Home Hemodialysis: Do We Need More Randomized Controlled Data?

When snake oil salesmen peddled their cure-alls, an undefended populace fell prey to the “best story,” the “best sell,” and the “most persuasive line.” Then, as remedy upon remedy failed to prove effective, to be safe, or to give value for money, greater scientific rigor was demanded of medical intervention. With statistical methods improving in parallel, “proof by clinical trial” emerged.

Trials of a single intervention—an active drug against a placebo, two active drugs head to head—are relatively simple to design and blind, the latter preferably applying to both investigators and subjects. If well designed and conducted, these trials commonly yield a useful and reliable answer.

Conversely, when multiple conflicting, competing, and differing factors exist in both trial arms—especially if these have an impact beyond the subject alone and affect the behavioral patterns, finances, and lifestyle of an entire family unit—conclusions blur, and outcome interpretation becomes impossible. This has been a core “insurmountable” for the recent attempts to conduct randomized control trials (RCTs) of home versus conventional hemodialysis.

Although the RCT is commonly regarded as the gold standard trial method—and so it is for assessing single or even multiple but controllable interventions—RCT design, subject selection, and trial conduct are crucial. An RCT that poses the wrong question, recruits insufficient subjects, lacks adequate statistical power, selects the wrong end points, or inappropriately selects surrogate outcomes may grievously mislead or misdirect.

Many believe that the two “landmark” RCTs in hemodialysis, the National Cooperative Dialysis Study (NCDS) (1) and the HEMO study (2), both inadvertently led to misinterpretations that fundamentally changed the course of dialysis by seriously impeding the very thing an RCT is supposed to ensure: improvement in outcomes. This has been most evident in the United States, where for 2 or 3 decades the blight of Kt/V urea, and gross inadequacies in dialysis duration that flowed from these RCTs, have directly diminished both patient well-being and survival. I dare suggest that the recent Frequent Hemodialysis Network 2nd arm trial—the FHN(2) trial (3)—of extended-hour home hemodialysis versus conventional hemodialysis is in the same category and risks leaving a similar misinterpretation. Yet, all three are RCT analyses…so, they must be right, eh?

Such is the aura of the epithet “RCT” that any conclusion drawn by an RCT becomes “lore” in the blink of an eye. Like apologies that follow media errors, printed much later, recorded in small print, appearing on page 47 near the obituary section, the error for which it “apologizes” long ingrained, dissenting opinions of RCT results either are dismissed or appear too late to alter the RCT “message.” This effect is amplified if the RCT in question has been long awaited and widely anticipated (advertised) to provide the “authoritative answer”—as was the case with the FHN trials. But although FHN(2) purported to yield an answer, it did not.

Even more concerning, now that RCTs have garnered such a bulletproof aura that legislators and funding agencies now structure practice regulations and funding models based on their outcomes, an incorrect conclusion from FHN(2) risks enshrining a further grave dialysis misstep.

Twardowski and Misra (4) nicely summarized the complexity—no, the impossibility—of achieving fair trial conduct and believable outcomes in an area as fraught as modality choice, wherein multiple competing comorbidities, socioeconomic circumstances, and expectations affect every dialysis patient differently. To be honest, dialysis patients are just not a homogenous group, and homogeneity of comparable groups is part and parcel of RCT selection and study.

There are now 60 or more years of worldwide accumulated experience in home hemodialysis, most of it outside the United States, with the past 15 years focusing on the two more recent iterations of home hemodialysis treatment—extended-hour and high-frequency (usually nocturnal) home hemodialysis, and short daily home hemodialysis. Carl Kjellstrand, a greatly respected savant of hemodialysis, compiled a lexicon of home hemodialysis publications appearing before 2000. This list, even then, exceeded 600 articles and abstracts; since then it has more than doubled. Although all these were observational and cohort studies, they were nonetheless studies by reputable experts in this field. Not one recorded a lesser outcome for home care. Nor have any since. Yet, given that these were criticized for selection bias and a range of other “shortcomings,” an RCT was demanded to settle the disquiet that these articles supporting home hemodialysis were somehow selling a lie.

Although I commend the authors of FHN(2) for their effort, a horrendous task gamely and honestly attempted, their fatally flawed study has seemingly swept aside this accumulated knowledge with one hopelessly underrecruited and underpowered RCT. FHN(2) suffered from all the recruitment biases it sought to exclude. Fewer than 33 percent of the subject numbers required by pretrial statistical analyses were recruited. Patients proved so resistant to randomization to center-based conventional care that the rules were altered in midtrial to permit the conventional arm to receive dialysis at home, where survival is twice that of in-center care (5). Even then, only 73 percent of those randomized to the home nocturnal arm performed their dialysis treatments as required by the trial design. Despite all this, the conclusion reported that extended-hour, high-frequency nocturnal home dialysis was no better than standard conventional hemodialysis.

A fairer conclusion might have been that despite a clear trend to near-significance in favor of home-extended hour and high-frequency dialysis from only a third of the predicted number of study participants, the trial was a design, recruitment, and statistical failure. A more honest conclusion should have ’fessed up to this. An excellent critique puts this brave, but failed, RCT into its proper perspective (6).

Do we need another RCT to resolve the issue? I contend, no. Ask yourself this question: has there ever been an RCT between transplantation and dialysis? No. If not, why not? It is simply because we know we can never fairly randomly assign patients to transplantation versus dialysis, any more than we can fairly randomize to one type of dialysis versus another. Yet, we universally acknowledge that “transplantation is the best therapy” and that “transplantation survival is better.”

Will yet another RCT of dialysis modality add anything further? No, it will not.

References

1. 

Anonymous. The National Cooperative Dialysis Study. Kidney Int Suppl 1983; 13:S1–S122.

2. 

Cheung AK, et al. Effects of hemodialyzer reuse on clearances of urea and beta2-microglobulin. The Hemodialysis (HEMO) Study Group. J Am Soc Nephrol 1999; 10:117–127.

3. 

Rocco MV, et al. The effects of frequent nocturnal home hemodialysis: The Frequent Hemodialysis Network Nocturnal Trial. Kidney Int 2011; 80:1080–1091.

4. 

Twardowski ZJ, et al. Con: Randomized controlled trials (RCT) have failed in the study of dialysis methods. Nephrol Dial Transplant 2013; 28:826–832.

5. 

Saner E, et al. Outcome of home haemodialysis patients: a case-cohort study. Nephrol Dial Transplant 2005; 20:604–610.

6. 

Kerr PG. Nocturnal hemodialysis—should we still provide this therapy?> Nat Rev Nephrol 2011; 7:611–612.