Effectiveness of blood transfusion in Malawi

Having been very impressed by Dr Fenton's study on the effectiveness of blood transfusions for Caesarean section in Malawi (Anaesthesia 1999; 54: 1055–8), I contacted him to discuss the results and enquire on specific concerns not answered in the original article. Dr Fenton was kind enough to share with me additional data and we would like to report further information on the effectiveness of blood transfusions in this context. The results reported originally are especially significant since they help resolve a crucial issue on which very little data are available, i.e. the effectiveness of allogeneic red cells to reduce mortality in haemorrhagic shock. However, from the original report in Anaesthesia, it was not possible to determine the proportion of mothers who survived/died while transfused or not in the Central vs. District areas. This may be important since transfusion practice appears to be quite different from one area of the country to another. Six thousand nine hundred and nine (6909) patients are now included in the database. Of these, 840 were deemed to require a transfusion and 558 were effectively transfused while 282 were not, owing to blood availability. Of the transfused mothers, 33 died (5.9%) and, of those who could not be transfused, 22 died (7.8%; p = 0.369 by Chi-squared). Thus, overall, transfusion of red cells does not appear to decrease mortality, as suggested in the original publication. However, the situation is very different in diverse parts of the country. Blood is more readily available in the District but much less readily available in the Central Hospital. In the District, 591 of the 4422 mothers (13.4%) were deemed to require blood and 432 (73%) were transfused. Of the transfused mothers, 25 died (5.8%) and, of those who could not be transfused, 18 died (11.3%), a statistically significant difference (p = 0.034 by Chi-squared). Hence, in the District, blood transfusion appeared to save lives, as we have been taught to expect. In the Central Hospital, 249 of the 2487 mothers (10%) were deemed to require blood and 126 (51%) were transfused. Of the transfused mothers, eight died (6.3%) and, of those who could not be transfused, only four died (3.2%; p = 0.399 by Chi-squared). Mortality in mothers deemed to require red blood cells, but not transfused, was lower in the Central Hospital compared with the District (3.2% vs. 11.3%, respectively; p = 0.022). Thus, in the Central Hospital, not only does blood transfusion not save lives, but it appears to worsen outcome. How can we explain this apparent paradox? As stated originally, fluid therapy is also very different in diverse parts of the country. Volume replacement with nonblood fluids is aggressive in the Central Hospital while the slow transfusion of blood is often the only volume administered in the District. The additional data presented here suggest that, globally, volume replacement is more important than red cell transfusion in mothers with anaemia or shock after Caesarean section. In mothers appropriately volume resuscitated, transfusions appear to be deleterious, a situation similar to that observed in industrialised countries where the benefits of transfusions are not as apparent as once thought [1]. These results are not those of a randomised clinical trial, but we must attempt to learn the most from the difficult African medical situation. Dr Fenton's observations are important for all of us, but particularly in Africa where blood may not always be readily available for fluid resuscitation and where, in parts of the continent, the safety of blood products is problematic (the incidence of endemic HIV may be as high as 20% [2–5]). In this context, we encourage aggressive nonblood fluid therapy for initial volume resuscitation. Erythrocyte transfusions should be considered later and only when specific indications for their administration exist.

Récupéré en direct depuis OpenAlex et désinversé. Les résumés ne sont pas conservés dans cette base de données : les index inversés représentent 8,6 Go des 9,3 Go de texte de la base, et le serveur dispose de 13 Go libres.