Effect of a machine learning-derived early warning system for hypotension vs standard care on depth and duration of intra- and postoperative hypotension in elective cardiac surgery - The HYPE-2 Randomized Clinical Trial

Hypotension during cardiac surgery and post-operative intensive care unit (ICU)
admission is associated with adverse outcomes. Yet, hypotension in the
operating room (OR) and post-operative ICU admission is common. In a recent
study performed in non-cardiac surgical patients in our center, we found that
up to 60% of patients endured hypotension (defined as MAP below 65 mmHg) during
anesthesia for an average of 10% of surgery time. The incidence of hypotension
in post cardiac surgery patients admitted to the Intensive Care Unit (ICU) of
our center was, with 84%, even higher (unpublished data PHYSIC 1 study,
conducted in the ICU of the Amsterdam UMC, location AMC). Hypotension is
preventable, however, current management of hypotensive episodes is
predominantly reactive and rather occurs with delay. Edwards Lifesciences
(Irvine, CA) has developed an algorithm using continuous invasively-measured
arterial waveforms to predict hypotension with high accuracy minutes before
blood pressure actually decreases, the so called Hypotension Prediction Index
(HPI). Two randomized controlled trials (RCT) in non-cardiac surgery have shown
significant reduction in hypotension when HPI with diagnostic guidance was
compared to standard care. No difference in median TWA of hypotension was found
in another clinical RCT. However, in the latter, HPI with diagnostic guidance
was associated with less hypotension when analysis was restricted to episodes
during which clinicians intervened. Hence, poor protocol compliance in this
trial is the most likely cause of the insignificant results. HPI has never been
tested in cardiac surgery patients or patients admitted to the ICU. We
hypothesize that the use of this algorithm will reduce hypotension as measured
by the time weighted average (TWA) during the off-pump phase of on-pump
coronary artery bypass graft (CABG) surgery (excluding cardiopulmonary bypassp
pump time) and the mechanically ventilated duration of post-operative ICU
admission.

Does HPI with diagnostic guidance affect severity of hypotension (defined as
MAP<65 mmHg) during both the off-pump phases of on-pump CABG surgery and the
mechanically ventilated phase of post-operative ICU admission (or 8 hours
maximum)?

This is a single center randomized controlled trial.

We will randomize the participants into two arms:

1) Treatment arm: HemoSphere Advanced Monitoring Platform (HemoSphere) with
Acumen IQ sensor with HPI software. The treating cardiac
anesthesiologist/anesthesia nurse (operating room) and critical care
nurse/intensivist (intensive care) are trained to use Acumen IQ variables and
HPI. Advanced hemodynamic variables provided by the HemoSphere include SVR,
SVV, SV(I), CO/CI, dp/dt, Eadyn. The treating anesthesiologist, anesthesia
nurse, intensivist and critical care nurse is provided with guidance by means
of a flowchart suggesting when to treat and what the cause of hypotension is.
Timing of treatment and choice of treatment is then left to the discretion of
the attending anesthesiologist, anesthesia nurse or intensivist. In the ICU a
nurse driven protocol will be provided, describing the suggested treatment per
cause of hypotension for the critical care nurses.

2) Conventional arm: Institutional Standard of Care with an intention to keep
MAP equal to or above (>=) 65 mmHg. Standard hemodynamic monitoring for CABG
patients during OR and ICU stay in our institutions includes: MAP, systolic
blood pressure (SBP), diastolic blood pressure (DBP), heart rate and pulse
pressure variation (PPV). The HemoSphere with Acumen IQ sensor will be
connected to collect data but the HemoSphere monitor will be covered completely
for the treating physicians and nurses (anesthesiologist, anesthesia nurse,
intensivist and critical care nurse). Timing and choice of treatment is conform
current clinical practice (and thus left completely to the discretion of the
attending anesthesiologist, anesthesia nurse, intensivist and critical care
nurse).

The intervention used in the treatment arm will consist of: HemoSphere with
Acumen IQ variables including HPI and diagnostic guidance. HPI will be
calculated via Acumen IQ sensor connected to the radial arterial line. The
treating anesthesiologist, anesthesia nurse, intensivist and critical care
nurse is trained to understand HemoSphere parameters and the meaning of HPI.
The treating anesthesiologist, anesthesia nurse, intensivist and critical care
nurse is provided with guidance concerning timing of treatment (HPI > 75) and
the causes of hypotension. HPI values above 75 translate to a 75% of
hypotension to occur in the following minutes. HPI values between 50-75
translates approximately to a 50-75% change of hypotension to occur in the
following minutes. When HPI falls in the 50-75 range the study investigator
starts diagnosing the cause of pending hypotension via the secondary screen.
The secondary screen of the HemoSphere provides the investigator and treating
anesthesiologist, anesthesia nurse, intensivist and critical care nurse with
variables such as dp/dt, dynamic elastance, systemic vascular resistance as
well as stroke volume, cardiac output and stroke volume variation. If HPI
reaches >75, treatment is suggested to be started and the investigator informs
the anesthesiologists of the most likely cause of hypotension. The
anesthesiologist, anesthesia nurse and intensivist treat the patient based on
these suggestions but may choose to deviate from the study protocol if deemed
necessary in both timing as well as treatment. The critical care nurse also
treats the patient based on these suggestions but treatment options are
described in a nurse driven protocol. The critical care nurse may also choose
to deviate from the study protocol if deemed necessary in both timing as well
as treatment, but treatment options are limited to those described in the nurse
driven protocol.

There are no additional risks associated with the use of the HemoSphere other
than described in the Instructions for Use. The study participants allocated to
treatment according to de HPI algorithm will receive diligently titrated
vasopressors, inotropes or fluids minutes before hypotension is predicted to
occur. These treatment options are not different to conventional therapy. The
anesthesiologist, anesthesia nurse, critical care nurse and intensivist treats
the patient based on our suggestions but may choose to deviate from the study
protocol if deemed necessary in both timing as well as treatment.
Theoretically, early treatment may lead to (transient) hypertension (MAP>100).
However, no increase in TWA of time spent in hypertension (defined as MAP >
100mmHg) was observed in a RCT, comparing HPI with diagnostic guidance to
standard care (Wijnberge, JAMA, 2020). In addition, no differences in the
incidence of serious complications, including cardiac arrhythmia, was found. No
other potential adverse events are expected from participation in this trials.