Using artificial intelligence as an aid to predict the risk of hospital readmission in patients with COVID-19
| ISRCTN | ISRCTN85858267 |
|---|---|
| DOI | https://doi.org/10.1186/ISRCTN85858267 |
| IRAS number | 126600 |
| Secondary identifying numbers | 11519663, IRAS 126600 |
- Submission date
- 09/02/2022
- Registration date
- 14/03/2022
- Last edited
- 19/10/2022
- Recruitment status
- No longer recruiting
- Overall study status
- Completed
- Condition category
- Infections and Infestations
Plain English summary of protocol
Background and study aims
Up to one-third of patients hospitalised with COVID-19 are readmitted to hospital within 4 months. This figure is higher than would be expected. These patients are more likely to have poorer long-term health, and some will die. It is not known why some patients are more likely to be readmitted, but it might be because they are older, living with other illnesses, living in lower-income areas, recovering from severe COVID-19, unvaccinated, receiving treatment or medication that suppress their immune system, or from an ethnic minority. The aim of this study is to use artificial intelligence as an aid to predict the risk of hospital readmission in patients with COVID-19.
Who can participate?
Patients aged 18 years and older with COVID-19 in England and Scotland
What does the study involve?
The researchers will use data from 220,000 hospital patients within the UK. They have linked these data to general practice and hospital NHS data in England and Scotland, vaccination data and data regarding virus variants. The dataset is necessary to provide detail on patients’ hospital stay, and NHS data will determine the details of readmission. Artificial intelligence approaches will be used to determine the risk of readmission using information about a patient’s disease, treatment and status at discharge. A risk calculator will be built, validated, and made available to the public and healthcare staff while undergoing regulatory approval.
What are the possible benefits and risks of participating?
Identifying those at risk of hospital readmission is important for three reasons. It will help identify patients most likely to have long-term health problems after COVID-19, It will allow safer discharge decisions, and it may enable targeted programmes to support patients at home and reduce the chance of readmission.
Where is the study run from?
University of Edinburgh (UK)
When is the study starting and how long is it expected to run for?
November 2021 to January 2023
Who is funding the study?
1. Health Data Research UK (HDRUK)
2. The Alan Turing Institute (UK)
Who is the main contact?
Prof. Ewen M Harrison
ewen.harrison@ed.ac.uk
Contact information
Principal Investigator
Centre for Medical Informatics, Usher Institute
Edinburgh
EH16 4UX
United Kingdom
 ORCID ID |
0000-0002-5018-3066 |
|---|---|
| Phone | +44 (0)1312423611 |
| ewen.harrison@ed.ac.uk |
Study information
| Study design | Observational cohort study |
|---|---|
| Primary study design | Observational |
| Secondary study design | Clinical prediction model development |
| Study setting(s) | Hospital |
| Study type | Diagnostic |
| Participant information sheet | Not available in web format, please use the contact details to request a participant information sheet |
| Scientific title | Predicting unplanned hospital readmission prior to discharge in patients with COVID-19: development, validation, and implementation of a machine-learning-based risk prediction model |
| Study acronym | 4C-R |
| Study objectives | Research questions: 1. What factors are associated with hospital readmission of COVID-19 patients? 2. What are the consequences of re-admission? 3. Can we reliably predict unplanned hospital readmission with machine learning? |
| Ethics approval(s) | Approved 21/02/2020, South Central - Oxford C Research Ethics Committee (address: not available; +44 (0)207 104 8226, +44 (0)207 104 8241, +44 (0)207 104 8256; oxfordc.rec@hra.nhs.uk), REC ref: 13/SC/0149 Approved 01/05/2020, Scotland A Research Ethics Committee (address: not available; +44 (0)131465 5680; Manx.Neill@nhslothian.scot.nhs.uk), REC ref: 20/SS/0028 |
| Health condition(s) or problem(s) studied | COVID-19 (SARS-CoV-2 infection) |
