THERAPY

Criteria abstracted from The Users' Guides to the Medical Literature series in JAMA

N Engl J Med 2007 356:1609-1619.[abstract]

Review posted December 18, 2007

1. What is being studied?:
   
   
   1. The study objective:
      
      The authors of this paper began by highlighting the point that transfusion strategies in pediatric ICU’s have a high degree of variability. The objective of this study was to determine if a restrictive strategy, using a lower threshold for transfusion, was as safe as a liberal strategy for stable patients in the pediatric ICU.
      
      
   2. The study design:
      
      This was a prospective, randomized, controlled, non-inferiority trial of critically ill, yet stable children in pediatric Intensive Care Units.
      
      
   3. The patients included:
      
      Patients were eligible for this study if, during their first 7 days of admission to an ICU, they had a hemoglobin level less than or equal to 9.5 g/dl. 5,399 patients met these criteria and were evaluated for this study.
      
      
   4. The patients excluded:
      
      Patients were excluded if the expected length of stay was less than 24 hours, predicted survival of less than 24 hours, a decision to withdraw or withhold critical care, there was no physician approval, less than 3 days old, greater than 14 years old, weight less than 3 kilograms, unstable hemodynamics, acute blood loss, cardiovascular problems, never discharged from the neonatal ICU, postconceptional age less than 40 weeks, had hemolytic anemia, severe thrombocytopenia, hypoxemia, previous enrollment into this and other studies, brain death, ECMO, hemofiltration, blood exchange transfusion, plasmapheresis, an inability to receive blood products, and pregnancy. A total of 4.372 patients were excluded based on these criteria. Of note, hemodynamic stability was defined as mean arterial pressure not less than 2 SD below the normal mean for age, and no changes in cardiovascular treatments for at least two hours prior to enrollment.
      
      
   5. The interventions compared:
      
      The study compared a liberal versus a restrictive transfusion strategy. In the restrictive strategy limb, the threshold to transfuse was a hemoglobin of 7 g/dl with a target range of 8.5-9.5 g/dl. In the liberal strategy limb, the threshold to transfuse was 9.5 g/dl with a target range of 11-12 g/dl. Only prestorage leukocyte reduced red blood cell units were used in this study. Each protocol was applied for up to 28 days of the ICU stay, or until death.
      
      
   6. The outcomes evaluated:
      
      The primary outcome this study was powered to measure was the development or progression of multiple organ dysfunction syndrome (MODS). In the Methods section the authors also mentioned that mortality during the 28 days after randomization was also a primary outcome, however, because of the overall low mortality rate among children, this study was not powered to measure this outcome. In the results mortality was reported as a secondary outcome. Other outcomes evaluated were: the Pediatric Logistic Organ Dysfunction (PELOD) score, sepsis, transfusion reactions, nosocomial respiratory infections, catheter related infections, adverse events, and length of ICU stay.
      
      
2. Are the results of the study valid?

Primary questions:

1. Was the assignment of patients to treatments randomized?
   
   Yes. Patients were randomized to each study group in blocks of two or four, which were randomly distributed and stratified according to age. The three age groups, into which the patients were stratified, were less than 28 days of age, 28 to 364 days of age, and greater than 365 days of age.
   
   
2. Were all patients who entered the trial properly accounted for and attributed at its conclusion?
   
   Was the follow-up complete?
   
   Follow up was complete in this study. Of the 626 patients in the study only 11 or 2% were lost to follow up. As the authors pointed out, this rate of loss was small enough to prevent any bias from sample size slippage.
   
   Were patients analyzed in the groups to which they were randomized?
   
   Yes. Analysis of the primary and secondary outcomes were based on an intention-to-treat analysis. The intention to treat analysis is designed to measure the original intention of the treatment protocol. Recognizing that in “real life� patients are not always in complete compliance with treatment protocols, this analysis measures the intention of the treatment protocol, but not necessarily the effects of the therapy itself. The authors of this study anticipated that some patients would be unable to fully comply with the protocol, leaving it up to the judgment of the attending physicians to briefly suspend the protocol during periods of instability. In this study adherence was defined as the proportion of days the hemoglobin concentration was kept above the threshold. Patients were considered adherent if the hemoglobin level was kept above the specific protocols threshold for at least 80% of the time. The intention-to-treat analysis accounted for all the patients in each protocol regardless of adherence. In order to preclude missing any significant differences between the transfusion strategies, the authors also conducted a per protocol analysis. The per protocol analysis is designed to measure the true efficacy of the therapy, and only takes into account patients who met the adherence criteria.

Secondary questions:

3. Were patients, health workers, and study personnel "blind" to treatment?
   
   No. Due to the nature of this study model, the healthcare workers involved in the treatment of the patients could not be blinded.
   
