The article "Actual Causes of Death in the United States," Journal of the American Medical Association (JAMA), Vol. 270, No. 18, pp. 2207-2212 by McGinnis and Foege (1993) was interesting as it included specific preventable causes of death, as opposed to general statements.

I used the deaths reported in that article as the foundation of this compilation to which was added a breakdown of the illicit drug deaths from the Substance Abuse and Mental Health ServicesSourcebook, August 1995 edition, as well as its percentage of deaths. Then a subschedule for the illicit use of drugs was compiled using figures from the National Institute on Drug Abuse, Annual Medical Examiner Data, 1990 publication.

Table 2.06A of the Annual Medical Examiner Data publication provided a listing of the number of mentions for each drug most frequently mentioned by medical examiners from drug related deaths in 1990. This report was based on a total of 14,081 drug mentions, and a breakdown of individual percentages was tabulated by dividing the number of respective mentions by the total.

The estimated number of deaths was then calculated taking the corresponding percentage of the 20,000 deaths attributed to illicit drug use from the McGinnis and Foege article.

Actual Causes of Death in the United States, JAMA 270 (18): 2207-2212. This report was a met-analysis¹ intended to identify and quantify the major external (non-genetic) factors that contribute to death in the United States. From the "Methods" section of the doctors' article:

"This article summarizes published reports that attributed deaths to (preventable contributors) and presents a composite approximation of the totals reported for each. Articles published between 1977 and 1993 were identified through MEDLINE searches, reference citations, and expert consultation. Government reports and compilations of vital statistics and surveillance data were also obtained. All relevant analysis were reviewed in full. Those selected for use in developing estimates were those most often cited and those for which the methodological assumptions could be identified.

The limitations in the data should be underscored both with respect to deficiencies in the primary databases and to the disparate approaches used in the studies reviewed to arrive at estimates of the contribution of a factor to a particular health outcome. In some cases, assignments were attempted through simple tallies of available information about the presence or absence of a factor in association with a given outcome. In other cases, population attributable risk calculations were used to arrive at estimates based on determinations of the relative risk for a particular health outcome of a population exposed to a specific health risk. Some of the studies presented meta-analyses of reports in the literature on a given topic. Estimates were often limited by the adequacy of information as to disease prevalence, and the nature of the relationship to other contributing risk factors for the disease.

Despite their limitations, the results of such studies provide a sense of the relative impact of various factors on health in the Untied States. Derivation of the numbers presented in the Table is explained (in the remaining portion of the article) in the discussion of each category. Where well-established methodologies have been developed for making the estimates, as with tobacco and alcohol, they have been used approximately as reported. For areas of greater uncertainty, such as diet and activity patterns and toxic agents, a sum of the lower boundaries of the estimates for various disease outcomes has been used. Although several of these factors are interrelated in their actions, care has been taken to avoid double counting. Given the fragility of the data base involved and the fact that the studies cited use different approaches to derive estimates, these numbers should be viewed as first approximation."