Vomiting Prevention in Children With Cancer
Trial statusRecruiting
Trial phaseNot applicable
Trial typeInterventional
Biological sexAll
AgeNot listed
SponsorThe Hospital for Sick Children
About this trial
The goal of this single arm trial is to learn if a machine learning (ML) model predicting the risk of vomiting within the next 96 hours will impact vomiting outcomes in inpatient cancer pediatric patients.
The main questions it aims to answer are whether an ML model predicting the risk of vomiting within the next 96 hours will:
Primary
1\. Reduce the proportion with any vomiting within the 96-hour window
Secondary
1. Reduce the number of vomiting episodes 2. Increase the proportion receiving care pathway-consistent care 3. Impact on number of administrations and costs of antiemetic medications
Newly admitted participants will have a ML model predict the risk of vomiting within the next 96 hours according to their medical admission information. The prediction will be made at 8:30 AM following admission. Pharmacists will be charged with bringing information about patients' vomiting risk to the attention of the medical team and implementing interventions.
Eligibility criteria
This trial does not accept healthy volunteersQualifiers
All pediatric patients admitted to the oncology service at SickKids
Disqualifiers
Pediatric patients admitted to the oncology service at SickKids that are discharged prior to prediction time
Trial design
Design model
Single group
Treatments tested in this trial
ML-based intervention
Other interventionFor each patient, a ML model will predict the risk of vomiting within the next 96 hours. Patients will then receive care pathway-consistent interventions based on the ML model predictions.
Treatment groups
1,332 Participants
are divided into 1 treatment groupGroup A: ML modelExperimental treatment 1 intervention
Trial outcomes
Primary outcomes
1
Vomiting post prediction time
The primary outcome will be any vomiting (a binary variable) 0-96 hours post prediction time. In the SickKids EHR, vomiting is described in the flowsheets by emesis volume, emesis count, emesis amount description, emesis color/appearance, and any vomiting/retching/gagging. Multiple descriptors can be used at a specific time stamp but no one descriptor is used consistently. Thus, the best measure of vomiting is a binary variable (yes/no) where yes represents any vomiting entry within the 96-hour window. Vomiting determination using this approach was validated. In a retrospective assessment, patients who received etoposide, ifosfamide or treosulfan were identified and 60 patients were randomly selected with stratification by age and HCT status. (Patel P et al, 2023)
Time frame
0-96 hours post prediction timeSecondary outcomes
1
Number of episodes of vomiting
The number of episodes of vomiting will be determined by counting the number of distinct time stamps with any vomiting description.
Time frame
0-96 hours post prediction time2
Care pathway-consistent care
Care pathway-consistent care will be measured as a binary outcome (yes/no).
Time frame
0-96 hours post prediction time3
Number of antiemetic administrations
All antiemetic administrations within the 96-hour window will be counted. The total number of administrations overall and by specific antiemetic agent will be tabulated.
Time frame
0-96 hours post prediction time4
Antiemetic costs
Medication costs are available in SEDAR. The total costs of antiemetics per kg overall and by specific antiemetic agent will be tabulated.
Time frame
0-96 hours post prediction timeOther outcomes
1
Number of SEDAR Predictions and Interventions
The number of times notes in SEDAR refer to the predictions and interventions to address them including antiemetic modifications during the prediction window will be descriptively reported. The role of the note writer will be documented (such as pharmacist or fellow).
Time frame
0-96 hours post prediction timeSponsors and contacts
Click on the lead sponsor to view all of their trials.
Source ClinicalTrials.gov