Trial no.:	PACTR202101512465690	Date of Approval:	13/01/2021
Trial Status:	Registered in accordance with WHO and ICMJE standards

TRIAL DESCRIPTION

Public title

Diabetic Retinopathy Screening Using Retinal Imaging and Automated Grading with Artificial Intelligence in Rwanda- A Randomized Control Trial

Official scientific title

Rwanda Artificial Intelligence Diabetic Retinopathy Screening Study (RAIDERS)

Brief summary describing the background and objectives of the trial

Blindness from Diabetic retinopathy (DR) has become another addition to the myriad demands on eye health systems in low resource countries such as Rwanda. Diabetic retinopathy (DR), a complication of diabetes, is the fastest-growing cause of vision loss in working-age adults. There is currently no national screening program for DR in Rwanda and people with DR are identified through sporadic screenings or on presentation to eye clinics due to vision loss. Early detection of DR through screening can prevent visual impairment and blindness. Screening however requires an abundance of trained eye health personnel of which there are very few in Rwanda, to grade images and provide feedback to the patients, The delayed feedback to patients about their DR status results in poor uptake of referral for interventions to treat the eye complications Artificial intelligence (AI) based grading of retinal images to identify diabetic retinopathy is a novel way of identifying people with diabetes who are at risk of blindness faster and using fewer human resources. Few countries have deployed and evaluated such a system in clinical practice. AI grading allows immediate feedback about disease grade to patients. The objective of this trial is to examine whether immediate feedback using AI grading of retinal images will increase uptake of referral to care following screening for DR. Patients with diabetes at three recruitment sites will be allocated to either immediate feedback using AI or normal delayed feedback in 3-5 days using human grading. We will measure the uptake of referral for treatment in the AI arm vs the human grading arm within 4 weeks of receiving appointment information.

Type of trial

RCT

Acronym (If the trial has an acronym then please provide)

RAIDERS

Disease(s) or condition(s) being studied

Eye Diseases

Sub-Disease(s) or condition(s) being studied

Purpose of the trial

Early detection /Screening

Anticipated trial start date

01/02/2021

Actual trial start date

31/03/2021

Anticipated date of last follow up

31/12/2021

Actual Last follow-up date

31/08/2021

Anticipated target sample size (number of participants)

250

Actual target sample size (number of participants)

273

Recruitment status

Completed

Publication URL

Pending registration of the trial protocol at OSF (Open Science Foundation) web site

Secondary Ids	Issuing authority/Trial register

STUDY DESIGN
Intervention assignment	Allocation to intervention	If randomised, describe how the allocation sequence was generated	Describe how the allocation sequence/code was concealed from the person allocating the participants to the intervention arms	Masking	If masking / blinding was used
Parallel: different groups receive different interventions at same time during study	Randomised	Simple randomization using by using procedures such as coin-tossing or dice-rolling	Sealed opaque envelopes	Masking/blinding used	Outcome Assessors

INTERVENTIONS
Intervention type	Intervention name	Dose	Duration	Intervention description	Group size	Nature of control
Experimental Group	Immediate Diabetic retinopathy grade feedback using Cybersight Artificial intelligence platform	Feedback on DR grade will be communicated in the form of a colour coded printout and explained further to the patient by a study counsellor. This will be done before the patient exits their regular diabetes care clinic. The report will include an interpretation as refer or no referral needed.	The AI diagnoses will be communicated to the patient immediately as a one time intervention while outcomes will be measured within 4 weeks of receiving the report and referral information.	Intervention arm - Once imaging is completed ALL images will be uploaded to Cybersight. Each uploaded image will be anonymized with a patient registration number. Cybersight will in turn give each a unique ID. The Unique IDs generated by Cybersight will be used to randomize and allocate which arm the patient belongs to offsite. The reading centre and on site investigators will be unaware of the allocation Cybersight will then send all images to both AI and to Labelbox (for onward forwarding to human grading). The Cybersight AI will return reports for patients for all recruited patients the appropriate feedback form within 60s. Only those with a Cybersight refer diagnoses of Refer positive will be enrolled in the trial. Those randomized to the intervention arm will receive the AI report as a printout immediately. For those randomized to the comparator arm the AI grading report will be stored. The patients will await the human grading report which will be received in 3-5 days The primary outcome that will be assessed is attendance to the scheduled examination, defined as the proportion of patients with DR requiring referral in each study group that presents for ophthalmology review as recommended within 4 weeks of receiving their referral diagnosis and appointment. For those graded as no DR (refer negative)an appointment on a specific date 6 months from the date of examination. They will not join the trial. The patient will be informed that all costs of travel to the ophthalmology clinic will be reimbursed and a voucher to claim the reimbursement will be handed to the patient together with his report and appointment card	125
Control Group	Delayed feedback after human grading of diabetic retinopathy status at a reading centre	Feedback will be provided to these patients after the human graders have confirmed the DR status. This report is expected to reach the patient within 3 to 5 days by telephone call and text messages. An SMS message will be sent to the patient followed by a voice call 24 hours later to confirm receipt of the SMS. The patient will receive a phone call from the same research assistant who saw the patient at the clinic, who will then explain to the patient the need to see an ophthalmologist on any working day within 4 weeks from the day of receiving their report, for further examination and treatment.	At the screening site, the research assistant will inform patient from the comparator arm that the grading report of his diabetic retinopathy will be received within 3 to 5 days and the outcome will be communicated by SMS and phone call. The outcome measurement (attendance at the referral clinic) will be measured within 4 weeks of the patient receiving their grading report and referral appointment	Comparator arm - After retina imaging, the images will be uploaded to Cybersight AI for grading as well as to Label Box for transmission to the human grading reading centre. Only those who receive an AI refer positive report will be enrolled in the trial. The AI report will be stored but will not be communicated to the patient. At the reading centre, human grading will be carried out, by a fully trained retina expert and one UK NHS grader with adjudication. The grading report is expected to be completed and communicated to the researchers within 3-5 days for those with and those without disease. This human grading report will be communicated to the patient via text , followed by a phone message 24 hours later. This represents current practice and this the placebo intervention. This phone call will come from the same ophthalmic worker who will then explain to the patient the need to see an ophthalmologist within 4 weeks for further examination and treatment. A telephone appointment will thus be given. The appointment period will remain open for any working day in the the four weeks following receipt of the report. The primary outcome that would be assessed is having come to the scheduled examination, defined as the proportion of patients with DR requiring referral in each study group that presents for ophthalmology review as recommended within 4 weeks of receiving a diagnosis and appointment. Those who are disease negative will be given a specific appointment date at a 6 month point from the date of examination and will remain open for one month. They will not be enrolled in the trial. Patients in the Control group who cannot be successfully e-contacted by short message services (SMS) or subsequent phone call (see below for protocol) will be included in the denominator as requiring referral. The patient will be informed that all costs of travel to the ophthalmology clinic will be reimbursed and a voucher to claim the reimbursement will be sent by SM	125	Active-Treatment of Control Group