   
4. Were the groups similar at the start of the trial?
   
   Yes. The patient groups were similar in terms of patient demographic, severity of illness (PRISM score), PELOD score, septic state, MODS involvement, hemoglobin level and prior transfusion exposure at the start of the trial.
   
   
5. Aside from the experimental intervention, were the groups treated equally?
   
   The attending physicians followed the protocol outlined for each study group. Adherence to the protocol was defined as the proportion of days that the hemoglobin concentration was kept above the threshold. Nearly 99% of the patients met the 80% adherence criteria, and in most centers the adherence to protocol rates exceeded 97%. The investigators did not implement any treatment protocols to control other therapeutic interventions. The potential confounders include ventilator management strategies, glycemic control, nutritional support and goal-directed therapy. They also did not mention therapies or interventions that could have affected the hemoglobin levels of the patients. For example, there was no mention of patients treated with erythropoietin or iron, how often the groups received blood draws, or how often the groups underwent invasive procedures not qualifying as surgery i.e. central line placement.

3. What were the results?
   
   
   1. How large was the treatment effect?
      
      The treatment effect, based on the results of the intention to treat analysis, was not statistically significant. The absolute risk reduction for developing new or progressive MODS was only 0.4% (95%CI, -4.6 to 5.5). When new or progressive MODS were stratified according to age, country and severity of illness there was still no significant reduction in the risk for development or progression of MODS. The results of the per-protocol analysis were quite similar to the results of the intention-to-treat analysis. The absolute risk reduction in the per-protocol analysis was 0.8% (95% CI, -4.3 to 5.9).
      
      The authors also conducted a time to event analysis using Kaplan-Meier curves and the Logrank test. Time to event data are used in survival analysis and when the outcome of interest is the time to a certain event, in this case the onset or progression of MODS (1). A hazard ratio was generated from the time to event analysis. The hazard ratio, which is essentially an estimate of the relative risk, demonstrated a ratio of 0.95 for the restrictive transfusion strategy. A hazard ratio of 0.95 suggests that there was no increase in risk for the onset or progression of MODS when using the restrictive strategy, compared to the liberal strategy.
      
      The greatest absolute risk reduction was evident when the two strategies were compared against new or progressive MODS in the patients for whom the protocol was suspended. The authors suggested that the protocol was suspended for more patients in the restrictive strategy group, because of the uneasiness of attending physicians in keeping the hemoglobin so low in critically ill children. The absolute risk reduction in this group was 18.9 (95%CI, -7.3 to 45), however, this was not statistically significant.
      
      
   2. How precise was the estimate of the treatment effect?
      
      One of the strongest points of this study was the number of patients enrolled and the subsequent precision of the confidence interval. The authors hypothesized that a restrictive strategy would not be inferior to a liberal strategy if the limits of the 95% confidence intervals did not exceed 10%. As stated above, the absolute risk reduction for the primary outcome using the restrictive strategy was only 0.4%, and using the 95% confidence interval the range was -4.6% to5.5%. When new or progressive MODS was stratified according to age, country, and severity of illness, only the patients with PRISM scores in the highest quartile had and upper limit of the 95% confidence interval that exceeded 10% (-11.1%-15.9%). This particular group exceeded the threshold of 10% set by the authors. This data suggests that the absolute risk reduction of new or progressive MODS in the most severely ill patients could be as much as 11.1% or an absolute risk increase of 15.9% with 95% confidence. There were no other confidence intervals whose upper limits exceeded 10%.
      
      Using the restrictive protocol, the number-needed-to-treat to prevent one RBC transfusion was only 2. The number of red blood cell units per patient in the restrictive group was 0.9, and in the liberal group was 1.7 (p<0.001). When comparing the two strategies there was a 96% reduction in the number of patients who had any transfusion exposure and a 44% decrease in the number of transfusions administered in the restrictive strategy.
      
      
4. Will the results help me in caring for my patients?
   
   
   1. Can the results be applied to my patient care?
      
      According to the protocol set in place at our institution the justification for transfusion is set at hemoglobin of 8 g/dl. The authors did well in powering this study to the designated end points, sufficiently randomizing the study patients and ensuring proper follow up. Based on the results of this study, I believe that it is safe to use a transfusion threshold of 7 g/dl for stable children in the PICU, with respect to the designated outcomes. Considering that the confidence interval did range from 11.1% to 15.9% in patients with PRISM scores >8, we may need to be cautious in applying this data to this group of patients. An absolute risk increase of nearly 16% could be significant in patients with highest quartile PRISM scores.
      
