Monday, December 5, 2005
Well, the debate evolves. The New England Journal of Medicine has published a study trying to compare the risk of death in elderly patients who take conventional vs atypical antipsychotics. (Check out the Oct. 20 posting for another blog about this.) Everyone has been a little excited recently about the apparent increased risk of death associated with elderly dementia patients on newer antipsychotics, and many have asked the obvious question "Do conventional antipsychotics have this increased risk also?" The NEJM article tries to answer this--it's suggestive--but overall I think it just muddies the water, despite getting a decent amount of media attention. It's a retrospective cohort study from a Pennsylvania prescription database that looks at people 65 years old & older who received either conventional or atypical antipsychotics at any time. 23,000 people were evaluated over a 10 year period; the authors looked at who died within 180 days of receiving an antipsychotic prescription. They found that 18% of people who received typical antipsychotics died within 180 days of the prescription vs 14.6% of those who received newer agents. Not a big difference, but highly significant in this large population. This mortality difference was largest immediately after the index prescription date and tapered off afterwards (that is, there was a wide difference in mortality within 30 days of the index prescription which became less apparent as time passed beyond the intial date of the prescription). The authors then did multiple subgroup analyses controlling for medical comorbidities and still found a higher risk amongst the conventional antipsychotics group.
My interpretation of the data presented, even with the subgroup analyses, is that--compared to atypical antipsychotics--the use of typical ones in this population is a marker of a higher risk of death, & that no grander conclusions can be made. Looking through the characteristics of the conventional vs atypical group it is clear that they are medically distinct groups (those who received conventional antipsychotics were more likely to have "medical" diseases, whereas the atypical group were more likely to have neuro or psychiatric illnesses such as dementia, delirium, mood disorders). The attempt to control for comorbidity (things like CHF or HIV) does not rule out an unmeasured factor being related to mortality and conventional antipsychotic use. The fact that the difference in mortality between conventional & atypical antipsychotics was more pronounced early on may mean nothing more than docs reaching for haloperiodol more often than quetiapine for patients they sense are doing poorly or dying. In addition, the data implicating atypicals comes from recent randomized placebo-controlled trials which unfortunately clearly show a small increase in the risk of death amongst elderly demented people taking atypicals chronically. Not only was this new study not a randomized trial, but there was no way of measuring in the study how much & for how long the patients received the antipsychotics--many of these precriptions could have been for one time orders, etc. My point is that what the authors of this article are catching here is a different population & circumstance than the recent trials involving atypicals.
Of course this does not mean that conventional antipsychotics are any more safer than newer ones! They may very well not be, & if the purported mechanisms of lethality from antipsychotics (anticholinergic effects, QT prolongation, etc) are true then it is likely that conventional agents are as harmful as newer ones. This paper is suggestive at best however and not much more should be taken from it. My guess is is that this will be looked at with increasing scrutiny in the future.
NEJM has also early-released a few articles & an editorial about the outbreak of the new strain of C. difficile in Quebec. Humbling stuff, that could seriously affect our patients lives if it becomes pandemic.
P.S. As I write this Monday PM, Blogspot seems to be having trouble with permalinks, so let me know if you have any trouble with links on the site. Thanks.