A Sad Direction in Home Dialysis Research: The FHN 2 Nocturnal Survival Analysis
First, I will paraphrase the introduction I used for last week’s post “The Absurd”. I hope it contextualizes this second in a duo of posts, though some may consider them “rants” or, more kindly perhaps, “Don Quixote moments” as I tilt (again) at the sacred windmills of crazy dialysis.
Last week I began something like this …
In recent days, the faith I have always had that somewhere, somehow, in the immense firmament of dialysis, I might be able to make a small difference, has been hit by not just one, but two broadsides of friendly-fire (or rather, fire that ought to have been friendly)…one of which is quite absurd, the other of which makes me sad. Floyd Patterson-like (the first real heavyweight “scalp” of Mohammed Ali), I have that feeling that I am sagging on the ropes!
Last week, I blogged The Absurd—the extraordinary PCORI project of the Arbor Research team (read DOPPS team) who have been funded a healthy $1.8M to develop an unhealthy dialysis decision aid that recognizes only two dialysis options: centre-based HD and home PD.
Yes, only two!
This week, it is time for me to consider “The Sad.”
#2: “The Sad”
Well, for two reasons:
First, and foremost, I think that the study this post addresses is a sad piece of research. But, is it just sad? Is it bad? Or, is it both?
Secondly, the research and the publications it has spawned is sad on a personal level, as I count some of the authors among my most respected, revered and good friends. Indeed, they are respected, revered and good friends to dialysis, to dialysis choices, to dialysis at home, and to dialysis patients everywhere. Between them, they have done some of the most seminal dialysis work performed anywhere in the dialysis world over the past 3 decades. It therefore hurts me to be critical of their most recent work—yet, I find I must be, and must say so.
The study in question is the FHN2 trial(1). While I have commented elsewhere on what I believe were the invalid conclusions drawn in the initial publication of the results of this trial, yet as trials do, it seems to keep on “giving.” And, with each “give,” the original trial errors and insufficiencies compound. The most recent “gift” from the FHN2 trial is a new “analysis”—this time of the “long-term survival data.”(2)
In my view, this second data set (on even fewer numbers) suffers from all the same errors as the parent paper, and more. Yet, as the data are derived from a randomized control trial (RCT), how could they possibly be wrong?
The true bête noir here is not the research itself, nor the researchers, but the relatively recent and often quite unrealistic “requirement” of funders, payers, and legislators to slavishly demand—and only act upon—the outcomes of RCTs. This action often flows without a thought for whether the trial (or trials) are valid, or whether the results are applicable across the spectrum of the disease, treatment, or program they have aspired to study.
I have written on this issue before, first in 2007 in Nephrology, News & Issues(3), and then in a requested editorial piece for the American Society of Nephrology (ASN) publication KidneyNews in 2013.(4)
In simple terms, it is relatively easy to do a RCT of a pill vs. placebo and get a relatively certain result. But, to try to randomize two dialysis modalities—especially when they “randomly” enforce hugely differing lifestyle adaptations onto the participants—well, that’s a different matter entirely.
So...what was wrong with the FHN2 trial, a trial which purported to compared frequent nocturnal home HD with (initially at least) conventional centre-based HD?
- First and foremost, recruitment was so fraught that the trial only achieved an enrolment of ~33% of the number that, prior to the start of the trial, was predicted to be needed to show a statistically sound difference. This is called “powering,” and the FHN2 trial was woefully underpowered. It was predicted that 275 patients would be needed to show a statistically significant result. Instead, just 87 were recruited and, of these 87, only 78 provided a full 12 months data (= 28% of the required “n”).
- Recruitment was difficult, with forced randomisation (home vs centre). As most recruits wanted home and not centre-based dialysis, the recruiting rules had to be altered mid-trial, allowing the conventional group to also dialyse at home. This alone—to be honest—should have rendered the trial invalid, as it made the conventional comparator group un-comparable. Even so, the trial went on…
- ~50% of the patients were incident patients. They had no prior dialysis exposure nor dialysis-dependent time to develop the left ventricular hypertrophy (LVH) from which the primary end-point of the trial required them to show regression (improvement).
- With an unusually high number of incident patients in the initially centre-based “conventional” group, many still had the benefits of residual renal function to augment their lesser dialysis provision.
- Only 87% of the 45 who were randomised to nocturnal home dialysis actually completed the full 12 months of the trial.
- Only 25% of those in the home nocturnal group actually performed the ≥5 nights/week that the trial required of them.
- Finally, and tellingly, there was the unknowable, and un-comparable impact(s) of the lifestyle differences that separate centre and home-based care, differences that were allowed to equalise and thus benefit the conventional group, mid-trial.
