I am responding to Dr. Leon Gussow's lively critique of our paper about sleeping pill risks. (EMN 2012;34:8; http://bit.ly/KE9EE1.)
Dr. Gussow wrote, “If sedatives and hypnotics taken at a dose of 1.5 pills per month can quadruple one's risk of death, they must be among the most toxic substances known to man.” Yes, perhaps they are. You may have noticed that Texas is now using pentobarbital alone to execute prisoners.
Dr. Gussow wrote, “If sleeping pills kill at least four times as many people in the United States as gunshot wounds and automobile accidents combined, there is an unprecedented and unappreciated ongoing public health catastrophe.” He is right about the unappreciated public health catastrophe but wrong about the lack of precedent. The risk from sleeping pills is in the same ballpark as the risk from cigarettes, which was also unappreciated for many years. As a matter of fact, the Cancer Prevention Study II, which is the World Health Organization standard for assessing cigarette risks, showed a rather comparable mortality risk associated with taking prescription sleeping pills. (Cancer 2002;94:500.) The studies on cigarettes and sleeping pills used exactly the same epidemiologic design. Incidentally, sleeping pill users have a higher rate of automobile crashes. Most emergency physicians have seen a patient injured in an automobile crash caused by the untimely use of a single sleeping pill.
The magnitude of the low-dose hypnotic mortality risk does seem quite surprising, but it occurred mainly among elderly patients with multiple comorbidities, who were often taking multiple interacting drugs such as narcotics and anticonvulsants. Most emergency physicians know that when a patient dies unexpectedly during sleep, there is often no autopsy and no way to recognize if drug-induced respiratory arrest caused the death, perhaps in combination with sleep apnea.
Dr. Gussow wants extraordinary evidence. Just like you won't find controlled trials proving that cigarettes cause cancer and death, you won't find that the drug companies have done controlled trials with sufficient patient exposure to determine if hypnotics cause excess mortality. Nineteen extraordinary studies, however, show significant mortality risks associated with hypnotic drugs but no studies suggesting life-preserving effects. There is uncertainty about the magnitude of the sleeping pill risk, but it definitely appears too large.
If you do not warn your patients about the mortality and cancer risks of sleeping pills, you might later hear from their attorneys.
Daniel F. Kripke, MD
Co-Director, Research Scripps Clinic Viterbi Family Sleep Center San Diego
Dr. Gussow responds: I am glad that Dr. Kripke found my criticism of his recent paper “lively,” and I appreciate his taking time to comment, but I find the arguments put forth in his letter — as I found those expressed in his paper — unconvincing.
The fact that a large amount of intravenous barbiturate can kill does not mean that a therapeutic dose of, say, zolpidem ranks among the most toxic substances known to man. This argument puts the “non” in non sequitur. A moderate amount of air, injected intravenously, will also be fatal. This does not mean that air is the toxic equivalent of cyanide.
It still seems to me that the health risks Dr. Kripke and his colleagues attribute to sleeping pills would indeed be unprecedented, if they were real. The abstract to his paper claims that: “Receiving hypnotic prescriptions was associated with greater than threefold increased hazards of death even when prescribed [more than] 18 pills/year.” This is equivalent to claiming that buying — not necessarily smoking — fewer than 18 cigarettes a year triples one's risk of lung cancer. Not even the most fervent anti-smoking advocate would make that claim.
The Cancer Prevention Study that Dr. Kripke cites deals with nutrition, and makes no mention of sleeping pills. It is not true that the evidence that demonstrated the risks of smoking “is exactly the same epidemiologic design” as that offered regarding sleeping pills. Over many years, numerous careful case-control studies and animal experiments established the biological plausibility of a causal connection between smoking and lung cancer (as well as other health risks). None of these studies involved querying electronic medical records. I'd like to know what “nineteen extraordinary studies” showing health risks associated with hypnotic medications Dr. Kripke has in mind. His paper references at most seven such studies, none of them extraordinary and several of which he himself co-authored.
To end on a note of concord, I do agree with Dr. Kripke that one should not take a sleeping pill and then go for a drive.