Use of Teleophthalmology for Evaluation of Ophthalmic Emergencies by Ophthalmology Residents in the Emergency Department

Introduction

Emergency departments (EDs) face a consistently increasing burden of patient volume, resulting in crowding and delays.¹ Between 2006 and 2011, eye-related problems accounted for 12 million of the ED visits taking place in the United States. Almost 45% of these visits were due to nonemergent causes, highlighting the importance of accurate and efficient triage, diagnosis, and management of these patients in already crowded EDs. Many specialties have introduced telemedicine services as one approach to reduce the growing burden on EDs and consulting physicians.² The role of telemedicine will likely continue to grow with improvements in technology and the consistent need for ED services.³

Utilization of telemedicine services in the ED may benefit education of resident and fellow trainees by improving diagnosis accuracy, practice-based learning, and medical knowledge (MK). Previous studies have shown that dermatology residents,^4–6 emergency medicine residents,⁷ and stroke fellows⁸ believe telemedicine to be an effective educational resource with both residents and patients satisfied with their experience. Another study showed that telemedicine services have a higher self-perceived educational benefit among health care professionals who are younger, such as resident physicians and attending physicians with <10 years of experience, compared to those with more experience.⁹

Multiple studies have shown that teleophthalmology is an effective method to diagnose patients with retinal disease, glaucoma, external eye trauma, and others.¹⁰ The Fundus Photography versus Ophthalmoscopy Trial Outcomes in the Emergency Department (FOTO-ED) study showed that nonmydriatic fundus photography was not only possible in the ED but also was used more often than direct ophthalmoscopy by ED physicians and could accurately detect relevant abnormal findings.¹¹ However, there are few studies that have evaluated the efficacy of teleophthalmology on resident education. This aspect is important as evaluation of ED patients is an important component of ophthalmology training. One small study from Tel-Aviv University showed that residents using a slit-lamp connected to a video camera combined with a telephone to communicate with a senior physician was effective and accurate in diagnosing patients,¹² but the impact of teleophthalmology on resident education merits further exploration. The purpose of our study was to describe the initial implementation of a teleophthalmology tool in the ED and its impact on resident education.

Methods

The first part of this study was a retrospective review of all patients on whom an ophthalmology consult was placed between December 1, 2019, and December 1, 2020, at the Johns Hopkins Hospital (JHH) ED. The patients were split into those who did and did not undergo imaging by an ophthalmology resident physician on the Topcon 3D Optical Coherence Tomography (OCT)-1 Maestro System (Topcon Medical Systems, Inc., Oakland, NJ). The second part of this study was a survey of ophthalmology residents using an online platform (Qualtrics, Provo, UT). The study protocols were submitted to the Institutional Review Board (IRB) of the Johns Hopkins University School of Medicine. The retrospective review was approved by the IRB, while the online survey was deemed exempt from IRB review. All study activities adhered to the tenants of the Declaration of Helsinki and followed all relevant local and national privacy laws (such as the Health Insurance Portability and Accountability Act [HIPAA]) and institutional guidelines. In 2018, the approximate cost of the Topcon 3D OCT-1 Maestro System with ImageNet 6 and All in One computer was $40,000.

The Wilmer Eye Institute (WEI) Eye Trauma Center (ETC) is located in the ED at the JHH, a tertiary urban academic medical center in Baltimore, Maryland, and is a designated ETC in the region. The WEI ETC is affiliated with an accredited ophthalmology residency training program, training five resident physicians per year. Ten first-year ophthalmology residents were trained on use of the device for capturing OCT images and photos of the retina and were asked to utilize the device in their evaluation of patients presenting to the ED with urgent eye complaints. Five residents were surveyed in October 2020 after completing their first year, while five additional residents were surveyed in February 2021 during their first year. Use of the device was at the discretion of the resident based on the clinical scenarios for which such images would be useful for diagnosis and clinical management, such as measurement of optic nerve thickness or documentation of retinal lesions. Images were exported to Zeiss Forum (Carl Zeiss AG, Oberkochen, Germany) after acquisition. Examples of images can be seen in Supplementary Figure SF1. Residents communicated their findings remotely with the on-call supervising ophthalmologist, who was either an attending physician or fellow.

