To determine if infants with very low birth weight who receive packed red blood cell (PRBC) transfusions have increased odds of developing necrotizing enterocolitis (NEC), to determine the rate of NEC after PRBC transfusion, and to characterize the blood transfused preceding the onset of NEC.
A retrospective cohort design was used. The study population included infants with a birth weight of <1500 g who were from a single center. NEC after transfusion was defined as NEC that occurred in the 48 hours after initiation of PRBC transfusion. Statistical analysis included unadjusted and multivariable analyses.
The study sample included 2311 infants. A total of 122 infants (5.3%) developed NEC, and 33 (27%) of 122 NEC cases occurred after transfusion. NEC occurred after 33 (1.4%) of 2315 total transfusions. Infants who received a transfusion had increased adjusted odds (odds ratio: 2.3 [95% confidence interval: 1.2–4.2]) of developing NEC compared with infants who did not receive a transfusion. PRBCs transfused before NEC were predominantly (83%) from male donors and were a median of 5 days old.
In our study sample, PRBC transfusion was associated with increased odds of NEC. The rate of NEC after transfusion was 1.4%. From our data we could not determine if PRBC transfusions were part of the causal pathway for NEC or were indicative of other factors that may be causal for NEC.
We have read with great interest the article by Paul et al about the association between necrotizing enterocolitis (NEC) and transfusion of red blood cells in premature infants, published in Pediatrics in April (1). However, we have identified several issues in the study that preclude accepting the authors' conclusions.
First, the authors mention that they conducted a retrospective cohort study. However, the findings are presented as if it were a case control study. The data shown in most of the tables compare the group of infants with NEC with those who did not have this outcome. The appropriate form of displaying the findings in cohort studies is by grouping the participants regarding the exposure (receiving or not transfusion of red blood cells, in this case) rather than by the occurrence of the endpoint. As presented, the tables do not allow readers to assess the comparability of infants exposed and non-exposed to the putative risk factor regarding other variables that could also affect the outcome (2,3). Therefore, it is not possible to judge if the variables included in the logistic regression models were appropriate.
Secondly, the figures presented in the tables make it not possible to replicate the unadjusted odds ratio (OR) presented by the authors. In fact, there are several inconsistencies between the data as presented in the tables and the text. According with Table 3, 91% of the 122 infants with NEC received transfusions (that is, 111 neonates). However, the text states that "by contrast 24% (30 of 122) of total NEC cases occurred in infants who never received a transfusion". This would mean that 76% of the infants with NEC had received a transfusion, instead of 91%, as presented in the table. But regardless of this, none of the 2 by 2 tables that could be assembled with these sets of inconsistent data provide the unadjusted OR of 8.9 depicted in Table 4 of the article. If one accepts that 91% and 48% of infants with NEC and no NEC received transfusions, respectively, the unadjusted OR would be 10.9 (95% CI 5.8 - 20.4). On the other hand, the assumption that 30 of the 122 infants with NEC never received transfusions and that these were given to 48% of the infants without NEC gives an unadjusted OR of 3.3 (95% CI 2.2 - 5.1). The use of the numbers for transfusions excluding those given after the diagnoses of NEC, also presented in Table 3, do not provide the OR presented by the authors either. In conclusion, it was not possible for us to reproduce the estimate of the increased odds of NEC after transfusion reported in the article.
All these facts cast a shadow of doubt over the findings of this study and, in fact, lead us to the conclusion the authors themselves provide: "From our data we could not determine if PRBC transfusions were part of the causal pathway for NEC or were indicative of other factors that may be causal for NEC." We agree, but not exactly for the same reasons.
1. Paul DA, Mackley A, Novitsky A, Zhao Y, Brooks A, Locke RG. Increased Odds of Necrotizing Enterocolitis after Transfusion of Red Blood Cells in Premature Infants. Pediatrics 2011;127:635-641 2. Levine M, Ioannidis J, Haines T, Guyatt G. Chapter 12. Harm (observational studies). In: Guyatt G, Rennie D, Meade MO, Cook DJ, eds. Users' Guides to the Medical Literature: A Manual for Evidence-Based Clinical Practice. 2nd ed. New York, NY: McGraw-Hill; 2008 3. Straus SE, Glasziou P, Richardson WS, Haynes RB. Evidence-Based Medicine. How to practice and teach EBM. Edinburg: Elsevier Churchill Livingstone, Fourth Edition, 2011
Conflict of Interest:
None declared
I read this article with interest. From earlier I´ve been convinced that there is a relationship between anemia and NEC, possibly through intestinal ischemia. The infants that recieve RBC transfusion can be expected (at least as a group) to be more anemic than the referance group. Therefor the most likely cause for increase in NEC in this group is not the transfusion per se, but rather the anemia, I would think. I can´t see to find a discussion of this possibility in the article.
Conflict of Interest:
None declared