This study evaluated the effectiveness of electronic clinical decision support (CDS) in reducing inappropriate blood transfusions in a large acute hospital. Researchers analysed the impact of electronic alerts on red blood cell (RBC) and platelet transfusions. They found that in general hospital areas, the alerts significantly and sustainably reduced RBC transfusions. In critical care (CC), the reduction was not sustained, and in haematology/oncology (HO), there was no impact on RBC transfusions. For platelet transfusions, alerts stopped the rising trend outside of CC and HO, and reduced transfusions in CC, but had little impact in HO. The study suggests that while CDS can immediately change user behaviour, it needs to account for specific circumstances in specialized areas like CC and HO to be effective.

Abstract

Background

Electronic clinical decision support (CDS) within Electronic Health Records has been used to improve patient safety, including reducing unnecessary blood product transfusions. We assessed the effectiveness of CDS in controlling inappropriate red blood cell (RBC) and platelet transfusion in a large acute hospital and how speciality specific behaviours changed in response.

Methods

We used segmented linear regression of interrupted time series models to analyse the instantaneous and long-term effect of introducing blood product electronic warnings to prescribers. We studied the impact on transfusions for patients in critical care (CC), haematology/oncology (HO) and elsewhere.

Results

In non-CC or HO, there was significant and sustained decrease in the numbers of RBC transfusions after introduction of alerts. In CC the alerts reduced transfusions, but this was not sustained, and in HO there was no impact on RBC transfusion. For platelet transfusions outside of CC and HO, the introduction of alerts stopped a rising trend of administration of platelets above recommended targets. In CC, alerts reduced platelet transfusions, but in HO alerts had little impact on clinician prescribing.

Conclusion

The findings suggest that CDS can result in immediate change in user behaviour which is more obvious outside specialist settings of CC and HO. It is important that this is then sustained. In CC and HO, blood transfusion practices differ. CDS thus needs to take specific circumstances into account. In this case there are acceptable reasons to transfuse outside of these crude targets and CDS should take these into account.

Authors: Jolene Atia , Felicity Evison , Suzy Gallier , Sophie Pettler , Mark Garrick , Simon Ball , Will Lester , Suzanne Morton , Jamie Coleman , Tanya Pankhurst