OK, not exactly this month, but “towards the end of last year” wouldn’t have been as good a title.
Three case-control studies, with some differences in design and implementation, but all 3 with the same result, no evidence that red cell transfusion causes NEC. So you could say that in Stanford, Gainesville Florida, and Riyadh transfusions don’t trigger NEC but in some other places they do, and in still other places they may even be protective.
Wallenstein MB, et al. Red blood cell transfusion is not associated with necrotizing enterocolitis: a review of consecutive transfusions in a tertiary neonatal intensive care unit. The Journal of pediatrics. 2014;165(4):678-82.
I think the most appropriate answer is “unproven”, I also think its great that these 3 negative case-control studies were published. There is probably less publication bias in neonatology for RCTs, although publishing negative trials can sometimes be a challenge, and sometimes doesn’t occur. Publication bias of observational studies is, I think a major problem, so kudos to these authors and journals. Even “completion bias”, a new term I just invented, is a real problem. If I run through a local database, and find no apparent association with NEC in my NICU, am I going to take the time and effort required to complete an IRB approved research project? Probably not, but if my first glance at the data looks positive, well then I might get a good publication out of it, and I’m much more likely to try to complete it.
I think that especially the first reports of an association, such as the association between NEC and blood transfusions, have a high likelihood of being biased, just because of such a completion bias. What I mean is that if someone has seen 3 babies in a row get NEC after a blood transfusion, they then might think, I wonder if blood transfusions might trigger NEC, and then search their local data and find that there are more episodes of NEC after transfusion, there is a high likelihood that it might just be due to the fact that they are searching a database that includes the cases that they have just seen.
Replications of that investigation by others would not have the same detection bias, but might well have the “completion bias” that I just described.
Many of the observational study articles have suggested that we need large randomized trials to be sure about whether this is a real effect or not, while I agree, I don’t think a trial of blood transfusion in at-risk neonates should be done just to address this issue. A randomized trial of differing threshold of transfusion would have other outcomes to investigate also, but an important secondary outcome to be prespecified should be NEC.
In the PINT trial, we did give fewer transfusions to one group compared to the other, an average of 4.9 in the low transfusion group, and 5.7 in the high group. The low group actually had more NEC 8.5% vs 5.3% (not statistically significant).
I don’t know if any future trial is likely to get more separation in hemoglobins, or in transfusion frequency than PINT, but powering such a trial to see if there is a difference in NEC would mean that the trial would have to be very large, and very expensive. Making large trials simpler, and less expensive is a major priority for the future.