Wednesday, July 29, 2009
Latest data on feeding tubes & dementia
Journal of the American Medical Directors Association recently published some more data on the natural history of feeding tube placement in nursing home patients with advanced dementia. No, not a controlled trial, nor even a prospective study (of which there have been hardly any), or even a comparative study of feeding tube outcomes (of which this as far as I know is the most recent and, while small, represents one of the better studies trying to see if there are measurable differences in survival between those who receive FT and those who don't), but this study does provide some good, recent, data on the natural history of FT placement including survival and complication rate.
The data largely come from the Minimum Data Set (short story: national mandatory database of patient characteristics collected on all patients admitted to a Medicare or Medicaid certified nursing home), and includes all patients admitted to a US nursing home with advanced dementia and no feeding tube at the time of admission in the year 2000. The study particularly focuses on ~5200 of these patients who received a feeding tube within a year of their first MDS assessment (presumably then within a year of their first admission to a NH). They did compare geographic variations in FT insertion rates, as well demographic characteristics of those who received FT vs. those who didn't.
There was a shocking regional variation found in rates of FT placement nationally. The average was 54/1000 patients (per year per NH resident with advanced dementia), with a low of 2.1/1000 patients in Utah and a high of 108/1000 patients in Mississippi: yes that's a 50-fold difference. Consistent with other data, non-white NH residents were much more likely to receive FT than white patients. The way they present this data is a little weird (not immediately easily interpretable) - it's not as rates of FT insertion in different racial/ethnic groups but instead percentage of patients in each group (received FT vs did not receive FT) represented by different groups. Either way - substantial differences: 89% of the no FT patient group were white vs. 67% of the +FT group; 8% vs. 25% were black; 2% vs, 6% were Hispanic. Far fewer percentage of patients in the +FT group had living wills/POAs/DNR orders as well.
2/3 of FT were placed in the context of an acute hospitalization: most common diagnosis (based on ICD9 codes associated with the hospitalization) were aspiration pneumonia, dehydration, dysphagia, and urinary tract infection (although the most common diagnosis, aspiration pneumonia, only accounted for 16.7 of all FT placements and you get the sense it was difficult to determine from diagnosis codes exactly why the FT was placed in many of these patients).
On to the natural history findings:
64% of patients died within a year of FT insertion; of those who died within 1 year median survival was 2 months. Median survival overall (for all patients who received a tube, not just those who died within 1 year) was ~160 days. To put it another way - half the patients who received a FT had died in 5 months (and most of those within 2 months). As far as complications go, about 20% of patients needed replacements or adjustments of their tubes in the 1st year.
The 50-fold difference in insertion rates is jaw-dropping, and clear indication that nationally decisions about TF insertion in advanced dementia, whatever they are based on, are not based on evidence or a consensus about best practice.
The survival data are within the ballpark of what is known - survival is poor - and clearly FT are commonly placed in dying patients. Until there's actually a well-designed, large, prospective study on FT we're really not going to know confidently if they impact survival; even if they do however this is further indication that (in advanced dementia) overall survival is terrible anyway. When I was a fellow I was a little obsessed about this and spent several months carefully combing through the data on survival in dementia & FT to 'figure it out.' I didn't, there is no definitive data, although pretty much any way it has been looked at no one has shown a survival benefit. My own conclusion is that if it's there, it's small, at least on a population level - there are probably some patients whose lives are dramatically shortened by FT placement, and some whose are greatly lengthened.
But lately I've more wondered does it really matter? Regarding patients with advanced dementia and aspiration pneumonia - what if there was a randomized controlled trial of FT placement and it showed that median survival was 5 months in those with FT placement and 2.5 months in those without. FTs double survival! FTs for everyone! Well that's not going to happen; because for a whole bunch of people 1) living an extra 2.5 months when you are bedbound and nonverbal and unable to recognize your family members is not a meaningful/valuable medical outcome, 2) especially if it means you have to undergo a painful, invasive procedure, with complication risks, which also means that 3) you are more likely to need to be restrained/tied down to your bed to make sure you don't remove that life-prolonging tube, and 4) to top it off it also means you lose one of your final remaining pleasures - eating - including the social pleasure of having a family member or (hopefully caring) nursing home staff member sit with you, interact with you, as they feed you, 'aspiration risk' be damned.
To me, that's a crappy deal, and I wouldn't care if TF are life-prolonging (particularly if my survival anyway would be 5 months) given 2-4.
That's for me, of course. There are some patients who feel differently, who feel that artificial nutrition should be provided no matter what if there's any suggestion of nutritional impairment or compromise (the tube feeds=love camp). This is a value system I don't share, personally, but further research showing TF don't improve survival aren't going to sway that group of people as decisions about TF placement are based in values/belief and not 'data' about survival. These patients will get their TF, and that's fine.
There are of course others for whom life-prolongation at any cost (see 2-4 above for the costs) is a goal - it's this group for whom better/more robust research about TF outcomes might be helpful. Included in this group of course are many, many doctors who might not choose this for themselves but who have a very, very hard time not recommending a treatment they think is life-prolonging (even if it's not QOL enhancing, again see 2-4 above), or (better yet) actively recommending against a treatment that possibly might be a little life-prolonging because it's likely QOL worsening. I.e. - I think we need this data for us (doctors), more than anyone. So it does matter, I guess, but perhaps for the wrong reasons.