Supporting early recognition of clinical deterioration for Mobile Vital signs monitoring of high-risk surgical patients

Patients admitted to the hospital for surgical care are at risk for developing
adverse events in the pre- or postoperative trajectory, which may lead to
deterioration requiring escalation of care. To promote patient safety, it is
important to recognize and start treatment in an early phase of deterioration.
Wireless sensors that allow continuous vital sign tracking with minimal patient
burden may aid early identification of deterioration in the hospital ward or
out-of-hospital setting.
To facilitate efficient implementation of continuous monitoring in these
settings, it is crucial that caregivers are supported in the interpretation of
the large amount of data. Current monitoring systems typically use
threshold-based alarm systems to detect abnormal vital sign levels. However,
these systems are known by high false alarm rates and may miss subtle but
relevant trends, related to the fact that the presentation of deterioration
differs per patient, per setting, and per type of underlying adverse event.
Therefore, there is need for patient specific methods to support early
identification of patterns of deterioration that associated with vital
instability.

Primary objective:
1. To develop and evaluate a model for prediction of deterioration
necessitating escalation of care, based on continuous vital signs recordings,
self-reported patient symptoms, nurse observations, clinical measurements, and
patient characteristics

Secondary objectives:
2. To gain insight in patterns of deterioration that can be observed in
continuous vital signs recordings, self-reported patient symptoms, nurse
observations and clinical measurements
3. To gain insight in effective methods for personalized detection of patterns
of deterioration
4. To explore the potential clinical value of the developed model for
prediction of deterioration

In the current study, we will develop a holistic model to predict deterioration
necessitating escalation of care in high-risk surgical ward patients, by
detecting patterns based on physiological assumptions. The model is developed
and evaluated using an existing study database (MoViSign; NL.65885.044,
2018-2019, ZGT Almelo) and data that is collected in a prospective
observational study.

The MoViSign study database includes continuous vital signs recordings and
clinical data obtained in high-risk surgical ward patients (N=33 upper GI
patients, N=27 hip fracture patients). In the prospective study part, we will
obtain additional data in the same population (N*40 upper GI patients, N*40 hip
fracture patients) and setting. Accordingly, a wireless sensor will be placed
on the patient*s chest during the peri-operative hospital stay for continuous
registration of vital signs (ECG, heart rate, respiratory frequency,
temperature) and activity level (accelerometry, posture, steps). In addition,
continuous vital sign measurements will be performed in the home situation for
elective patients two days prior to hospital admission and seven days after
discharge. These home measurements aim to collect reference data and explore
the potential use of the model for detection of late complications in a home
setting respectively. In addition to the vital sign measurements, patients will
register symptoms using a daily diary during ward stay and home measurements.
Last, nurse worry will be registered by nurses during ward stay.
After completion of data collection, the presence and expected timing of
deterioration will be defined for each patient included in the database by two
clinical specialists based on the patient record, as gold standard for
deterioration events. Next, the total database will be split in a model
development set and a model evaluation set. In the model development phase,
patterns of deterioration will be described based on physiological assumptions
and clinical experience. Next, these pattern descriptions will be translated to
detection criteria to predict deterioration based on vital signs recordings,
self-reported patient symptoms and nurse observations, clinical measurements,
and patient characteristics. Last, the detection criteria will be optimized
using case observation in the model development set.
The performance of the resulting model will be evaluated in the model
evaluation set in terms of sensitivity and false detection rate, and compared
with the performance of the Modified Early Warning Score (MEWS). Last, it will
be explored whether this model could contribute to earlier recognition and
treatment of deterioration, by evaluating the time between prediction and
current clinical response.

This study includes two patient groups which have high risk of deterioration
due to high rate or large impact of complications during the perioperative
trajectory. Accordingly, these patients may particularly benefit from improved
methods to support detection of deterioration. As patterns of deterioration may
vary per patient group, it is important to conduct the study in the target
population and setting.
The study is a non-therapeutic study, and patients will receive care as usual.
Individual patients will not experience benefit from participating in this
study.
The burden to patients related with the sensor recordings is minimal given the
fact that patients will likely not experience any physical discomfort, because
the sensor device is a small and noninvasive. As with any measurement using
adhesive electrodes or plasters, it is possible that some patients will
experience skin irritation in response to the adhesive patches. If skin
irritation occurs, the sensors will be removed. The sensor does not restrict
daily activities by patients and do not require actions by patients. As an
exception, patients will be asked to keep the mobile data receiver charged and
in range as much as possible during home measurements.
The patient diary will be provided to the patient on a paper form, including a
standard daily questionnaire for registration of the patient*s condition,
symptoms, and daily activities (home measurements only). The nature of
questions and time needed to fill in the diary will provide minimal burden to
the patient.