The first randomised controlled trial in dialysis

In 1980 the NCDS led to the first minimum standards for dialysis
The National Cooperative Dialysis Study (NCDS) reported in March 1980. It led to the widespread acceptance of minimum standards for ‘dialysis dose’ worldwide. However few people have read the full study. It is hard to find in libraries, and you can’t get it online.

Dialysis in the early 1980s
In the 1970s, ‘Generally, any change in a patient’s treatment was influenced more by the need to remove fluid than it was by the need to remove metabolic waste’ wrote Ed Lowrie in his introduction to the NCDS report. The trial was designed to investigate whether the removal of waste products could be usefully measured. Urea clearance was chosen as the method to test.
Clearance by dialysis was not routinely estimated at the time. A standard prescription was usual. In 1967, the Charing Cross Hospital regimen was described as 10 hours 3 times weekly at home, or 14h twice weekly overnight in hospital. Times shortened as dialysers became more efficient, and as pressure of patient numbers encouraged higher throughput.
The NCDS was a collaborative randomized controlled trial (RCT) between 9 renal units, a rare event in nephrology in the 1970s, and new to dialysis. The patients’ mean age was 49, and they had been on dialysis for an average of 4.2 years. Half were white, there were no diabetics, many were taking androgens in attempt to raise haemoglobin in the pre-erythropoietin era, and in these ways the population was probably fairly typical of the time.
151 patients were randomised to two different average urea concentrations (adjusted by varying dialyser size and flow rates), and to two different durations of dialysis (2.5-3.5h and 4.5-5h three times weekly), in all 4 possible combinations, for 6-12 months.
Patients in the groups with higher average ureas (predialysis urea 38 versus 26 mmol/l (107 versus 72 mg/dl)) were more likely to be withdrawn from the study or admitted to hospital, and this led to the study being stopped in 1980. There was no significant mortality difference – in fact there were only 3 deaths in the study, but two of these were in the high-urea, short-dialysis group, which also had the highest rate of cardiovascular events (44%) and hospitalisation. However at the end of the (prematurely finished) study, the effects of increased urea removal looked greater than the effects of longer dialysis, and only the urea difference reached conventional levels of statistical significance. The long vs. short dialysis results were suggestive, but were dismissed as not significant in the main report, and mostly overlooked in subsequent analysis.
Proportion of patients who had no hospital admisssions with time during the NCDS. I = higher Kt/V, longer dialysis; II = lower Kt/V, longer dialysis; III = higher Kt/V, shorter dialysis; IV = lower Kt/V, shorter dialysis. From Parker et al, fig 5, full NCDS report.
The major publication from the NCDS focused on the analysis of urea clearance, and it has become one of nephrology’s classic papers. The follow-on (1985) publication by Gotch and Sargent introduced the measurement of Kt/V, and this used together with the NCDS headline results have been very influential. An equilibrated Kt/V of 1 was their recommended minimum. It is equivalent to a single-pool Kt/V, the usual one we measure, of about 1.21. Kt/V remains a central element haemodialysis in guidelines 30 years later.
However targeting a desired Kt/V had some unintended consequences. Increasingly efficient dialysers made it possible to reduce treatment times dramatically, likely to lead to cumulative fluid overload and cardiovascular complications. In that way overinterpretation of the NCDS may have led to some avoidable deaths.
By modern standards, the NCDS was very small, certainly too small to be confident that shorter treatments were safe. Further studies were certainly needed, but the NCDS was pioneering and important.
Dialysis dose has been tested further. After the NCDS helped set a minimum Kt/V, it was noted in many observational (non-randomized) studies that patients with higher Kt/V had better survival. People still believed that this was a causative relationship until the HEMO study of 1846 patients (more than 10 times the size of the NCDS) published in 2002 definitively showed that increasing single pool Kt/V from 1.3 to 1.7 gave no extra benefit in patients on fixed duration thrice weekly dialysis. (Equilibrated Kt/V rose from 1.16 to 1.53.)
Treatment time is receiving much greater attention again but we have still not had an adequate RCT comparing different dialysis durations or frequency. Nearly everyone believes more frequent or longer treatments are better, but these are expensive and difficult for patients, and as the HEMO study showed, we could be wrong. We really need a trial.
de Wardener HE, 1967. The Kidney. London, Churchill. The classic early nephrology textbook.
Lowrie EG et al, 1981. Effect of the hemodialysis prescription on patient morbidity. N Engl J Med 305:1176-81
Lowrie EG, NM Laird (eds), 1983. The Cooperative Dialysis Study. The full report was Supplement 13 to Kidney International
Gotch FA, JA Sargent, 1985. A mechanistic analysis to the National Cooperative Dialysis Study. Kidney Int 28:526-34
Eknoyan G et al, 2002 Effect of dialysis dose and membrane flux in maintenance hemodialysis. N Engl J Med 347:2010-9
This article will also be published in the Journal of Renal Nursing in March 2010.
, ,

One response to “The first randomised controlled trial in dialysis”

  1. Correction made 19 Mar 2010 – groups II and III were transposed in the figure legend. The top two lines are the lower average urea (high Kt/V) groups, with the longer dialysis group doing better. The lower two lines are the higher average urea groups (low Kt/V), with the short and less dialysis group faring worst.

Leave a Reply

Your email address will not be published. Required fields are marked *