An otherwise well 3 week old infant born at 28 weeks gestation has a haemoglobin level of 68 g/l and is prescribed a blood transfusion. The departmental protocol states feeds should be withheld during the transfusion to decrease the risk of development of necrotising enterocolitis (NEC). What is the evidence that blood transfusion increases the risk of NEC?

Structured clinical question

In a preterm infant [patient] does blood transfusion [intervention] increase the risk of NEC [outcome]?

Search strategy and outcome

Search words: “transfusion” AND “necrotizing enterocolitis” (excluding exchange transfusion).

Secondary sources—Cochrane Library (Issue 3, 2003): no relevant systematic review.

Pub Med (1975–2003). Limits: newborn. Search outcome: 85 articles, of which two were relevant.

Pub Med (1975–2003) using clinical queries with methodology filters (category aetiology, emphasis: sensitivity). Limits: newborn. Search outcome: 34 articles, of which one was relevant (already retrieved by Pub Med).

Embase (1974–2003). Search outcome: 111 articles, of which one relevant (already retrieved by Pub Med).

Cinahl (1982–2004). Search outcome: 15 articles, none relevant.

Sum Search. Search outcome: 29 articles, none relevant.

See table 2.

Commentary

In the two reported studies,^1,2 the indications for transfusion were not standardised, the time interval between transfusion and NEC was not available, and any transfusion at any time between birth and NEC was analysed.

The results of the ecological study¹ are difficult to interpret as the association found between transfusion and NEC was at the level of the NICU but was not studied at the individual neonate level.

Bias in the published results of the two studies is possible, as the findings may be related to other practices in the specific neonatal intensive care unit (e.g. restricted transfusion policy). It may also reflect confounding by the indication for transfusion (e.g, infants who have NEC may require more transfusions). It could also be that the anaemia for which a blood transfusion was requested was an independent risk factor for NEC, or an early manifestation of NEC still developing, which then becomes recognised several hours later (during or after the transfusion).

While anecdotal reports suggest that NEC has developed quickly after a blood transfusion, such information is not available in published studies. However, neonatal exchange transfusion^3,4 and intrauterine transfusion,⁵ both via umbilical vessels, have been shown to be associated with an increased incidence of NEC.

Further studies minimising bias and confounding are needed to prove or disprove an association between blood transfusion and the risk of NEC, but even then, association is not necessarily synonymous with causality. It should be possible to undertake randomised controlled studies on the effect of withholding feeds versus feeding during blood transfusions on the rate of NEC, although blinding would be impossible and the sample size required for adequate power would likely be extremely large.