ELIGIBILITY CRITERIA
List inclusion criteria	List exclusion criteria	Age Category	Minimum age	Maximum age	Gender
Recruitment: Consecutive patients with confirmed Type 1 or Type 2 diabetes diagnoses who are attending at a participating diabetes clinic at three sites in Rwanda will receive a patient information leaflet with all the details of the trial while in the waiting room of their diabetes clinic appointment. In addition, a researcher nurse will be available to explained verbally the details of the trial and answer any questions. At the end of their diabetes consultation, their regular clinician will ascertain whether they wish to participate in the trial. If they express an interest, or needed further details, a trained researcher will give further explanation and on reaching agreement to participate, written informed consent will obtained. The individuals will then be enrolled to undergo screening for diabetic retinopathy using retinal imaging. All images will then be uploaded to Cybersight AI for interpretation. ONLY THOSE WHO GET A REFER POSITIVE AI REPORT WILL BE ENROLLED INT THE RCT Summary of inclusion criteria: 1)Adults, young people and children who are aged ≥18years. 2) Are Registered with a participating diabetes clinic . 3) Have gradeable digital retinal images in at least one eye. 4) Give their informed consent for participation. 5) Can travel to the referral clinic at RIIO 6) Are not involved in any ongoing treatment or trial investigating a treatment with ongoing appointments for care 7) For inclusion into the trial: Receive a refer positive Cybersight AI report	Exclusion criteria will be: 1) Patients who do not think that they can return for follow-up e.g. prisoners or visitors to Rwanda ; 2)Adults, young people and children who are aged under age 18years. 3)Are not registered with a participating diabetes clinic. 4) Are ineligible for screening for whatever reason, including inability to be imaged, having ungradable digital retinal images, or have both eyes ungradable eye with no visual potential. 5) Do not give consent for participation in the RCT. 7)Are not involved in any ongoing treatment or trial investigating a treatment with ongoing appointments for care 8) Those who receive a normal Cybersight AI report (Refer negative) will not be enrolled into the trial All these patients will receive appropriate clinical care but will not be included in the trial	80 and over: 80+ Year,Adult: 19 Year-44 Year,Aged: 65+ Year(s),Middle Aged: 45 Year(s)-64 Year(s)	18 Year(s)	100 Year(s)	Both

ETHICS APPROVAL

Has the study received appropriate ethics committee approval

Date the study will be submitted for approval

Date of approval

Name of the ethics committee

Yes

02/11/2020

Rwanda National Ethics Committee Ministry of Health

Ethics Committee Address
Street address	City	Postal code	Country
Kigali	Kigali	PO Box 84	Rwanda

