Screening for Diabetic Retinopathy in Pharmacies With Artificial Intelligence Enhanced Retinophotography (DIABeyeIA)

Screening for Diabetic Retinopathy in Pharmacies With Artificial Intelligence Enhanced Retinophotography

Acceptability and Efficiency of Screening for Diabetic Retinopathy in Pharmacies With Artificial Intelligence Enhanced Retinophotography : a French Pilot Study

Diabetic retinopathy is frequent, potentially severe with visual threat, health costly and represents a major public health problem. However, screening compliance for retinopathy remains too low in France, approximately 40% patients with diabetes laking diabetic retinopathy screening for at least 2 years. DIABeyeIA is a prospective pilot study evaluating the effectiveness and acceptability of diabetic retinopathy screening in 11 pharmacies in Normandy (north of France) using a non-mydriatic portable retinophotograph enhanced by artificial intelligence software. The main goal of this work is to evaluate a potential increase rate of diabetic retinopathy screening, compared to the actual rate (64% in France). Secondary goals are faisability, satisfaction and economical considerations for implementation of such a new screening program.

DIABeyeIA is a 6 months prospective study in 11 pharmacies of Normandy (a french region in north of France), to assess a systematic diabetic retinopathy screening proposed to all patients with diabetes, identified on their usual treatment, who regularly visit their pharmacy. Participating pharmacies will offer to all their patients with diabetes the opportunity to benefit from this screening when they visit the pharmacy to pick-up their treatment. If the patient accepts, after written consent and completion of the clinical data form, he will benefit from diabetic retinopathy screening (Arm 1) : two retinophotographies per eye will be performed by the pharmacist and will be immediately interpreted by an artificial intelligence system through an interface made available to pharmacists. In a few seconds, the system will answer: (i) that there is no diabetic retinopathy. In this case, a letter is sent to the patient, reminding him of the modalities and french recommendations for diabetic retinopathy screening. (ii) that images are doubtful and in this case, they will be re-read by an expert ophthalmologist who will transmit his interpretation to the patient via the pharmacist. In case of threatening damage, an emergency ophthalmologist visit in the university hospital will be proposed to the patient via his pharmacist. Patients who will not meet inclusion criteria or will refuse the study will be included on the list of refusals/impossibility to participate to the DIABeyeIA study (Arm 2). In all cases, a letter will be given to the patient by the pharmacist, for his general practitioner and his usual ophthalmologist, in order to inform them of the screening and its result. Clinical and therapeutic data of each participating patient will be informed by the pharmacist. Satisfaction questionnaires will be completed by the participating pharmacists at the end of the study and by the patients at the end of the screening.

Diabetic Retinopathy, Artificial Intelligence, Telemedicine, Pharmacy, Retinophotography, Screening, Primary care

Patients who accepted to participate to this study and had retinophotography for diabetic retinopathy screening.

The screening rate for diabetic retinopathy will be assessed in the 11 participating pharmacies and then compared to literature data (actual diabetic retinopathy screening rate = 64%) The screening rate will be calculated as follows : Number of patients screened(Arm 1)/(number of patients screened (Arm 1) + number of patients who refused screening or failed screening (Arm2)). We hypothesise an improvement of 10 points (≥74%)

Cost of screening per patients will be calculated, by dividing total cost of this study by the number of patient screened for diabetic retinopathy.

To assess the possibility to identify diabetic retinopathy lesions at earlier stages than the conventional screening strategy

To evaluate consistance between artificial intelligence and ophthalmologist re-read for doubtful images

Inclusion Criteria: Adult patients with diabetes Ongoing diabetes treatment (oral hypoglycaemic agents, injectable GLP-1 receptor agonists and insulin) Regular customer of the pharmacy (at least 3 previous visits) Informed consent to participate to the study Exclusion Criteria: Patient unable to read, write or give consent Patient refusing to share results with their general practitioner or ophthalmologist