology of the major death certificate causes of death. They came up with 9 leading causes of death that they estimate are responsible for at least half of all deaths. There are no diseases on this list. The number one killer of Americans, by a healthy margin (pardon my word choice) is tobacco. Number two is bad diet and lack of exercise -- basically eating too much saturated fat and being obese. Then we have "toxic agents," which doesn't mean drinking drano, but rather occupational or environmental exposures. Deaths from illicit drug use are principally because of HIV and other infections resulting from drug injection, not from the direct effects of the drugs themselves. Sexual behavior kills mostly because of AIDS as well, plus other sexually transmitted diseases and a few maternal deaths resulting from unintended pregnancies. The researchers avoid double counting. For example, motor vehicle deaths resulting from alcohol intoxication are counted under alcohol rather than motor vehicles.

These estimates are very conservative. They have excluded any etiology that is not generally considered well established. Furthermore, all of these numbers are based on the lowest bounds of available estimates.

McGinnis and Foege have introduced a paradigm shift, from the biomedical paradigm -- what diseases kill people -- to the public health paradigm -- why do people die? They don't quite make it all the way. For example, they refer to "accidents", which is neither a biomedical nor a public health concept. The biomedical cause of death in these cases would presumably be cerebral concussion, shock, ruptured aorta, or whatever. But the term "accident" is alien to the public health paradigm since it implies randomness, an act of God or the cosmos, which cannot be foreseen or prevented.

Similarly, McGinnis and Foege list "microbial agents", which belong to the biomedical paradigm. From a public health standpoint we would ask why people succumb to microbial agents. In the case of HIV, they have asked this question. Hence they do not list HIV, but rather sexual behavior and injection drug use, and deaths from HIV are not counted under "microbial agents". They also point out that most influenza and pneumonia deaths occur among people who are already sick and weak, often due to tobacco, diet and alcohol. Some of these they assign to those agents.

However, nosocomial infections -- that is, infections people get while in hospitals, which are very dangerous and disease ridden places -- as deaths due to microbial agents. From a public health standpoint, these deaths are caused by biomedical intervention in the form of hospitalization. And we could make public health arguments for many other deaths counted as due to microbial agents.

This paradigm shift has enormous implications. McGinnis and Foege point out that the U.S. spendt nearly $1 trillion per year on biomedical services (it's a lot more now!). That comes to more than $14,000 a year for a family of four, on average. Most of that sum will be spent to treat the physiopathological processes recorded on death certificates as the nation's leading causes of death. At the same time, the U.S. Centers for Disease control estimated total national spending to control etiologic factors as less than 5% of that trillion dollars, that is less than $700 per family -- and thanks to the Bush Administration, proportionately it is even less today. Furthermore, that trillion dollars in biomedical intervention obviously isn't doing much good in preventing the leading causes of death -- once people have cancer or heart disease, or AIDS, they are likely ultimately to die from those conditions regardless of the biomedical resources that are brought to bear.

So we have a major misallocation of society's resources. This is a political issue which is gravely underappreciated.