Yet, despite all this, the nocturnal arm very nearly reached the magical p ≤ 0.05 to record a statistical benefit…but not quite! So, near enough ain’t good enough and the conclusion? Frequent nocturnal dialysis at home is no better then conventional dialysis in a centre (or at home).
In a word, it was bunkum!
And…for confirmation that it was “bunkum,” ask a home HD patient.
Better, ask a few who have been party to both.
Though I commend and applaud the trial team for trying, I believe that the results of this trial should never have been published, and certainly not as a negative trial.
Now, 2-3 years later, a new analysis of an un-analysable trial has appeared—this time, on survival(2). An avalanche of papers (probably mostly not read by US researchers) have emerged from ANZ(5), EU countries, and Canada(6) that disagree and report better long term survival with home HD regimens. Yet, here, a pitifully small number of patients (a little over 30 in each arm) who were followed for only a median period of 3½ years were reported to show that survival in the nocturnal group was less than in the conventional “centre+home” group. Again, where is the power in this study?
There is one supplemental graph (see below) in this latest report that shows the clear problem the authors had with performing and presenting this study. It compares the duration and frequency data for the “conventional centre-based HD group” (in the top three panels) with the home nocturnal group (in the bottom three panels).
At the start, the conventional group all began a centre-based 3 x low hour/week regimen with the dots (representing patients) nicely grouped in the bottom corner (top left graph). Already problematically, the home nocturnal group had a spread of hours—commonly more, and commonly more frequent, but not consistently so—with a wide scatter of frequencies and hours (bottom left graph).
So, not all of the home nocturnal patients were doing the required high frequency, extended hour program the trial demanded.
But, worse was to come …
By trial years’ end, many of the “conventional” group patients had “morphed” into longer hour and more frequent regimens. In the second two sets of comparator panels, the “conventional” dots at 1 year and >1 year post trial (middle and right top) now look much more like the dots for the nocturnal patients in the bottom middle and right graphs.
Indeed, the duration and frequency of the so-called “conventional patients” were now proving to be indistinguishable from the nocturnal group with who they were being compared.
This “morphing,” or convergence of a trial “control” group into a replica of a group with whom they were being compared should make any comparison or conclusion untenable. Yet, the FHN2 trial concluded that as the outcomes of the two groups were similar, home nocturnal HD could not be shown to have any benefit when compared to conventional centre-based care.
An ANZ study(5) reported 26,016 patients with a follow-up of 856,007 months and showed a clear survival advantage for those at home. A further study from the same authors (currently in preparation) will further tease out the conventional home HD vs. extended hour home HD outcomes from an even larger data set.
The abstract for the FHN2 survival paper does at least report the trial’s limitations as: …“these results should be interpreted cautiously due to (1) a surprisingly low mortality rate for the individuals randomly assigned to conventional home hemodialysis (2) the low statistical power for a mortality comparison due to (3) the small sample size, and (4) the high rate of hemodialysis prescription changes”
Yet…they still concluded!
So will I—and let you also draw yours, as you wish.
- The FHN2 Trial. Found at: http://www.nature.com/ki/journal/v80/n10/abs/ki2011213a.html
- Rocco MV et al. Long-term Effects of Frequent Nocturnal Hemodialysis on Mortality: The Frequent Hemodialysis Network (FHN) Nocturnal Trial. Am J Kidney Dis. 2015 Apr 8 [Epub ahead of print]. http://www.ncbi.nlm.nih.gov/pubmed/25863828
- Agar JWM. Should the Medicare ESRD Program fund daily and nocturnal hemodialysis? Neph News & Issues. 21(12): 48-56. November 2007. http://www.ncbi.nlm.nih.gov/pubmed/?term=agar+j+medicare
- Agar JWM. Home Hemodialysis: Do we need more randomized controlled data? ASN Kidney News. September 2014. 9-10. http://onlinedigeditions.com/publication/?i=224633
- Marshall MR, Hawley CM, Kerr PG, Polkinghorne KR, Marshall RJ, Agar JWM, Home Hemodialysis and Mortality Risk in Australian and New Zealand Populations. Abstract to be found at http://www.ajkd.org/article/S0272-6386%2811%2900942-5/abstract
- Pauly RP, Gill JS, Rose CL, Asad RA, Chery A, Pierratos A, Chan CT Survival among nocturnal home haemodialysis patients compared to kidney transplant recipients. Abstract to this paper can be found at http://ndt.oxfordjournals.org/content/24/9/2915.abstract