| Intervention | Trajectory modelling: understanding readmission and its consequences: First, the researchers will examine the factors associated with readmission with time-to-event models, accounting for death as a competing event. Multilevel binary logistic regression will be used to examine the consequences associated with readmission, including the complications of disease and mortality. Risk prediction model development: TRIPOD best practice guidelines will be used. Temporal and geographical validation sets will be held out. Derivation data will be used to define baseline models using logistic regression. A systematic examination of additional algorithms will be undertaken, including discriminative and generative approaches, as well as different neural network architectures. The researchers will pay particular attention to model explainabillity by design. K-fold cross-validation will be used, and performance assessed with area under receiver operator curves, sensitivity (recall), specificity, precision, and accuracy. Calibration will be performed, and validation using the holdout datasets. |
| Intervention type | Other |
| Primary outcome measure | Hospital readmission measured using NHS data at 30 and 90 days |
| Secondary outcome measures | Mortality measured using NHS data at 30 and 90 days |
| Overall study start date | 01/11/2021 |
| Completion date | 31/01/2023 |
Eligibility
| Participant type(s) | Patient |
|---|---|
| Age group | Adult |
| Lower age limit | 18 Years |
| Sex | Both |
| Target number of participants | 5261 |
| Key inclusion criteria | Consecutive patients (ISARIC4C/CO-CIN) aged 18 years and older with a completed index admission for COVID-19 in England and Scotland |
| Key exclusion criteria | Age <18 years |
| Date of first enrolment | 14/02/2022 |
| Date of final enrolment | 30/04/2022 |
Locations
Countries of recruitment
- England
- Scotland
- United Kingdom
Study participating centres
South Bridge
Edinburgh
EH8 9YL
United Kingdom
L1 8JX
United Kingdom
Sponsor information
University/education
NINE Bioquarter
Edinburgh
EH16 4UX
Scotland
United Kingdom
| Phone | +44 (0)131 650 1000 |
|---|---|
| enquiries@ed.ac.uk | |
| Website | https://www.ed.ac.uk |
"ROR" |
https://ror.org/01nrxwf90 |
Funders
Funder type
Charity
No information available
Government organisation / Research institutes and centers
- Alternative name(s)
- The Alan Turing Institute, ATI
- Location
- United Kingdom
Results and Publications
| Intention to publish date | 31/01/2023 |
|---|---|
| Individual participant data (IPD) Intention to share | Yes |
| IPD sharing plan summary | Stored in non-publicly available repository |
| Publication and dissemination plan | The researchers are committed to meaningfully collaborating with public members throughout this project and a lay member (Weatherill) and PPI Research Fellow (Jackson) jointly lead a dedicated PPIE Work Package. The researchers will continue to involve the EAVE II PAG in PPIE activities throughout the project lifecycle. In addition, they have existing relationships with the Long COVID Scotland Action Group and minority ethnic and faith groups who they will engage with to ensure their perspectives are included in this work. The EAVE II PAG and Weatherill have actively contributed to developing the research questions, designing the research, and producing this application. Weatherill is a member of the project management group and will attend project meetings ensuring the public perspective is included throughout the project. PPIE members will be involved in decision-making and have influence over the direction of the research. Their involvement in interpreting findings will bring invaluable public perspective to discussions and helping keep the research grounded in the real world. The public members are viewed as colleagues and will be included in opportunities to co-author papers and attend conferences. The researchers will report on their PPIE activities, measure impact and disseminate in journals and conferences to promote the best practice of PPIE in healthcare research. To help increase public confidence in healthcare data usage, the researchers will work with their PPIE groups to create engaging messages to showcase their project to a lay audience. They will evaluate their PPIE activities to continue to improve how we work with patients, the public, and communities. |
| IPD sharing plan | Applications for use of this data by researchers can be made via the HDRUK Gateway. |
Study outputs
| Output type | Details | Date created | Date added | Peer reviewed? | Patient-facing? |
|---|---|---|---|---|---|
| HRA research summary | 28/06/2023 | No | No |
Editorial Notes
19/10/2022: The following changes were made to the trial record:
1. The overall end date was changed from 31/10/2022 to 31/01/2023.
2. The intention to publish date was changed from 31/10/2022 to 31/01/2023.
3. The plain English summary was updated to reflect these changes.
11/02/2022: Trial's existence confirmed by Health Data Research UK.
