Introduction

Thrombocytopaenia is highly prevalent in neonatology affecting around 25% (22–35%) of all neonates admitted to neonatal intensive care units.1 Clinical signs of bleeding are also commonly documented in preterm neonates. But the close temporal association commonly noted in sick babies between low platelet counts and the occurrence of bleeding does not establish cause and effect. Thrombocytopaenia is a risk factor for poorer neonatal outcomes, although it is unclear whether it is largely a marker of severity of illness and comorbidity.2 Platelet transfusion remains the only readily available specific treatment for this condition. Decisions about when to treat thrombocytopaenia are therefore linked to defining safe and effective platelet transfusion practices.

Platelet transfusion practice

Policies for neonatal platelet transfusion therapy vary widely between clinicians, institutions and countries, and are inevitably based on specified threshold counts of platelets, although platelet counts provide no information on changes in platelet function. Alternative criterion of the need for platelet transfusions, such as platelet mass (which is based on a sum of platelet count and platelet volume, based on the rationale that larger platelets may be more effective haemostatically), have been proposed, but larger studies to assess clinical outcomes have not been undertaken.3

A large web based survey of neonatologists in Canada and USA reported significant variation between neonates units and indicated that platelet transfusions were frequently administered to non-bleeding neonates with platelet counts >50×10⁹/l.4 In the UK, a telephone survey of all tertiary level neonatal units demonstrated similar variation in practice but with the most common thresholds for transfusion in well or stable term and preterm infants being 25×10⁹ and 30×10⁹/l, respectively.5

Bleeding and thrombocytopaenia

It should be stressed that the safety and efficacy of these platelet count thresholds for prophylactic platelet transfusions have never been assessed in randomised trials.6 Murray …