A propaganda video against modern medicine made by the notorious pseudoscientist Mike Adams is making its way around the Internet. The video is entitled “Why doctors are more dangerous than guns” and can be found here. The idea behind the video is that since more people supposedly die in association by health care related treatments such as prescription drugs and surgery than by getting killed by a homicidal shooter, the risk of modern medicine is higher than the risk of homicidal shooters. Predictably, it is the deadly doctor gambit (named by C0nc0rdance). The comparison is done between the absolute number of deaths associated with each category, when it should be about comparing the number of fatalities divided by the number of individuals exposed. This makes the argument statistically flawed, and therefore invalid.
To be more precise, the statistical flaw is known as floating numerator or denominator neglect. It means that you forget to ask the important statistical question “out of how many?”. Every year, a certain number of people will die from prescription drugs. But out of how many that took prescription drugs that year? Once we start looking into this, the deadly doctor gambit asserted by Mike Adams collapses.
The propaganda video references Starfield (2002). The strongest emphasis in the video is placed on the number of deaths per year from non-error, adverse effects of medications, so I will focus on this one in more detail, although the general argument I made can be applied for all categories.
This article says that 106000 hospitalized patients die every year from non-error, adverse effects of medications. 106000 deaths per year is, as the video correctly calculates, an average of around 290 people per day. But as we saw, this figure is meaningless on its own if we are interested in relative risk or odds ratio. 106000 out of how many?
As it turns out, the figure comes from another article by Lazarou, Pomeranz and Corey (1998). The article calculated that the number of fatal adverse drug reactions in hospital (fatal ADRIn) by multiplying the incidence of fatal ADRIn with the number of total hospital admissions in the United States during that same period. This is strictly speaking wrong as not all hospital admissions come with prescription drugs, but we can say that the error is not going to be that large. Most hospital stays involve giving the patient prescription drugs. The total number of hospital admissions during that period was, according to the article, around 33 million.
Guess what the incidence of fatal adverse drug reactions in hospital was according to Lazarou et. al.? It was 0.0019. That means that the incidence was less than 0.2%. If you find yourself in a room where a violent shooter is emptying the clip of his firearm, do you think that your chance of survival is more than 99.8%? That is highly doubtful. Therefore, it is wrong to say that the risk of death from prescription drugs is higher than death by homicidal shootings, because there has been denominator neglect. C0nc0rdance (2010) points out in his video about the deadly doctor gambit that there are fewer deaths from motorcycles than from cars, but that is only because cars are so many more than motorcycles. Making ambulances into motorcycles is a bad idea.
If we look at the bigger picture, Gu, Dillon and Burt (2010) shows that about 48.3% of the U. S. population uses prescription drugs during a specific time period (1 month before being asked). That is about 150 million people. The total amount of fatal adverse drug reactions during such a time period, both in and out of hospital has to be put in relation to this 150 million figure. Also note that for individuals not in hospital, they could have died due to overdose or drug abuse, which can hardly be directly blamed on doctors, especially if those drugs have been smuggled or stolen.
References and further reading
C0nc0rdance. (2010). The Deadly Doctor Gambit. Youtube video. Accessed 2012-08-04.
Starfield, B. (2000). Is US health really the best in the world? JAMA, 284(4), 483-485.
Lazarou J, P. B. H. C. P. N. (1998). Incidence of adverse drug reactions in hospitalized patients: A meta-analysis of prospective studies. JAMA: The Journal of the American Medical Association, 279(15), 1200-1205. doi: 10.1001/jama.279.15.1200
Gu Q, Dillon CF, Burt VL. (2010) Prescription drug use continues to increase: U.S. prescription drug data for 2007-2008. NCHS Data Brief. (42):1-8.