They Died From Pot! Oops, Nevermind

"Drug-induced" Death Originally, I got started on investigating drug-induced deaths as a result of having read the latest annual National Drug Control Strategy from the drug czar's office (ONDCP) a few years ago. I noticed that, in tiny print below the table of drug-induced deaths, there was a series of alpha-numeric codes used to create the death tally. They also mentioned that the data was available on-line from the Centers for Disease Control. When I went to the CDC's mortality database system (WONDER) and plugged in the codes, I did, indeed, find that the numbers tallied as claimed, but I learned something more important: most of those deaths are actually due to legal drugs. Of course, I also learned that several people were said to have died directly as a result of using cannabis (marijuana), and this report is the culmination of my efforts to dig for the truth in the matter. It Just Has to be a Mistake! When I first built my webpages exposing the realities behind the drug-induced death claims, I made a tactical blunder: I assumed that the cannabis deaths listed in the CDC database simply had to be the result of human error. Given that the database relies heavily on a short-hand coding system (the ICD - or International Classification of Death), and my knowledge of data systems, I was confident that the deaths could probably be best explained as something called a "one-off" error. Such an error is caused when the wrong number is typed as part of a data entry. Since the codes for cocaine related deaths were the closest likely candidate for such an error, that is how I described the issue on my web site. Ask the Owners I was called on my assessment in a usenet post, so became motivated to put the effort into finding the reality behind the numbers. I sent an email to the organization charged with the responsibility for the mortality database at the National Center for Health Statistics (NCHS), specifically stating my theory. In response, they assured me that such an error would likely be caught thanks to their multiple layers of data assurance. I had to take that one on faith, however, as the database contained over 43 million death records. With a system that large, there is an almost absolute certainty that there will be errors. Humans, after all, remain a large part of the process, and some of the records date to a time well before the era of personal computers. But, I ultimately had to accept their explanation that the data were faithfully and accurately transcribed from the death certificate into the database. In other words, if there were any errors in the process, then they weren't coming from the deacons of the death data. Mental Illness? The other striking feature about the deaths attributed to marijuana is that none are described as being the result of poisoning or overdose. That part made sense at least, as marijuana is essentially non-toxic. There is a scale to indicate the toxicity of various substances, and THC (marijuana's active ingredient) has a solid claim to be being "non-toxic." The best estimate is that it would require about 40,000 times the effective dose to lead to a lethal dose in 50 percent of people subjected to such a dose. Instead, the cannabis deaths were coded as being due to some form of mental illness involving marijuana. I emailed the NCHS again, expressing my astonishment that any sort of mental illness could be described as the cause of someone's death. They then explained that a death certificate actually has a wealth of data associated with it, including as much as could be determined about the conditions and factors surrounding a death. Thus, there are usually "multiple causes" of death listed on the certificate, one of which is designated as the underlying cause -- the factor that, if avoided, may have prevented the death. That specific cause is what then gets tallied in the official mortality counts. "Multiple" Causes of Death The following is quoted from the 2003 edition of Physician's Handbook on Medical Certification of Death, DHHS Publication No. (PHS) 2003-1108, p.12: "The cause-of-death section consists of two parts. Part I is for reporting a chain of events leading directly to death, with the immediate cause of death (the final disease, injury, or complication directly causing death) on line (a) and the underlying cause of death (the disease or injury that initiated the chain of events that led directly and inevitably to death) on the lowest used line. Part II is for reporting all other significant diseases, conditions, or injuries that contributed to death but which did not result in the underlying cause of death given in Part I." So, in essence, a death could be attributed to marijuana use in this type of scenario: someone gets high, walks in front of a car and thus: "because he got high, he was discombobulated and walked in front of a car. The "immediate" cause of death would be hit by car, while the "underlying" cause of death would be because he got high. Obviously, I Needed More Data I emailed the NCHS folks again enquiring whether or not the detailed data about the multiple causes of death was also available for on-line research. They said it wasn't but that they would be happy to send me a CD-ROM of their current set of public use files covering the year 2001. I was warned that the data was just raw form ASCII, but that complete documentation was included on the CD. Fortunately, I've spent part of my career as a computer geek, so busting apart the data was not very difficult. Once I wrote the code I needed to find the data and extract the "multiple causes of death," I was able to determine the full set of circumstances recorded on the death certificates for the three "marijuana-induced" deaths that occurred during 2001. There Are Rules For This One of the important things about what is recorded in the cause of death part of a death certificate, is that the order in which the conditions are listed is intended to provide a chain of events sequence between the underlying and immediate causes of death. This information is also encoded within the data stream for each record in the mortality data itself. After I extracted all the data for the cannabis deaths, I was able to decipher some, but not all of the codes I found. In addition, I discovered that the rules of which particular code was used for the underlying cause were apparently being violated for each of the deaths being blamed on marijuana use. The last entry in Part I is supposed to be entered as the underlying cause of death, and the additional data listed in Part II is specifically not supposed to be used as the underlying cause at all. However, I found that for the deaths blamed on marijuana, these rules were not followed. Indeed, the specifics I found on the three deaths clearly indicate that the underlying cause of death is indeed designated, and that the references to marijuana use as a contributory factor is listed in Part II in every one of them. I refer you again to the definition from the official handbook I referenced above which states quite clearly that: "Part II is for reporting all other significant diseases, conditions, or injuries that contributed to death but which did not result in the underlying cause of death given in Part I." Of Course, There Are Exceptions Too I sent another email to the NCHS and disclosed my findings: according to the rulebooks on filling out death certificates and managing the mortality data, the three cannabis deaths should have been coded as due to "Y19," "R56.8," and "R09.2": Y19 - Poisoning by and exposure to other and unspecified chemicals and noxious substances, undetermined intent R56.8 - Other and unspecified convulsions R09.2 - Respiratory arrest In reply, the folks at NCHS informed me that their mission, ultimately, is to help prevent unnecessary deaths. When attempting to define a preventable condition, it is of course better to be as specific as possible. In these cases, I was told, the deaths coded as due to respiratory arrest and unspecified convulsions qualify as conditions that just aren't very informative in terms of discovering mortality trends or designing preventive measures. In such cases, the procedure is to use whatever other specific data is available -- even if it is in Part II of the death certificate. In essence then, the deaths are blamed on cannabis simply because there was nothing else to blame them on. In the case of the death that should have been coded as "Y19," I was told that I am correct. However, the data will eternally remain the way it is, as once the records are made "public" it is quite impractical to attempt such a minor modification. Public Health Concern? As the drug czar's war machine continues to grind against those who use the cannabis plant to amuse themselves, it is frequently said to be a matter of the government doing its job protecting the health of the citizens. From the perspective of mortality, it appears there is very little justification for "protecting" people from marijuana. Indeed, all of the available data the government collects about the impacts of marijuana use bears out one consistent finding: the greatest danger of using marijuana is simply being caught with it. But what about deaths that are marijuana "related"?