Trial Review

The safety and scientific validity of this study is the responsibility of the study sponsor and investigators. Listing a study does not mean it has been endorsed by the ANZCTR. Before participating in a study, talk to your health care provider and refer to this information for consumers
Trial registered on ANZCTR


Registration number
ACTRN12620001319965
Ethics application status
Approved
Date submitted
11/06/2020
Date registered
7/12/2020
Date last updated
25/04/2024
Date data sharing statement initially provided
7/12/2020
Date results information initially provided
25/04/2024
Type of registration
Prospectively registered

Titles & IDs
Public title
Re-engineering the clinical approach to suspected cardiac chest pain assessment in the emergency department by expediting research evidence to practice using artificial intelligence
Scientific title
Re-engineering the clinical approach to suspected cardiac chest pain assessment in the emergency department by expediting research evidence to practice using artificial intelligence
Secondary ID [1] 301497 0
Nil Known
Universal Trial Number (UTN)
U1111-1260-3563
Trial acronym
RAPIDx AI
Linked study record

Health condition
Health condition(s) or problem(s) studied:
Acute Coronary Syndromes 317831 0
Condition category
Condition code
Cardiovascular 315887 315887 0 0
Coronary heart disease
Emergency medicine 316140 316140 0 0
Other emergency care

Intervention/exposure
Study type
Interventional
Description of intervention(s) / exposure
This study will use AI-based support for clinical decision in a system-wide approach. Real-time health data will be assimilated within the clinical decision-support tool which will be deployed in South Australian emergency departments. This AI clinical support aid will provide clinicians with objective patient-specific diagnostic probabilities and prognostic assessments alongside recommended evidence-based clinical management.

The AI-based support will be assimilated systems wide thus obtaining an opt-in participant consent form will be operationally unfeasible. However, all patients will be able to opt-out of providing their data at any point of the study. To enable informed patient participation, posters and information sheets will be accessible to patients, their clinical team/clinician and/or their next-of-kin.

In the hospitals allocated to the intervention arm, real-time health data will be assimilated within the electronic decision-support tool being developed in partnership with Siemens Healthineers. This will be a web-based tool, accessible on any computer or mobile device, and will require login credentials. Data assimilation will occur in an automated manner where possible, with manual entry occurring only as required (determined by electronic data system maturity at site). The decision-support tool will then provide clinicians with objective patient-specific diagnostic probabilities (i.e. the likelihood for Type 1 MI, vs Type 2 MI, vs cardiac injury etc.) and prognostic assessments alongside recommended clinical management. Specifically, the AI algorithm will report the probabilities for the various types of myocardial infarction and myocardial injury. Importantly, since a therapeutic evidence base exists only for Type 1 myocardial infarction, treatment recommendations will only be presented when the probability for this condition exceeds 90%. In all other scenarios, only ongoing diagnostic recommendations will be made.

This intervention does not mandate any clinical procedures. However if any procedures are required, the participant will undergo the standard medical procedure consent process, conducted by appropriately qualified medical personnel in line with SA Health policies, which is independent of this study.

This AI-based support will be trialed in South Australian Hospital Emergency Departments for a period of 12 months post-index presentation for the last randomised participant.

Adherence to the study will be monitored through data linkage of health system data and interrogation of medical records as required.
Intervention code [1] 317804 0
Diagnosis / Prognosis
Comparator / control treatment
For the control arm of the study, the same variables will be collected however no risk prediction will be calculated nor suggested clinical management provided.
Control group
Active

Outcomes
Primary outcome [1] 324095 0
The primary outcome will be the composite endpoint (EP):
-Cardiovascular mortality
-New or recurrent MI defined by the current 4th Universal Definition of Myocardial Infarction
-Unplanned hospital re-admission: non-elective coronary revascularisation; cerebrovascular accidents; atrial or ventricular arrhythmia; congestive cardiac failure without MI; PE and aortic dissection as documented by a hospital discharge summary.
N.B All in-hospital myocardial infarctions (MI) within 12 hours of presentation will be excluded from primary outcomes.

All data for in-hospital care will be obtained by data linkage of public and private hospital admissions and care through electronic medical records systems and linkage to national databases (e.g. NDI, MBS/PBS).. Where required, paper medical records will be sought.

Clinical outcomes will be assessed by an independent clinical events committee and use blinded end-point adjudication.
Timepoint [1] 324095 0
30 days post-index presentation
Secondary outcome [1] 383754 0
All-cause mortality

All data for in-hospital care will be obtained by data linkage of public and private hospital admissions and care through electronic medical records systems and linkage to national databases (e.g. NDI, MBS/PBS).. Where required, paper medical records will be sought.
Timepoint [1] 383754 0
30-days and 12 months post-index presentation
Secondary outcome [2] 383756 0
Representation with myocardial infarction or unstable angina

All data for in-hospital care will be obtained by data linkage of public and private hospital admissions and care through electronic medical records systems and linkage to national databases (e.g. NDI, MBS/PBS).. Where required, paper medical records will be sought.

