Clinical decision-support with Viscoelastic Haemostatic Assay; towards a more practical guided and proactive use of clotting agents

Cardiac surgery is frequently associated with blood loss which forms a great
challenge to teams in the operation room and intensive care unit. Blood loss
and consecutive blood transfusion are associated with increased morbidity,
mortality and costs . To reduce the amount of blood transfusions actions like
coagulation support, also known as haemostatic resuscitation, are important
measures

To improve adequate therapy and save time between blood draw and result, point
of care coagulation testing is available in the form of viscoelastic
haemostatic assays (VHA). VHA analysis can be executed with several techniques
like rotational thromboelastometry® (ROTEM®) and rotational thrombelastography®
(TEG®). Both techniques are able to measure the resistance of a blood sample,
representing viscoelastic changes .

VHA analysis can be used for quick and accurate assessment of the clot function
during cardiac surgery. The use of VHA analysis was validated for haemostasis
testing in cardiac surgery by several studies. Standardized implementation of
VHA analysis reduced blood transfusion, decreased mortality and was shown to be
cost effective. In a pediatric cardiac surgery setting, the use of VHA analysis
decreased mean cardiopulmonary bypass surgery time with 15 minutes.

However, interpretation of VHA is more complex compared to conventional tests
and shows a learning curve. Even though VHA analysis is a validated test that
improves patient outcome, it is not yet frequently used during (cardiac)
surgery. For the interpretation of VHA analysis several flowcharts have been
built, suggesting the difficulty of direct VHA analysis interpretation. VHA
interpretation can be simplified by creating a model able to interpret the
parameters and able to support the physicians in the optimal strategy to
correct coagulation and is therefore of interest.

The primary objective of this study is to provide a supportive model for
optimal coagulation management in the OR and/or ICU to reduce blood loss. This
support will be divided into two models to first assess a problem in the
coagulation cascade and second, to give guidance about the kind and amount of
coagulation therapy necessary.
Both models will be trained on the *gold standard*, stated as the consensus of
at least 2 experts on the interpretation of VHA testing and the advice on the
kind and amount of coagulation therapy. The gold standard will be based on all
VHA analysis, platelet aggregation analysis, and patient/surgery specific data
up to that point, like demographics and intermittently recorded pre- intra and
postoperative data.

The second objective is to compare the actual choice made by the during
surgery/at the ICU with the decision of the expert panel. The products used or
ordered within one hour after VHA analysis will be counted as the choice of the
physician (who is not a member of the expert team).

This is a non-randomized prospective data collection study. We evaluate VHA
parameters at several time point during and after surgery. Furthermore,
demographic data of the patient and the surgical setting are registered. This
study is divided into two phases:

Phase 1:
Phase 1 will be a pilot phase, in which 100 patients scheduled for elective
cardiac surgery are included. Several models will be created based on VHA
parameters, in order to assess the probability whether a supportive model can
be created based on the input parameters of 500 patients. This probability will
be tested calculating the accuracy of the models. Two supportive models will be
created to first interpreted VHA analysis and secondly advise on the optimal
coagulation management.

Phase 2:
In case of a go-decision after phase 1, we will increase the patient population
with 400 to a total maximum of 500 cardiac surgery patients. With this increase
in data, the supportive models will be optimized, resulting in two validated
machine learning models for the VHA interpretation and the advice on the
optimal coagulation management.

Depending on administration of medication or blood products that affect
coagulation status , there will be 4 up to 8 time points where blood is
withdrawn via standardly used arterial line. Executing VHA analysis at 3 time
points is standard of care. In case of an additional need to improve
haemostasis, an extra analysis will be performed for clinical reasons. This
will result in a low extra burden for participants in this study.