The Threshold for Platelets study: a prospective randomised trial to define the platelet count which critically ill patients should receive a platelet transfusion prior to an invasive procedure

Prophylactic platelet transfusions are frequently administered to intensive care unit (ICU) patients with low platelet counts to reduce bleeding risk before they undergo a procedure. However, evidence to support this practice is poor and has been identified as a high research priority by Australian and international transfusion organisations. It is unclear whether pre-procedure prophylactic platelet transfusions reduce bleeding or are otherwise beneficial to ICU patients, and there is evidence that platelet transfusions can cause patient harm. In addition, platelets collected for transfusion are expensive and in limited supply. Clinicians primarily use a low platelet count to decide whether to administer a prophylactic platelet transfusion. However, the uncertainty around the benefits and risks of prophylactic transfusions has led to variation in clinical practice in the platelet count threshold at which clinicians give a pre-procedure prophylactic transfusion.

The T4P trial is a novel, randomised Bayesian adaptive trial in ICU patients with a low platelet count who need an invasive procedure. Patients will be randomly allocated to one of five platelet transfusion thresholds. The primary outcome is 90-day all-cause mortality. It aims to identify the optimum platelet count for clinicians to use as the basis of a decision to give or not give a pre-procedural prophylactic platelet transfusion to an ICU patient. It will model the platelet threshold-response curve, evaluate whether the threshold varies according to patient characteristics, and evaluate the cost-effectiveness of the intervention with an Australian health economic evaluation.

This trial will provide high quality evidence that can be rapidly translated into clinical practice to improve patient outcomes and provide economic benefits in Australia and internationally.