Clinical outcomes will be assessed by an independent clinical events committee and use blinded end-point adjudication.
Timepoint [2] 383756 0
30-days and 12 months post-index presentation
Secondary outcome [3] 383757 0
Representation with acute and chronic myocardial injury.

All data for in-hospital care will be obtained by data linkage of public and private hospital admissions and care through electronic medical records systems and linkage to national databases (e.g. NDI, MBS/PBS).. Where required, paper medical records will be sought.

Clinical outcomes will be assessed by an independent clinical events committee and use blinded end-point adjudication.
Timepoint [3] 383757 0
30-days and 12 month post-index presentation
Secondary outcome [4] 383758 0
Re-presentation to any emergency department with suspected cardiac chest pain.

All data for in-hospital care will be obtained by data linkage of public and private hospital admissions and care through electronic medical records systems and linkage to national databases (e.g. NDI, MBS/PBS).. Where required, paper medical records will be sought.

Clinical outcomes will be assessed by an independent clinical events committee and use blinded end-point adjudication.
Timepoint [4] 383758 0
30-days and 12-months post-index presentation
Secondary outcome [5] 383759 0
Significant bleeding using Bleeding Academic Research Consortium (BARC) criteria.

All data for in-hospital care will be obtained by data linkage of public and private hospital admissions and care through electronic medical records systems and linkage to national databases (e.g. NDI, MBS/PBS).. Where required, paper medical records will be sought.

Clinical outcomes will be assessed by an independent clinical events committee and use blinded end-point adjudication.
Timepoint [5] 383759 0
30-days and 12-months post-index presentation
Secondary outcome [6] 383760 0
Measures of in-hospital care: stress testing at discharge consistent with guidelines

This outcome will be assessed through linkage of health system data systems.
Timepoint [6] 383760 0
30-days and 12-months post-index presentation
Secondary outcome [7] 383761 0
Health-related quality of life (EQ-5D-5L)

This data will be acquired through surveying of participants at specified timepoints.
Timepoint [7] 383761 0
30-days, 6-and 12-months post-index presentation.
Secondary outcome [8] 383762 0
Resource utilisation over 12 months: Medicare data among consenting patients using Medical Benefits Schedule (MBS), medication use from Pharmaceutical Benefits Schedule (PBS) and inpatient admissions from the AN-Diagnosis Related Group (DRG) version 8.0.

This secondary outcome data will be acquired through the SA Health data infrastructure for all SA residents. This data will be obtained by data linkage of public and private hospital admissions and care through systems and, from Medicare (Medical Benefits Schedule (MBS), Pharmaceutical Benefits Schedule (PBS)), and in-patient admissions from the AN-Diagnosis Related Group (DRG).
Timepoint [8] 383762 0
12 months post-index presentation
Secondary outcome [9] 384349 0
Cardiovascular mortality

All data for in-hospital care will be obtained by data linkage of public and private hospital admissions and care through electronic medical records systems and linkage to national databases (e.g. NDI, MBS/PBS).. Where required, paper medical records will be sought.

Clinical outcomes will be assessed by an independent clinical events committee and use blinded end-point adjudication.
Timepoint [9] 384349 0
30-days and 12 months post-index presentation
Secondary outcome [10] 384350 0
New or recurrent MI defined by the current 4th Universal Definition of Myocardial Infarction.

All data for in-hospital care will be obtained by data linkage of public and private hospital admissions and care through electronic medical records systems and linkage to national databases (e.g. NDI, MBS/PBS).. Where required, paper medical records will be sought.

Clinical outcomes will be assessed by an independent clinical events committee and use blinded end-point adjudication.
Timepoint [10] 384350 0
30-days and 12 months post-index presentation
Secondary outcome [11] 384351 0
Unplanned hospital re-admission: non-elective coronary revascularisation; cerebrovascular accidents; atrial or ventricular arrhythmias; congestive cardiac failure without MI; PE and aortic dissection as documented by a hospital discharge summary.

All data for in-hospital care will be obtained by data linkage of public and private hospital admissions and care through electronic medical records systems and linkage to national databases (e.g. NDI, MBS/PBS).. Where required, paper medical records will be sought.

