Criteria abstracted from The Users' Guide to Medical Literature, from the Health Information Research Unit and Clinical Epidemiology and Biostatistics, McMaster University

Highlighted lines and questions below provide links to the pertinent description of criteria in The EBM User's Guide, now available at the Canadian Centres for Health Evidence

Article Reviewed:

Please visit the new Evidence Based Journal Club Reviews

A multicenter, randomized, controlled clinical trial of transfusion requirements in critical care.

Hebert PC, Wells G, Blajchman MA, Marshall J et al., Transfusion Requirements in Critical Care Investigators, Canadian Critical Care Trials Group.

N Engl J Med 1999 340:409-17. abstract] With an accompanying editorial: Transfusions in Critically Ill Patients. Ely EW. New Engl J Med 1999; 340:467-8. [citation]

Reviewed by Fiona H. Levy MD, and Barry Markovitz MD, St. Louis Children's Hospital, Washington University School of Medicine

Review posted September 12, 1999

---

I. What is being studied?:

The study objective:

To determine whether a restrictive strategy of red-cell transfusion (maintaining hemoglobin between 7.0 and 9.0 g/dl) and a liberal strategy (hemoglobin between 10.0 and 12.0 g/dl) provided equivalent results in critically ill adult patients.

The study design:

Randomized, multicenter, unblinded, controlled clinical trial

The patients included:

838 critically ill patients admitted to tertiary (n=22) and community (n=3) ICUs in Canada between November 1994 and November 1997. The patients needed to be euvolemic after initial treatment, have a Hgb < 9.0 gm/dl and be enrolled within 72 hours of admission to the ICU.

The patients excluded:

Age < 16 yr., inability to receive blood products, chronic anemia, pregnancy, active blood loss, imminent death, brain death, post-op routine cardiac surgery. Patients with a question by the attending physician about withholding or withdrawing therapy were also excluded.

The patients excluded for reasons of consent were slightly older, but were similar in APACHE scores and diagnosis (except for cardiac disease).

The interventions compared:

A restrictive strategy of transfusion- using a Hgb < 7 gm/dl as the indication for blood transfusion - and maintenance of Hgb 7-9 gm/dl, was compared with a liberal transfusion strategy with parameters of 10-12 gm/dl.

The outcomes evaluated:

Rates of death from all causes at 30 days post randomization was the primary outcome measure. Secondary outcomes included 60 day death rates, ICU mortality, ICU and hospital length of stay, survival times in the first 30 days. In addition, severity of organ dysfunction as measured by rates of organ failure and the multiple-organ-dysfunction score.

II. Are the results of the study valid?

Primary questions:

1. Was the assignment of patients to treatments randomized?

Consecutive patients were randomly allocated to groups, stratified based upon degree of illness and institution.

2. Were all patients who entered the trial properly accounted for and attributed at its conclusion?

Was followup complete?

Nine patients failed to complete the study; 4 in the liberal group, 5 in the restrictive group, and reasons were not offered. Three patients were lost to follow-up at 6 months.

Were patients analyzed in the groups to which they were randomized?

There was a non-compliance rate of 4.3% (4.3% in the liberal group and 1.4% in the restrictive group) and an overall crossover rate of 1%. However, the final analysis was on an intention-to-treat basis.

Secondary questions:

3. Were patients, health workers, and study personnel "blind" to treatment?

No.

4. Were the groups similar at the start of the trial?

Yes. There were no differences between the groups in diagnoses, severity of illness, interventions, location before admission to the ICU, and oxygen delivery variables.

5. Aside from the experimental intervention, were the groups treated equally?

There were no differences between groups in the use of medications, vasoactive drugs, fluids, pulmonary artery catheter use, dialysis, mechanical ventilation, or surgical procedures. Such comparisons are extremely important since this study was not blinded; caregivers could have treated one group differently in other ways that might account for outcome differences. Relatively few measures are compared here to draw solid conclusions.

III. What were the results?

1. How large was the treatment effect?

First, the strategies were effective in meeting goals; the daily hemoglobin was 8.5 +/- 0.7 gm/dl vs. 10.7 +/- 0.7 gm/dl (p=0.01) between the two groups. In addition 33% of the restricted group required no transfusions (with an average of 2.6 +/- 4.1 units) while none of the liberal group went without transfusion and they had received an average of 5.6 +/- 5.3 units of blood (p= 0.01).

The rate of death from all causes in the 30 day period following ICU admission was 18.7% in the restrictive-strategy group, and 23.3% in the liberal-strategy group. This represents a relative risk of 0.8, or a relative risk reduction of 20%. The absolute difference is 4.7%. The overall hospital mortality rates were 22.2% and 28.1%, with a difference, favoring the restrictive-strategy, of 5.8%.

Kaplan-Meier survival curves overall showed no differences between the two groups, but subgroup analysis of patients with APACHE II scores of 20 or less (specified a priori) revealed a significant difference favoring the restrictive-strategy group (p=0.02). This is confirmed by a 7.4% difference in 30 day mortality (restrictive group 8.7%, liberal group 16.1%).

There was no difference between groups in the number of patients with multiple organ failure. The mean mulitple-organ-dysfunction score was somewhat lower in the restrictive-strategy group (8.3 +/- 4.6 vs. 8.8 +/- 4.4, p=0.10). Pulmonary edema and myocardial infarction were more common in the liberal-strategy group.

2. How precise was the estimate of the treatment effect?

The 95% confidence intervals for the 30 day mortality rate between the groups are -0.84 and 10.2. In other words, the true mortality rate (with 95% confidence) could be almost 1% worse or 10% better in the restrictive-strategy group. The 95% confidence intervals for the overall hospital mortality rate are -0.3 and 11.7, barely breaching the accepted statistical significance margin (p=0.05). The 95% confidence limits for the 30 day mortality rate in the subgroup of patients with APACHE II scores of 20 or less were 1.0 and 13.6%, suggesting a true improvement with the restrictive-strategy in less critically ill patients.

Of note, by the authors' sample size analysis, with an estimated combined 30 day mortality rate of 23%, and type 1 and 2 error rates of 5%, 1620 patients would have had to be enrolled to detect a 5.5% difference in rates.

IV. Will the results help me in caring for my patients?

1. Can the results be applied to my patient care?

Not directly. All the patients in this study were adults. It is interesting that the younger and less ill patients appeared to benefit most from the restrictive transfusion strategy, but the implications of this on the pediatric population are unclear. One could reasonably infer, however, that children, with potentially inherently more robust cardiorespiratory reserve, might also tolerate lower hemoglobin levels during critical illness, and therefore also reap the benefits that avoiding transfusions appears to confer.

2. Were all clinically important outcomes considered?

The most important outcomes were clearly measured, but there is no data to suggest how or why the liberally transfused patients apparently did worse. Although cardiac complications overall where higher in the liberal-strategy group, no such complications were higher in the 48 hour period prior to death in those who died.

Health care providers today are, in general, more restrictive in their use of blood transfusions than a decade ago, primarily for fear of infectious complications. There is no evidence in this study to support or refute this concern.

3. Are the likely treatment benefits worth the potential harms and costs?

This study lends further support to the current trend of a more restrictive-transfusion approach. There does not appear to be any additional cost or harm from lowering the "transfusion trigger," although costs and long-term risks were not addressed in this investigation.

Comments

Back to the EB Journal Club Index