Author's Reply.

Abstract

We thank Harper et al. for their interest in our paper. We agree that the safety of commissioned services is of importance, and we believe that novel services should be commissioned with inbuilt service evaluation, but we acknowledge the inherent barriers to doing this. We were grateful that the companies involved in our study were keen to have independent evaluation that enabled system learning and improvement of services. As Harper et al. note, clinical record abstraction has some downsides. This has further disadvantages for an acute eye service appointment for symptomatic individuals. For example, a practitioner could record no tear/detachment for a patient presenting with flashing lights and/ or floaters which, according to the records, the practitioner may appropriately decide not to refer the patient onwards for treatment. This does not tell us, however, whether the patient had a retinal tear or detachment that was simply not detected. It is likely that a practitioner's clinical record would justify their choice of action. Direct patient contact, therefore, may be an appropriate method of determining major errors in an acute eye care service. We therefore welcome Harper et al. bringing a discussion of the challenges in appropriately documenting the safety of primary eye care services. In addition to the lack of falsenegative data being an issue when determining the safety of a service, this also poses questions for the system's costeffectiveness. Despite work attempting to elucidate the costeffectiveness of optometryled acute eye care services, 5 without knowing the precise details of the erroneous management (e.g., falsenegatives), true costeffectiveness cannot be ascertained. For example, mismanagement of patients with early symptoms of stroke has the potential to incur a significantly larger cost to the health care system and society, relative to a falsepositive hospital referral. The lack of specifics in the methodology of Williams et al. makes comparison to other studies difficult. However, the results of a teleophthalmology service delivered at Moorfields Eye Hospital are interesting and in line with our study, when examining representation rates. Specifically, 35.5% of patients who had a teleconsultation by an ophthalmologist represented, unplanned, to the emergency eye service within one month of their initial consultation. For patients in our study, 26.8% reattended (11.1% to the phone line and 15.7% elsewhere) after teleconsultation. In both studies, rates of representation were lower after a facetoface appointment. Specifically, 19.8% fewer patients represented after a facetoface ophthalmology appointment (15.7% represent rate) and 26% fewer patients represented after a facetoface optometry appointment (0.7% represent rate). Whilst the teleophthalmology service differed markedly from that of the service in our study, one distinction is that in the service examined by Li et al., patients were able to choose whether they were seen in person, or virtually. Accordingly, it may be that patients who chose teleconsultation thought their condition was less serious, relative to those who opted to attend in person which reduced instances of actual harm. This is also in concordance with patient centred care. In the optometry teleconsultations examined by both Williams et al. and ourselves, all patients had a telephone consultation before the clinician determined whether a facetoface appointment was necessary. An important aspect of recent studies through the COVID19 pandemic is that the data suggests that facetoface emergency eye care delivered by optometrists is safe with low rates of adverse events. Ansari et al. reported low rates of referral to hospital ophthalmology departments (5%), with all patients retained in primary care managed to the resolution of their condition by independent prescribing optometrists. Our study revealed only two patients (0.4%) who were seen in person where their condition deteriorated that wasn't a ‘major’ error (worsening dry eye type symptoms). In addition, there was a single patient defined as a ‘major’ error (0.2%), namely a case of recurrent corneal epithelial erosion misdiagnosed as conjunctivitis. Our study reported higher rates of referral to hospital ophthalmology departments (27% of facetoface consultations, 14.8% of all consultations) than Ansari et al., which is unsurprising given the differences in qualifications and scope of practice of the optometrists delivering facetoface care between the studies. This supports the integration of independent prescribing into such services. Reassuringly, 85% of referrals from facetoface consultations in our study were determined to be truepositives. Ultimately, we believe that digital integration between community and hospital eye care, with a minimum data set and independent systemwide audit and service evaluation is necessary to ensure patients in both primary and secondary care are not coming to harm. Whilst our paper and this response focus mainly on the safety of primary care, this should be balanced against secondary care, including, but not limited to, the impact of waiting lists and potential delays in follow up.