Monday, June 11, 2007
Annals of Internal Medicine has published another paper suggesting atypical antipsychotics increase mortality in elderly demented patients. This one is a large retrospective study from Ontario using administrative database data comparing demented patients who received antipsychotics to those who didn't (patients with comorbid mental illnesses like schizophrenia or who were receiving palliative care services were excluded). I'm not going to belabor their findings because they're similar to other ones: atypical antipsychotic use is associated with higher mortality (albeit a tiny increase) - in this study it was ~1% increase and evident by as little as 30 days (this is consistent with prospective data). What this study adds is that it also compares risk of death from conventional vs. atypical antipsychotics and finds that the risk of death with conventional antipsychotics is significantly higher (significant in that it is a statistically sound conclusion - actual magnitude of increase is ~2.6%...tiny). The problem with this is that this is all uncontrolled data & doesn't exclude the possibility, or even likelihood, of confounding, although I personally believe that if atypicals increase one's risk of mortality why wouldn't conventional ones? Frankly if these agents were wonderfully effective for behaviors in dementia I'd be arguing, & I think most people would too, that a few percentage increase in mortality is acceptable given the high mortality anyway of this population, quality of life considerations, etc. However they aren't particularly effective, so... I have this vivid memory from a geriatrics rotation in my residency in which the geriatrician was liberally prescribing Geodon & Valproate to his demented patients with behaviors and proudly noted to me how he was on the cutting edge of things. I've always wondered how his practice changed in the subsequent years.
(Related posts here, here.)
For you public radio fans out there: WNYC's Radio Lab recently aired a show about the placebo effect. It is an interesting listen and addresses, amongst many things, pain physiology and bedside manner. Radio Lab is quite an interesting show - hour long looks at a variety of topics - all science-related - but usually told through stories (of patients, researchers, etc.) instead of typical reportage. Think of it as what This American Life would be like if it was about science/the natural world. The show goes straight for the Big Ideas - space, time, morality, identity, etc. (You can download or stream the show from the website. Podcast also available through iTunes).
Back to the dementia theme: JAMA has a review on diagnosing dementia as part of their "rational (i.e. evidence-based) clinical exam" series. EBM aside this is actually a nice review of dementia & its symptoms & real-life manifestations - a good one for the teaching file. Its conclusion: there are a variety of brief, validated tools to diagnose (or at least accurately screen) for dementia at the bedside, etc. etc. They didn't however mention the palliative care method: simply asking the patient, "Are you demented?"
And finally, for you arrest-prognosis buffs out there, American Heart Journal has an article about long-term prognosis after in-hospital cardiac arrest. It is a retrospective study from a single US veterans' hospital looking at >700 consecutive in-house arrests from 1995-2004. The point of the study was to look at what was associated with long term survival for those who actually survived an arrest. 6.6% of patients actually survived to discharge. 1- and 3-year survival rates were 68% & 41% for those who survived to discharge (making the overall 1- and 3- year survival rates for in-house arrest 5% & 3%). They note that during this time the hospital had a QI project to improve their response to in-house arrests which increased their immediate survival from CPR by 2.6 times. This, however, had zero effect on long-term survival. (The authors go on to look at things like receiving an ICD and long-term survival but, frankly, there were only 9 patients who both survived to discharge and received an ICD and it's not clear to me that any conclusions can be drawn from so few patients.) For me, the most interesting aspect of the article was the notice given to the fact that the QI project to improve in-house survival didn't affect long-term survival. It corroborates what we tell patients with whom we are discussing code status - even if it temporarily restores life--a beating heart--it will not restore one to health/ameliorate the underlying condition causing the arrest (patients with isolated ventricular arrhythmias aside).