Clinical outcomes will be assessed by an independent clinical events committee and use blinded end-point adjudication.
Timepoint [11] 384351 0
30-days and 12-months post-index presentation
Secondary outcome [12] 384352 0
Measures of in-hospital care: echocardiography at discharge consistent with guidelines

This outcome will be assessed through linkage of health system data systems.
Timepoint [12] 384352 0
30-days and 12 months post-index presentation
Secondary outcome [13] 384353 0
Measures of in-hospital care: coronary angiography at discharge consistent with guidelines

This outcome will be assessed through linkage of health system data systems.
Timepoint [13] 384353 0
30-days and 12 months post-index presentation
Secondary outcome [14] 384354 0
Measures of in-hospital care: cardiac medications at discharge consistent with guidelines

This outcome will be assessed through linkage of health system data systems.
Timepoint [14] 384354 0
30-days and 12 months post-index presentation.

Eligibility
Key inclusion criteria
Patients presenting to the emergency department will be considered eligible for analysis if they meet all of the following:
a) Clinical features of chest pain or suspected ACS as the principal cause; and
b) At least one high-sensitivity troponin T assay is drawn; and
c) Age of 18 years or older
Minimum age
18 Years
Maximum age
No limit
Sex
Both males and females
Can healthy volunteers participate?
No
Key exclusion criteria
Patients presenting to the ED will be considered ineligible for analysis if they meet any of the following:
a) Are re-presenting with suspected cardiac chest pain within 30 days of last presentation for suspected cardiac chest pain; or
b) Arrive as a transfer after initial assessment within another hospital ED; or
c) Reside interstate or overseas; or
d) Wish to opt-out.

Study design
Purpose of the study
Diagnosis
Allocation to intervention
Randomised controlled trial
Procedure for enrolling a subject and allocating the treatment (allocation concealment procedures)
Allocation is not concealed.
Methods used to generate the sequence in which subjects will be randomised (sequence generation)
Masking / blinding
Blinded (masking used)
Who is / are masked / blinded?


The people assessing the outcomes
The people analysing the results/data
Intervention assignment
Parallel
Other design features
Phase
Not Applicable
Type of endpoint/s
Safety/efficacy
Statistical methods / analysis
Sample size calculation: This study is powered to detect a hazard ratio of 0.776 (i.e. event rate 25% versus 20% representing a NNT of 20). Assuming an intraclass correlation coefficient (ICC) of 0.001, as derived from our recent RCTs, a total of 1338 patients per arm from 12 hospitals will be required (PASS version 16.0.5). However, assuming 75% of patients receiving troponin testing do not have an elevated result, a total sample size of n=9600 (i.e. 12x800 per participating emergency department) is planned and will maintain >80% power.

Primary analysis: The primary analysis will employ the intention-to-treat (ITT) population including all eligible randomized patients. The study will be reported according to CONSORT guidelines. The two groups will be compared on baseline characteristics, with Chi-squared tests undertaken for categorical variables, and independent samples t-tests for continuous variables. Non-parametric analyses will be used where necessary. The initial analysis will compare the time to the composite primary outcome within 12 months between the decision-support and standard care groups, using Cox proportional hazards models with shared frailty to account for correlation within hospitals.

Economic analysis: Patient-level measures of utility derived from the EQ-5D instrument will be integrated with survival curves to estimate quality-adjusted life years in each trial arm using the quality-adjusted survival analysis (QASA) method. Time and resources required for decision-support implementation will be included in the cost-effectiveness analysis. Within-trial incremental costs associated with the decision-support and with standard care will be estimated from patient data on MBS, PBS and hospital use (estimated 50% of cohort). Within-trial cost-effectiveness will then be analysed allowing for bivariate uncertainty with bootstrapping of patient costs and effects to maintain covariance structure. This analysis will include cost-effectiveness, acceptability, net benefit and expected net loss curves to inform decision makers of the optimal strategy at any given threshold, uncertainty around this decision, and the value of further research locally and internationally.

