Real-World Evaluation of the Eye+Dot Online Triage Support Tool in Community Optometry Practices: Mixed Methods Evaluation Study.

Background: Many patients attending hospital emergency services with recent-onset eye symptoms could have been managed in the community. This may reflect a lack of specialist experience or triage capacity among primary care providers. Online triage tools collect patient symptomatology and relevant ophthalmic and medical history virtually, compile a report, and suggest an outcome decision which can support the streaming and direction of patients to a suitable service within an appropriate timeframe. Objective: Our objectives were to implement and analyze the real-world use of a novel online triage tool in a community optometry setting. Methods: This was a real-world mixed methods evaluation of a web app "eye+dot" (Cambridge Medical Innovation Ltd), which uses a conditional logic multiple-choice questionnaire to produce a symptom report and automated triage recommendation. A preimplementation survey was used to gather views regarding current triage practices and interest in trialing an online triage tool among community optometrists. The app was then implemented in participating practices, with optometrists receiving the symptom report but not the automated triage disposition. The sensitivity and specificity of optometrists and the app at identifying urgent cases (those requiring specialist examination within 24 h) were compared to a gold standard, generated by retrospective review of the symptom reports by hospital specialists. Metadata including patient age, test duration, and usability feedback from patients and optometry staff were analyzed. Results: Seven optometry practices trialed the app over 5 months. Records of 209 patients with a mean age of 53.5 (SD 17.5; range 13-90) years were analyzed. Three patients with potentially sight- or life-threatening symptoms (as defined by Royal College of Ophthalmologists guidelines) received appropriate urgent hospital dispositions by both optometrists and eye+dot. The app's automated disposition had similar sensitivity to optometry triage (75.5% vs 77.3% respectively; P=.99) but significantly better specificity (82.7% vs 54.4%; P<.001) for identification of the 53 cases categorized as requiring either hospital or community specialist examination within 24 hours by gold-standard review. A total of 147 patients gave acceptability feedback, with 121 (82%) of them giving a rating in the top 2 points of a 5-point visual Likert scale. Optometrists in 5 of the 7 practices found the app helpful and were interested in continuing its use. Conclusions: This real-world study demonstrates the triage tool's accuracy for identifying high-risk symptomatology and its acceptability among patients and optometry services. The significantly lower optometric triage specificity may reflect risk-averse decision-making following direct communication with patients and appointment availability in primary care. In addition, the relatively low mean age of patients suggests possible barriers to uptake among older patients. This and other symptom-based triage tools can provide useful support to eye care providers in the community where access to specialist triage may be limited.