      Decreasing the transfusion threshold from 9.5 g/dl to 7 g/dl significantly decreased the number of transfusions, and the number-needed-to-treat to reduce one transfusion was only two patients. It has been described in the literature, that blood transfusions can be independent risk factors for nosocomial blood stream infections. One study, a prospective cohort study performed on over two thousand adult ICU patients, showed that red blood cell transfusions were a significant risk factor for nosocomial blood stream infections even when corrected for survival probability. Using leuko-reduced red blood cells decreased the risk of infection slightly but not significantly(2). Another prospective cohort study, performed on over two thousand pediatric ICU patients, showed that a higher number of red blood cell transfusions were associated with an increased risk of blood stream infections (3). By using a lower transfusion threshold we can decrease the number of blood transfusion and potentially decrease the number of blood stream infections.
      
      It was not the purpose of this study to look at safe transfusion thresholds in unstable, critically ill children. Understandably, it would be nearly impossible to determine what the safe transfusion threshold is for unstable PICU patients. The pathophysiologic causes of instability in our patients are too numerous. However, by only looking at stable children, a great number of our patients are excluded by this study. Of the 5,399 patients eligible for the study, 81% were excluded prior to the screening process because of the multiple exclusion criteria. Much the same, when these exclusion criteria are applied to our average patient population many would not qualify. We are then left with other guidelines, such as early goal directed therapy (4), and the clinical practice parameters for the support of pediatric patients in septic shock (5), which have significantly higher transfusion thresholds. On the other hand, over time as our patients progress to a more stable condition we can institute a restrictive transfusion strategy.
      
      
   2. Were all clinically important outcomes considered?
      
      The authors were careful to consider all of the potential short-term outcomes, but it did not look at any of the potential long-term outcomes. The study by Jonas, Wypij, and Roth et al. (6), which connected a lower hematocrit during cardiopulmonary bypass to poorer neurodevolpmental outcomes, should at least cause us to question how a lower hematocrit in a stressed critically ill child affects long-term neurodevelopment.
      
      
   3. Are the likely treatment benefits worth the potential harms and costs?
      
      This study succeeded in determining that a restrictive transfusion strategy is as safe as a liberal transfusion strategy. To then recommend that we adopt a universal restrictive strategy in stable critically ill children, presupposes that there is a benefit to using fewer blood transfusions. My teaching has always stressed that we should not transfuse patients unnecessarily because of the risk of transfusion related infections, transfusion reactions, and the unnecessary use of a limited blood supply. With the institution of leukocyte-reduced red blood cells, complications with red blood cell transfusions may have been reduced, however, they still exist (7).
      
      Considering the literature, it appears that red blood cell transfusions may be an independent risk factor for nosocomial blood stream infections (2,3). Even though this study could not show a significant difference in the number of transfusion reactions, there were certainly more in the liberal strategy group than in the restrictive strategy group (2% vs. 1%). If this study could have been powered to look at such an end point, it is likely that there would have been a significant difference, given the substantial reduction in the total number of transfusions. In view of the current evidence, it appears that the institution of a restrictive transfusion strategy in stable critically ill children is worth the potential harms and costs.

References

1. Altman DG, Bland JM. Time to event (survival) data. BMJ 1998;317:468-469.
   
   
2. Taylor RW, Obrien J, Veramakis C, et al. Red blood cell transfusions and nosocomial infections in critically ill patients. Crit Care Med. 2006 Sep;34(9):2302-8.
   
   
3. Elward AM, Fraser VJ. Risk factors for nosocomial primary bloodstream infection in pediatric intensive care unit patients: A 2-year prospective cohort study. Infect Control Hosp Epidemiol. 2006 Jun;27(6):553-60
   
   
4. Rivers E, Nguyen B, Tomlanovich M, et al. Early goal directed therapy in the treatment of severe sepsis and septic shock. N Engl J Med. 2001 Nov 8;345(19):1368-77.
   
   
5. Carcillo JA, Fields AI, et al. Clinical practice parameters for hemodynamic support of pediatric and neonatal patients in septic shock. Crit Care Med 2002 June;30(6):1365-1378.
   
   
6. Jonas RA, Wypjj D, Roth SJ, et al. The influence of hemodilution on outcome after hypothermic cardiopulmonary bypass: results of a randomized trail of infants. J Thoracic Cardiovasc Surg 2003;126:1765-74.
   
   
7. Hebert PC, Fergusson D, Sher GD, et al. Clinical outcomes following institution of the Canadian universal leukoreduction program for red blood cell transfusions. JAMA. 2003 Apr 16;289(15):1941-9.

Last Updated: December 18, 2007

Fill out our Feedback Form | Editorial Notes | About Us