OUTCOMES
Type of outcome	Outcome	Timepoint(s) at which outcome measured
Primary Outcome	Primary Outcome Measurement - The primary outcome is the attendance rate for treatment and further care within 4 weeks of receiving appointment information in the two arms of the study. Non-attendance is defined as failure to attend any treatment appointment within 4 weeks of being invited for follow up. This outcome will be measured only among those who are disease positive as they are the ones who will have a referral within 4 weeks.	Within 4 weeks of receiving appointment information
Secondary Outcome	Patient acceptability and satisfaction measurements of delayed human grading vs immediate grading by AI including perception of impact on time in clinic of including DR screening will bemeasured among providers and patients using a questionnaire designed for the study	At exit from screening and at completion of referral visit
Secondary Outcome	Secondary Outcome Measurements - Diagnostic test accuracy of Cybersight AI platform and human graders diagnoses for identifying any DR needing referral, and for identifying sight-threatening or vision-threatening DR (STDR / VTDR) ; i.e., sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV) and area under the receiver operating characteristic curve (AUC of ROC); and inter-rater agreement compared to gold standard (i.e., adjudicated response agreed to between the retina specialist and the UK NHS grader is the gold standard).	End of the study
Secondary Outcome	Inter-rater agreement of level of gradeability of the images as identified in Cybersight AI platform compared to the adjudicated human graders’ grading.	After 4 weeks

RECRUITMENT CENTRES
Name of recruitment centre	Street address	City	Postal code	Country
Rwanda Diabetes Association	KN 8 Ave, No 27 Kinamba	Kigali	PO Box 22	Rwanda
University Teaching Hospital of Kigali	KN 4 Ave	Kigali	Bp. 655	Rwanda
Legacy Clinics Private Hospital	KK 3 Road	Kigali	PO 1589	Rwanda
Rwamagana District Hospital	Kigali-Kayonza Road	Rwamagana		Rwanda
Nyamata District Hospital	Kinazi	Nyamata		Rwanda
Rwanda Military Hospital	KK 739 Street Kanombe, Kicukiro District	Kigali	3377	Rwanda

FUNDING SOURCES
Name of source	Street address	City	Postal code	Country
ORBIS International	520 8th Ave 12th Floor	New York	10018	United States of America

SPONSORS
Sponsor level	Name	Street address	City	Postal code	Country	Nature of sponsor
Secondary Sponsor	Rwanda International Institute of Ophthalmology	2nd Floor MTn Centre KG 9 Avenue	Kigali	POBox312	Rwanda	Charities/Societies/Foundation

COLLABORATORS
Name	Street address	City	Postal code	Country
Dr Damas Kabakambira	University Teaching Hospital of Kigali	Kigali	BP 655	Rwanda

CONTACT PEOPLE
Role	Name	Email	Phone	Street address
Principal Investigator	Wanjiku Mathenge	ciku@email.com	0025788384440	No 15 KG 29 Avenue
City	Postal code	Country	Position/Affiliation
Kigali	POBOX7088	Rwanda	Rwanda International institute of Ophthalmology and Orbis International
Role	Name	Email	Phone	Street address
Public Enquiries	John Nkurikiye	nkurikiye.john@gmail.com	0025788307810	RIIO MTN Centre KG 9
City	Postal code	Country	Position/Affiliation
Kigali	POBOX312	Rwanda	Rwanda International Institute of Ophthalmology
Role	Name	Email	Phone	Street address
Scientific Enquiries	Nathan Congdon	nathan.congdon@orbis.org	00442890976350	Centre for Public Health - Queens University - Institute of Clinical Science - Block A - Royal Victoria Hospital
City	Postal code	Country	Position/Affiliation
Belfast	BT126BA	United Kingdom	Queens University Belfast and Orbis International

REPORTING
Share IPD	Description	Additional Document Types	Sharing Time Frame	Key Access Criteria
Yes	Individual participant data that underlie the results and outcomes that will be presented in the publications will be available under a public domain after ‘de-identification’.	Informed Consent Form,Study Protocol	The data will be available immediately after publications without an end-date for a free domain or 5 years for a paid domain or a third party website.	-Investigators whose proposed use of the data has been approved by an independent review committee will be identified for this purpose. -The requests or proposals should be submitted to the principal investigator. -To gain access, data requestors will need to sign a data access agreement with the current trial investigators. -Access to the Cybersight AI web application used for this study can be requested. -Authors of secondary analyses using shared data must attest that their use was in accordance with the terms (if any) agreed to upon their receipt. -They must also reference the source of the data using its unique, persistent identifier to provide appropriate credit to those who generated it and allow searching for the studies it has supported. -Authors of secondary analyses must explain completely how theirs differ from previous analyses. -In addition, those who generate and then share clinical trial data sets deserve substantial credit for their efforts. -Those using data collected by others should seek collaboration with those who collected the data. -As collaboration will not always be possible, practical, or desired, the efforts of those who generated the data must be recognized.
URL	Results Available	Results Summary	Result Posting Date	First Journal Publication Date
(Pending registration at the OSF - Open Science Foundation site)	No
Result Upload 1:	Result Upload 2:	Result Upload 3:	Result Upload 4:	Result Upload 5:

Result URL Hyperlinks	Link To Protocol
Result URL Hyperlinks

Changes to trial information

Pan African Clinical Trials Registry