Mortality associated with restrictive threshold for red blood cell transfusion in pediatric patients with sepsis

The authors reply: As mortality is low in pediatric intensive care patients, other outcome measures have been developed. Because a surrogate outcome is defined as “a variable that provides an indirect measurement of effect in situations where direct measurement of clinical effect is not feasible,” (1) most would agree that organ dysfunction scores could be a valid surrogate for death. In our study (2), the primary outcome was the development or progression of a multiple organ dysfunction syndrome, as defined by Proulx et al (3). We also measured other multiple daily morbidity outcomes, such as the Pediatric Logistic Organ Dysfunction Score (4), PaO2/FIO2 ratio, blood lactate level, nosocomial infections, duration of mechanical ventilation, pediatric intensive care unit length of stay, pediatric intensive care unit mortality rate, and mortality at 28 days. None of these outcome measures differed significantly between the patients in the two groups. Among the 137 patients, the primary outcome, new or progressive multiple organ dysfunction syndrome, was 13 (18.8%) vs. 13 (19.1%) (p = .97) for patients in the restrictive and liberal transfusion groups, respectively. The 28-day mortality was 7 (10%) vs. 2 (3%) (p = .08). Because death can be viewed as a dysfunction of all organs, it was also measured in the primary outcome. As both the primary outcome and all the secondary outcomes were so similar in both groups, we believe that observation of this nonsignificant difference in mortality at 28 days is nothing more than statistical artifact, as no other variable correlates with the observed mortality. This is why we wrote in our manuscript: “The number of deaths during [pediatric intensive care unit] stay and at 28 days were not significantly different between both groups, but these results should be interpreted cautiously, as the number of deaths was low. Nevertheless, no other secondary outcome measuring morbidity was different between the two groups, which further supports the findings with regard to mortality.” (2) Furthermore, mortality is a secondary outcome in a secondary analysis. Therefore, we would advise against overinterpreting this outcome (5). Indeed, the sample size of the main trial was based on the main outcome and not on a rare secondary outcome. All subgroup analyses suffer de facto from a lack of power. This is why we wrote “the most important limitation of our study are the pitfalls inherent to any subgroup analysis which preclude the possibility of generating definitive conclusions and at best allow for hypothesis generation only. Hence, no definitive recommendations regarding transfusion thresholds in stabilized septic children can be made; it is nonetheless striking that the frequency of the primary outcome in the two transfusion groups was very similar.” (2). We agree that absence of evidence is not evidence of absence and that our sample size does not allow us to demonstrate the noninferiority of a restrictive strategy, but it might seem misleading to compute a post hoc sample size and extrapolate a hypothetically increased mortality in such a way (5). Therefore, as both the main study (6) and all three subgroup analyses (2,7,8) do not find evidence that a restrictive red cell transfusion strategy, when compared to a liberal one, increase the rate of morbidity, we maintain our suggestion that a hemoglobin level of 7/dL is safe in stabilized critically ill children. The authors have not disclosed any potential conflicts of interest. Oliver Karam, MD, MSc, Marisa Tucci, MD, BSc, Thierry Ducruet, MSc, Heather Ann Hume, MD, Jacques Lacroix, MD, France Gauvin, MD, MSc, CHU Sainte-Justine, Montreal, Quebec, Canada

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.