Recruitment
Recruitment status
Completed
Date of first participant enrolment
Anticipated
1/03/2023
Actual
16/05/2023
Date of last participant enrolment
Anticipated
16/05/2024
Actual
31/12/2023
Date of last data collection
Anticipated
Actual
31/12/2023
Sample size
Target
20000
Accrual to date
Final
14812
Recruitment in Australia
Recruitment state(s)
SA
Recruitment hospital [1] 16876 0
Flinders Medical Centre - Bedford Park
Recruitment hospital [2] 16877 0
The Royal Adelaide Hospital - Adelaide
Recruitment hospital [3] 16878 0
The Queen Elizabeth Hospital - Woodville
Recruitment hospital [4] 16880 0
Modbury Hospital - Modbury
Recruitment hospital [5] 16882 0
Murray Bridge Soldiers Memorial Hospital - Murray Bridge
Recruitment hospital [6] 16883 0
South Coast District Hospital - Victor Harbor
Recruitment hospital [7] 16884 0
Riverland Regional Health Service - Berri Hospital - Berri
Recruitment hospital [8] 16885 0
Port Augusta Hospital - Port Augusta
Recruitment hospital [9] 16886 0
Whyalla Hospital - Whyalla
Recruitment hospital [10] 16891 0
Noarlunga Health Service - Noarlunga Centre
Recruitment hospital [11] 16892 0
Mount Gambier and Districts Health Service - Mount Gambier
Recruitment hospital [12] 16893 0
Lyell McEwin Hospital - Elizabeth Vale
Recruitment hospital [13] 23738 0
Port Pirie Hospital - Port Pirie
Recruitment postcode(s) [1] 30526 0
5042 - Bedford Park
Recruitment postcode(s) [2] 30527 0
5000 - Adelaide
Recruitment postcode(s) [3] 30528 0
5011 - Woodville
Recruitment postcode(s) [4] 30530 0
5092 - Modbury
Recruitment postcode(s) [5] 30532 0
5253 - Murray Bridge
Recruitment postcode(s) [6] 30533 0
5211 - Victor Harbor
Recruitment postcode(s) [7] 30534 0
5343 - Berri
Recruitment postcode(s) [8] 30535 0
5700 - Port Augusta
Recruitment postcode(s) [9] 30536 0
5600 - Whyalla
Recruitment postcode(s) [10] 30542 0
5168 - Noarlunga Centre
Recruitment postcode(s) [11] 30543 0
5290 - Mount Gambier
Recruitment postcode(s) [12] 30544 0
5112 - Elizabeth Vale
Recruitment postcode(s) [13] 39178 0
5540 - Port Pirie

Funding & Sponsors
Funding source category [1] 305936 0
Government body
Name [1] 305936 0
National Health and Medical Research Council (NHMRC)
Country [1] 305936 0
Australia
Primary sponsor type
University
Name
Flinders University of South Australia
Address
1 Flinders Drive, Bedford Park 5042, South Australia
Country
Australia
Secondary sponsor category [1] 306390 0
None
Name [1] 306390 0
Address [1] 306390 0
Country [1] 306390 0

Ethics approval
Ethics application status
Approved
Ethics committee name [1] 306182 0
The Southern Adelaide Clincal Human Research Ethics Committee
Ethics committee address [1] 306182 0
1 Flinders Drive, Flinders Medical Centre, Bedford Park, South Australia-5042
Ethics committee country [1] 306182 0
Australia
Date submitted for ethics approval [1] 306182 0
26/10/2020
Approval date [1] 306182 0
04/02/2021
Ethics approval number [1] 306182 0

Summary
Brief summary
Within Australia, suspected cardiac chest pain represents nearly 1 million emergency department (ED) presentations every year. Most patients are eventually not diagnosed with acute coronary syndromes (ACS). Further, the clinical work-up of patients with suspected ACS is laborious and complex, often leading to unnecessary invasive tests. In an effort to improve decision-making, and thereby reduce unnecessary risk to patients and associated health-economic impacts, this study will elucidate the pivotal role of artificial intelligence (AI) as an aid in ACS diagnosis. This study will implement and evaluate the system-level intervention of AI-based decision support for clinical assessment of suspected cardiac chest pain in the reduction of death, myocardial infarction and 12-month readmissions. Further, the study will also provide the cost-effectiveness of embedding AI-based decision support in routine clinical assessment of suspected chest-pain and ACS.
Trial website
Trial related presentations / publications
Public notes

Contacts
Principal investigator
Name 102998 0
Prof Derek Chew
Address 102998 0
Flinders University
Sturt Road
Bedford Park SA 5042
Country 102998 0
Australia
Phone 102998 0
+61 03 7511 1680
Fax 102998 0
Email 102998 0
[email protected]
Contact person for public queries
Name 102999 0
Ms Kristina Lambrakis
Address 102999 0
Flinders University
Sturt Road
Bedford Park SA 5042
Country 102999 0
Australia
Phone 102999 0
+61375111854
Fax 102999 0
Email 102999 0
[email protected]
Contact person for scientific queries
Name 103000 0
Ms Kristina Lambrakis
Address 103000 0
Flinders University
Sturt Road
Bedford Park SA 5042
Country 103000 0
Australia
Phone 103000 0
+610375111854
Fax 103000 0
Email 103000 0
[email protected]

Data sharing statement
Will individual participant data (IPD) for this trial be available (including data dictionaries)?
No
No/undecided IPD sharing reason/comment
This study will report outcomes on collective and de-identified datasets over a course of 12 months from index presentation on a large volume of participants. Thus, providing IPD will be operationally unfeasible.


What supporting documents are/will be available?

No Supporting Document Provided



Results publications and other study-related documents

Documents added manually
No documents have been uploaded by study researchers.

Documents added automatically
No additional documents have been identified.