Abstract
Purpose :
To assess the clinical utility and diagnostic drifts of a cloud-based referral platform for referrals from community optometrists and hospital-based eye care services.
Methods :
Retrospective cohort study conducted at Moorfields Eye Hospital, Croydon (NHS Foundation Trust, London, UK). Patients referral into the hospital eye service referral pathway by nine contributing optometrists covering a population of 90 000 individual were analysed (January 2018 to December 2019). Main outcome measures were the analysis of diagnostic entities, and the diagnostic drifts of referrals over the 2-year period.
Results :
A total of 203 referrals were analysed retrospectively. In 2018, a total of 98 patients were reviewed; 83.7% were graded as having none, mild, or moderate diagnoses and 16.3% were graded as acute diagnoses/requiring referral and timely appointments in hospital eye services. 2019 saw a total of 105 patients seen in community optometry practices who were referred using the cloud-based referral pathway. A reduction of patients none or mild/moderate eye conditions was observed (71.4%), and an increase in acute diagnoses and those requiring referral referrals to 28.6%. Notably, a marked changed was observed in “potential” wet age-related macular degeneration referrals, that were reduced from 42% to 31% from 2018 to 2019. Sight-threatening diabetic eye disease, the second most common reason for referral, was largely unchanged, from 25.4% in 2018 to 20/5% in 2019. There was an increase in “other” pathologies being referred; 32.6% (2018) to 48.5% (2019). The latter included more complex pathology and ranged from conditions such as previously undetected pattern dystrophies to urgent referrals such as papilloedema.
Conclusions :
Diagnostic drifts are a reality in telemedicine-based referral pathways and should play a role in the maintenance of a high quality and safe systems. Continual audits of high quality, structured, and linked clinical and imaging data will facilitate early and accurate diagnoses; imperative if treatment is to be effective. These data may further contribute to the development of robust analytical models to develop a methodology for further understanding of drift detection and characterization in the field of healthcare.
This is a 2020 ARVO Annual Meeting abstract.