Marjon Borgert

182 Chapter 8 Transfusion bundle During the baseline period 466 units of RBCs were transfused, 284 during the implementation period and 378 units post-implementation. After introducing the transfusion bundle, compliance was 50% (141/284) during implementation, and 37% (141/378) during the post-implementation period (di erence -13%, 95% CI: 0.045 to 0.202, P -value = 0.002). Red Blood Cell transfusions In Figure 1, the preset Hb thresholds and the last recorded pre-transfusion Hb levels are shown. The mean pre-transfusion Hb level was 7.3 g/dL -1 (SD=1.15) during the baseline period. After introducing the bundle the mean pre-transfusion Hb level decreased to 7.1 g/dL -1 (SD=1.04), di erence 0.2 g/dL -1 , 95% CI: 0.009 to 0.308, P -value = 0.037. In Figure 2, the percentage deviation between the preset Hb thresholds and the last recorded pre-transfusion Hb levels is shown. In the vast majority of transfusions the pre- transfusion Hb was below the Hb thresholds, i.e. appropriately transfused. Compared to the baseline period, the number of inappropriate transfusions signi cantly decreased during implementation from 25% (111/439) to 15% (42/280) during implementation, di erence 10%, 95% CI: -0.164 to -0.042, P -value 0.001. During the post-implementation period, the number of inappropriate transfusion further decreased to 12% (45/370). Table 3 shows the results from the univariate analysis. The covariates ‘intervention’, i.e. transfusion bundle, and ‘pre-transfusion Hb’ showed a statistical signi cant e ect on the appropriateness of transfusions. The logistic regression analysis shows that the chance (Odds) to nd an appropriate transfusion is approximately twice as high after the implementation of the transfusion bundle (Table 4). The pre-transfusion Hb level in uences this e ect. The direction of this odds ratio shows that a lower pre-transfusion Hb level is associated with better protocol compliance, resulting in more appropriate transfusions when lower pre-transfusion Hb levels were measured.