Sanne de Bruin

109 InPUT- pilot study in the Netherlands When the patient or their legal representative did not give consent for study partici- pation, the patient was excluded from the study and data was not used for analysis. Data collection Data was collected prospectively by a research student (see appendix 3 for CRF). Baseline characteristics included information about the demographics, comorbidities, Acute Physiologic Assessment and Chronic Health Evaluation (APACHE) lV score, type and reason of admission. Clinical and laboratory data were collected daily to compare patients who received a transfusion to patients who did not receive a transfusion, in- cluding nadir haemoglobin (Hb), nadir platelet count, highest prothrombin time (PT) and highest activated partial thromboplastin time (aPTT). For each transfusion event a short transfusion CRF was completed. A transfusion event was defined as a transfusion of a blood derived product or the administration of a coagulation factor, ordered within one order in the electronic patient system. For each type of blood product or transfusion (RBCs, plasma, platelets, coagulation factors including prothrombin complex, fibrin- ogen and tranexamic acid and activation of massive transfusion protocol) a separate CRF was designed. The 28-day mortality and the location of the patient (home, ward, ICU) were also collected. Outcomes The primary outcome of this study was to assess the current transfusion practice at our ICU, described by the number of RBC, platelet and plasma transfusions and the admin- istration of coagulation factors, correlated with Hb levels, platelet count, INR/PT and fibrinogen prior to and post transfusion. The secondary outcome was the feasibility of the study protocol, assessed by the workload, missing data and the number of transfu- sion and administration of coagulation factors in this patient cohort. Furthermore, the design of a worldwide follow-up study was based on this pilot data. Data analysis The statistical analysis was performed using the statistical program R version 3.5.2. Continuous normally distributed data are presented in means (standard deviation) and not normally distributed data as medians (first-third quartile). The Wilcoxon sum rank test was used to test the dependence of two grouping variables on continuous, non-normally distributed data. Categorical variables were expressed as number and percentage. Categorical data were compared using the chi-square test. P-values below 0.05 were considered to be statistically significant. 4