Retrospective chart review was conducted to obtain information on demographics, visual acuity (VA), time of arrival to the ED, time of ophthalmology consult in the ED, the patient’s final disposition, and final diagnosis or reason for workup as documented by the ophthalmology resident. Images were extracted from the device and reviewed by two independent graders (authors Y.S.S. and M.J.F.) for subjective image quality on a scale of 1 (lowest) to 3 (highest). A score of 1 indicated poor quality with no or almost no findings observable (with or without significant glare or visual artifact), a score of 2 indicated fair quality with some findings observable, and a 3 indicated good quality with all findings observable. An intraclass correlation coefficient (ICC) was calculated to compare the inter-rater reliability of image quality scores. Moderate agreement was defined as an ICC of 0.5–0.75 with good agreement between 0.75 and 0.9.¹³ In addition, a survey (Supplementary Table A) was administered to the residents asking them to rate the features of the device based on ease of use, technical reliability, quality of images, and educational value. Responses were either on a scale of 1 (least favorable) to 10 (most favorable) or a scale of “Strongly Disagree” to “Strongly Agree,” which were then recoded to a 1 to 5 scale, respectively.

Primary outcomes that we assessed included the most common diagnoses or reasons for workup that the device was used for and resident ratings of educational value. Secondary outcomes were ratings of image quality and residents’ subjective ratings of device features, such as technical reliability and ease of use.

Results

There were a total of 1,715 ophthalmology consults in the adult ED. Out of those, the device was used in 109 patient encounters by 10 residents to assess 107 unique patients, which were included in our study.

Patient Characteristics

Demographics of patients who underwent imaging with the Topcon are summarized in Table 1. Patients in our cohort had an average age of 48.5 years (±17.2, range 17–90), and 59.6% were female. In terms of race, 43.1% (n = 47) of patients were White, 38.5% (n = 42) were Black, 8.3% (n = 9) were Hispanic/Latino, 4.6% (n = 5) were Asian, 0.9% (n = 1) were American Indian or Alaskan Native, and 4.6% (n = 5) were other. The average logMAR VA of patients upon presentation to the ED was 0.50 (Snellen 20/63, standard deviation [SD] 0.8, range 0–3).

Trends in Utilization of Teleophthalmology Device

The ophthalmic conditions and diagnoses present in patients who were imaged using the teleophthalmology device are listed in Table 2. The most common reasons for utilization of the device were papilledema (n = 21, 18.6%), new-onset VA or visual field (VF) defects that did not result in a specific diagnosis (n = 12, 10.6%), retinal tears or detachments (n = 8, 7.1%), and traumatic eye injury (n = 8, 7.1%). Other common reasons for use of the device were suspicion for an autoimmune disorder (n = 7, 6.2%), diabetic retinopathy (n = 7, 6.2%), and nondiabetic retinopathy or retinal lesion (n = 5, 4.4%).

Length of Stay and Disposition

Patients who were imaged with the Topcon had an average length of time in the ED of 13.1 h (±8.6, range 3.6–54.4). The average time from ophthalmology consult to ED discharge was 9.3 h (±7.6, range 0.3–51.8). In terms of disposition from the ED, 67.9% of patients (n = 74) were discharged, 13.8% (n = 15) were admitted, 12.8% (n = 14) underwent hospitalized observation, 1.8% (n = 2) eloped, 0.9% (n = 1) left against medical advice, 0.9% (n = 1) was sent to the ophthalmology department, and 2 (1.8%) patients’ dispositions were not available.

Patients who were not evaluated with the Topcon had an average time of 7.9 h from ophthalmology consult to ED discharge (SD 7.3, range 0.7–92.4). The time from ophthalmology consult to ED discharge was similar between patients who were and were not evaluated with the Topcon (9.3 vs. 7.9 h, p = 0.07). Among all patients who were consulted for an evaluation for papilledema (n = 135), 21 underwent imaging with the Topcon (15.6%). Patients who were evaluated for papilledema with and without the Topcon had no statistical difference in length of time between ophthalmology consult to ED discharge time (10.2 vs. 11.7 h, p = 0.39).

Image Quality

A total of 887 images were taken using the device during the study period for the 109 patient encounters. There was an average of 8.1 images/patient. Between the two graders, the average image was rated at a 1.79 on a scale of 1–3, with 1 being the worst and 3 being the best. Grader 1 (M.J.F.) had an average image rating of 1.80 and Grader 2 (Y.S.S.) had an average image rating of 1.78. When only the highest quality image from each patient was considered, the overall average image quality was 2.24 with Graders 1 and 2 averaging 2.23 and 2.26, respectively. The inter-rater reliability measure was moderate to good with an ICC of 0.75 (95% confidence interval 0.72–0.78, p < 0.001).

