What's the punchline?

Daily, as opposed to alternate day, dialysis improves outcomes in critical illness

Verdict: 1/2 An important contribution to the critical care literature with some methodological flaws

The implications of this study are different from what you would expect looking at the title. Although the patients in the daily dialysis group were more frequently dialysed, this was less aggressive, in terms of fluid removal. The implications are:
1. In critical illness, acute renal failure is associated with poor outcome: the reason is unknown.
2. Critical illness is a catabolic state, with increased production of toxic by-products of metabolism, which accumulate in renal failure.
3. Daily dialysis is more effective for solute removal than (conventional) alternate day dialysis.
4. Hemodynamically, critically ill patients do not tolerate aggressive fluid removal. Patients dialysed daily require less fluid removal pre session, and this leads to less hypotension.
5. Patients dialysed daily  have lower mortality (ARR 18%, NNT 5.5), and significantly lower incidence of infection, respiratory failure and gi bleeds. It is unclear whether these outcomes relate to accumulation of toxic metabolites or transient hypotension in the alternate day dialysis group.
See critical appraisal below

Abstract

Background Intermittent hemodialysis is widely used as renal-replacement therapy in patients with acute renal failure, but an adequate dose has not been defined. We performed a prospective study to determine the effect of daily intermittent hemodialysis, as compared with conventional (alternate-day) intermittent hemodialysis, on survival among patients with acute renal failure.

Methods A total of 160 patients with acute renal failure were assigned to receive daily or conventional intermittent hemodialysis. Survival was the primary end point of the study. The duration of acute renal failure and the frequency of therapy-related complications were secondary end points.

Results The two study groups were similar with respect to age, sex, cause and severity of acute renal failure, medical or surgical intensive care setting, and the score on the Acute Physiology, Age, and Chronic Health Evaluation. Daily hemodialysis resulted in better control of uremia, fewer hypotensive episodes during hemodialysis, and more rapid resolution of acute renal failure (mean [±SD], 9±2 vs. 16±6 days; P=0.001) than did conventional hemodialysis. The mortality rate, according to the intention-to-treat analysis, was 28 percent for daily dialysis and 46 percent for alternate-day dialysis (P=0.01). In a multiple regression analysis, less frequent hemodialysis (on alternate days, as opposed to daily) was an independent risk factor for death.

Conclusions The high mortality rate among critically ill patients with acute renal failure who require renal-replacement therapy is related to both coexisting conditions and uremic damage to other organ systems. Intensive hemodialysis reduces mortality without increasing hemodynamically induced morbidity.

Critical Appraisal

Question: does daily dialysis improve outcome in critically ill patients with acute renal failure?
Setting: prospective controlled trial at single institution (multiple hospitals) in Germany between January 1993 and September 1998.
 

Methodology

160 patients with acute renal failure (ARF) requiring dialysis were alternately assigned to daily (80) or alternate day (80) intermittent hemodialysis. Two thirds of these patients were medical, one third surgical. Over 50% of the patients developed ARF as a consequence of hypotension. There is no information given about the use of CRRT (continuous renal replacement therapy) at the institution, leading to the possibility of selection bias: sicker patients may have received CRRT, and information about outcomes would have been useful.
The inclusion criteria, and evidence of renal failure requiring dialysis, were in keeping with standard practice. We assume that the patients were hemodynamically stable and suitable for IHD rather than CRRT, although this information is not given (e.g. how many were being treated with pressors on enrollment etc). This suggests selection bias: CRRT was available at the institution, and we do not know if sicker patients were dialysed with this mode as would be standard practice in European ICUs. Patients were excluded if they had chronic renal insufficiency, rapidly progressive glomerulonephritis, interstitial nephritis or post renal obstruction. Two patients with Goodpasteur's syndrome "sneaked" into the study, and were included in intention to treat analysis.
Randomization was by alternating order, and physicians caring for the patients were blinded until after the first dialysis session. 
The dose of hemodialysis was calculated with a urea kinetics model: none of the patients actually received the prescribed dose.
The primary outcome measure was mortality at 14 days after last session of dialysis. This is an unusual measure in critical care: usually (and controversially) it is 28 day mortality. We are not informed when this day occurred - a mean time with a range would have been helpful. We are told, however that renal failure resolved in the treatment group (daily dialysis - TG) at 9 +/- 2 days, and in the conventional group (alternate day dialysis - AG) at 16 +/-6 days.  Thus it would appear that the endpoint for each group was at significantly different timescales, and this may have led to bias (cumulative mortality in critically ill patients will always be higher at 30 days than at 23 days), and this is a flaw in the study.
Secondary endpoints were: presence or absence of sepsis, oliguria,&  hypotension.
Statistical analysis was performed using paired t tests and Fisher's exact test.

Results

All patients were accounted for: six patients were switched to CRRT and six underwent surgery. Data was analysed on intention to treat basis and data is presented including and excluding these patients.
Although patients were similar to baseline, daily dialysis was clearly superior to alternate day dialysis for creatinine clearance. Interestingly ultrafiltration volume (fluid removal) was lower (including cumulatively) in the daily dialysis group. Unsurprisingly, the alternate day removal of 3.5 litres of fluid (as opposed to the daily removal of 1.2 litres) was associated with significantly more hypotension (20% absolute increase in risk). The AG also had a significantly higher incidence of sepis/SIRS (24% absoulute increase), respiratory failure (34% absolute increase), GI bleeding (15% absolute increase).
The overall mortality rate was 37%, surprisingly high, 28% in the TG and 46% in the AG. The ARR (absolute risk reduction) was 18% with NNT (number needed to treat) of 5.5. Alternate day dialysis, sepsis and higher Apache III scores at randomization were associated with worse outcomes.

Impression

This study is a useful contribution to the critical care literature. This represents clear evidence that alternate day intermittent hemodialysis with aggressive fluid removal, leads to poor outcomes in critically ill patients. As to whether daily dialysis, which is associated with better solute removal, with less aggressive fluid removal, is a magic bullet, or less harmful is an interesting distinction. The mortality rates appear high by ICU standards, but are, in fact lower than most published series. The accompanying editorial cites the preponderance of medical ICU patients as the reason for lower mortality; medical ICU, universally, has higher mortality rates than surgical units. The more likely scenario is that sicker patients were preselected out of the study e.g. they received CRRT and were not enrolled.
The mortality endpoint is irritating. So what if patients in the AG had a lower mortality 14 days after end of dialysis? How many of them were alive at 28 days? How many of them were discharged home? How many of them were alive at 6 and 12 months? These would be more conventional mortality endpoints.
What really needs to be studied now is daily dialysis versus continuous veno-venous hemodiafiltration, which many of us would consider the goal standard in this patient population.

Overall I have learned three things from this study:
1. In consult based ICU systems it is essential to discuss dialysis plans with the nephrologists. Aggressive fluid removal is associated with worsened outcomes, and dialysis, for the most part should be for solute removal.
2. Dialysis should be performed more frequently than before, and it is probably no longer acceptable for patients to be dialysed Friday and then Monday "because we don't have enough staff at the weekend". This has enormous resource allocation implications.
3. The hypothesis touted in the CRRT literature about the importance of hemodynamic instability in critically ill patients undergoing dialysis is probably correct.

Reviewed by Patrick Neligan , University of Pennsylvania, March 2002