Tuesday, February 8, 2011
Journal of the American Geriatrics Society has a short article about illness trajectories in patients with ESRD.
It's part of a small trend I've noticed the last few years of more critically appraising (ie, with observational data) the important and widely promulgated 'illness trajectory' models used in palliative care education. This article (free full-text available from BMJ) is a concise summary of the illness trajectory concept and importance. It recapitulates these 3 trajectories (click on image on the right to see in full): rapid steady decline in cancer, slow/steady decline in dementia, and steady overall decline punctuated by marked declines followed by (incomplete) improvements seen in the organ failures such as CHF, COPD. They have been proposed as a way of helping patients understand their expected futures, as well as identifying patients appropriate for palliative care-type interventions, and have probably been part of countless palliative care didactic presentations - I've certainly seen them dozens of times, and have used them personally.
Regarding the 'small trend' I mentioned above - see for instance this paper looking at illness trajectories leading up to death in patients with CHF which essentially found there was no identifiable trajectory at all. On the other hand, this large analysis seems to support the distinct trajectories. As with most things, different methodology, patient populations, etc. make all of this difficult to interpret.
This paper looks at illness trajectories in patients with ESRD who were managed conservatively (without dialysis). None of the prior empirical studies on illness trajectories have included patients with ESRD (to my knowledge, and the authors of the paper suggest this as well). The analysis can be best thought of as a retrospective analysis (mortality follow-back) of data about functional status that were gathered prospectively - this paper comes from a larger study following 74 patients in the UK who chose conservative management of their ESRD. The patients (mean age 80 years) had ESRD and could have received dialysis (mean GFR at study entry was ~11 ml/min) but chose not to. The patients were followed for 2 years; 49 patients (66%) died in this time. They don't explicitly define median survival but they indicate median 'follow-up' was 8 months which probably represents the median survival.
What they found was that, in the year prior to death, functional status declined minimally (average KPS of 65 to 55) until the last month of life at which time it precipitously declined. The curve actually looks something like a hybrid of the classic cancer curve (with a rapid-but-several-month decline and the sudden death 'curve' which is really a sharp 90 degree right turn).
So, there we go, that's what we should tell our older patients who ask what will happen if they choose not to initiate dialysis.
Or not, and that's what I wanted to talk about - is how we use these ideas at the bedside, as well as in teaching. I'm curious as to what others do: do you draw those curves for patients and families at times? Do you find them useful in teaching housestaff etc. about how to recognize when someone is 'dying' or entering a 'terminal trajectory' and needs palliative care issues addressed?
I myself have tended to be skeptical of the curves, rightly or not, as actually describing reality, in the sense of being predictive. That is, can you take a COPD patient and tell them 'this will happen to you'? I do use them when I see it happening, when I see someone who appears to be following a classic trajectory as a way of normalizing and explaining the patient's experience to them: 'this is what happens with advanced COPD', etc. (and may or may not draw the curve, usually not, but some patients/families seem amenable to looking at the actual graph). I can't count the number of times I've had family members start talking about the last 6 months and you realize they're describing the organ failure curve ('every time they went to the hospital, they were never quite got back to how they were before'): in fact people seem to say that most with dementia patients in my experience, even thought they supposedly follow the frailty curve. Nonetheless there is a selection bias here - I am keying in on this for patients who clearly are following the curve, and don't think about it for the COPD patient who does fine for a while then precipitously dies, etc. Ie useful for explaining, less so for predicting. Certainly an over-reliance on looking for patients who fit the curve will lead one to miss a whole bunch of people who are dying.
I'm also not trying to set up a straw man argument here, none of the groups who have proposed and investigated illness trajectories have made sweeping predictive claims for them and I know that. That doesn't however answer the question of: why, and how, to use these models? Clearly, they have resonated with our community and medicine at large - they have legs!
I'll be in Vancouver next week - say Hi.