Resident Survey Responses

Eight out of 10 residents who used the device completed the survey (response rate 80%). Results of the survey are shown in Table 3. Residents reported that they most often utilized the fundus photo of the disc or macula (n = 7), followed by macular OCT (n = 6). All residents shared images with a supervising ophthalmologist. On a scale of 1 (least favorable) to 10 (most favorable), the mean resident rating for the ease of use was 6.4 (±2.4), technical reliability at a 5 (±2.8), and quality of images at a 7.3 (±2.1). Most residents (n = 7) rated “Agree” or “Strongly Agree” with the statement “The Topcon device allowed me to diagnose patients more accurately.” Half of the residents (n = 4) rated “Agree” or “Strongly Agree” with the statement “The Topcon device is easy to use.”

There were several obstacles that residents reported with the device. Issues intrinsic to the device included an inability to log in, being unable to find a patient in the work list after placing an order, machine malfunction, image artifacts, and an inability to capture an image. There were also several issues with connecting the device to connected platforms and services.

Residents reported issues linking the device to the institutional clinical information systems, which provided a list of patient demographics. Even when images were successfully uploaded, they sometimes had a “blue hue.” Residents also reported issues connecting the device to the standalone computer that runs the image capture and upload software. While these issues could sometimes be resolved after closing out of the program and logging in again, it added extra time to the process.

Discussion

The most common use of teleophthalmology in the ED setting was to rule out an acute neuro-ophthalmologic or retinal concern. Teleophthalmology was by far most commonly used to evaluate for papilledema, likely due to the necessity of viewing the optic nerve and quantifying the extent of swelling that has occurred to avoid the devastating sequela of blindness that may occur if not treated in a timely manner.^14,15 The next most common condition for which the device was used was a new-onset VA or VF defect. Since this symptom can be relatively nonspecific, use of a fundus examination and OCT imaging were likely helpful to residents to rule out any acute pathology (such as a retinal tear or detachment) in patients with new-onset VA or VF defects.¹⁶ Pathology of the retina, such as a retinal tear/detachment, diabetic retinopathy, and nondiabetic retinopathy, as well as workup for a traumatic injury were also quite common in our cohort, again likely due to their acuity, the utility of fundus examination and OCT imaging for diagnosis,¹⁶ and the possibility of worsening vision if not appropriately treated. Other conditions such as an autoimmune workup, nondiabetic retinopathy, optic neuropathy, and a mass/tumor were less common. For conditions that need to be immediately ruled out, teleophthalmology shows promise, as residents can promptly show patient findings to supervising ophthalmologists and reduce the burden on consultant attendings. As conditions affecting the optic nerve and retina are often acute in nature and require a fundus examination or OCT imaging, this device shows great value in being incorporated into the ED setting.

Previous studies have shown success in using nonmydriatic fundus photography to diagnose specific conditions, such as diabetic retinopathy,¹⁰ retinopathy of prematurity,¹⁰ age-related macular degeneration,¹⁰ glaucoma,¹⁰ and trauma. The FOTO-ED study also showed the feasibility of nonmydriatic fundus photography and teleophthalmology in an ED setting, diagnosing conditions such as grade III/IV hypertensive retinopathy, optic disc edema, intraocular hemorrhage, retinal vascular occlusion, and optic disc pallor.¹⁷ Our observational study adds to the important findings of the FOTO-ED study in that we describe the overall real-world patterns of use of this device at a large academic eye referral center and determine the most common indications for this device in the future. This is important clinically, as it will allow us to streamline the most common reasons for using teleophthalmology in the ED setting among both resident and attending physicians.

One previous study implemented teleophthalmology by using a camera attached to a slit lamp microscope.¹² Residents took pictures with the camera and sent them to a remote supervising ophthalmologist. They found this was helpful in the diagnosis of disorders affecting the ocular surface, anterior segment, pupils, optic nerve, and macular disorders. Resident diagnoses with a senior physician’s virtual consult and a senior physician’s diagnosis seeing the patient in person the next day were the same in every case, indicating good accuracy. Our study built upon this by examining the use of a nonmydriatic fundus camera as a tool for teleophthalmology and surveying resident’s perceived educational benefits.

In addition, in light of the coronavirus disease-2019 (COVID-19) pandemic, patients and physicians have developed a renewed interest in teleophthalmology,¹⁸ with a number of potential applications being developed to reduce the number of health care workers interacting with patients.¹⁹ Ophthalmology practices have increased the number of services provided remotely,^20,21 although there are challenges with replicating the entire ophthalmic examination virtually.^21,22 However, devices such as the one used in this study have the promise of improving communication between providers not in the same physical location by providing images to the consulting physician rather than a verbal description of examination findings.

In terms of time in the ED, using the device did not correlate with a longer length of stay, although the difference did trend toward significance. This may be because more complex patients were more likely to undergo imaging and more likely to have a longer length of stay. Patients were therefore also evaluated by diagnosis. Patients who were being evaluated for papilledema, the most common diagnosis in our cohort, also did not have a difference in the length of stay, implying the usefulness of this device without increasing time for patients.

In previous investigations of teleophthalmology applications, images were obtained by attending physicians or nurse practioners,^10,17 but our study is the first to evaluate resident-obtained images and the impact of teleophthalmology on their educational experience. First year ophthalmology residents are very early on in the development of their ophthalmologic physical examination and photography skills. The majority of residents stated that teleophthalmology helped them more accurately diagnose patients and indicated they would use the technology again. The Accreditation Council for Graduate Medical Education (ACGME) mandates several core milestone domains for all ophthalmology residents to improve their clinical competency and education.²³ In this study, residents engaged in the core milestone domains of practice-based learning and improvement (PBLI), patient care (PC), and MK, which likely contributed to their self-perceived educational benefit. Practice-based learning may be especially relevant in acute neuro-ophthalmic and retinal cases. In most academic centers, residents take most of the call and there is a learning curve to visualize the fundus. Nonmydriatic fundus photography allows a resident to quickly confirm or deny their physical examination findings, such as when quantifying the extent of optic disc swelling, thus engaging in the important educational domains of PBLI and MK. Teleophthalmology also likely improved the accuracy of a resident’s diagnosis due to faster communication with attending physicians, allowing for asynchronous viewing of the examination findings rather than hearing a description. We hypothesize that this benefits resident education due to the ability to discuss a differential with the attending immediately, rather than having to wait for the superior to come examine the patient themselves, thus contributing to both PC and MK competency domains.

We also explored residents’ individual experiences with the device. On average, images were poor to fair quality, but the best image of each patient encounter was of better quality, usually fair to good. This may be because it took residents several tries to get a high-quality image of the patient’s eye due to artifact or poor positioning, which would also explain the high number of images taken per patient. There were a number of logistical issues with the Topcon device, such as artifacts in the images and images appearing different once uploaded compared to when they were taken on the device. There were also several issues connecting the device to the computer, electronic health record (EHR), and the platform used to view ophthalmic images. While these issues may be specific to the camera device itself, the computer attached to the device, the imaging software used to capture images, or individual problems with our hospital’s information technology system, they greatly impact the utility of the device as success of teleophthalmology requires communication between different devices and platforms. In an acute setting in which a resident wants to quickly communicate patient findings to an attending, technical issues may hinder the process, take up valuable time, and render teleophthalmology less useful. Therefore, it is important to resolve these technical issues before expanding the use of this device in the ED setting.

Resident survey responses did indicate that the device was comfortable to use for patients. Furthermore, there was no statistical difference in the time to discharge from ophthalmology consult based on whether the device was used. This may be beneficial in the future to speed up ophthalmologic diagnosis and ease the burden of overcrowding on the ED.

We present a full year of data on the most common uses of teleophthalmology with a single device at a busy tertiary academic eye referral center, with over 100 patient encounters. Some limitations of this study include its retrospective observational nature, which does not allow us to randomize the use of teleophthalmology and determine the quantitative impact on resident education. This study evaluated patients from a single tertiary care center, which may limit generalization of most common uses. Based on the findings from this study, we believe future studies should include a randomized study of certain diagnostic categories, with assessment of presurveys and postsurveys to quantify residents’ educational attainment, patient satisfaction, and ED performance metrics.

Conclusions

The most common use of teleophthalmology in the ED setting was for papilledema and the majority of trainees perceived an educational benefit to teleophthalmology. While the images obtained from the Topcon device were of fair to good quality, there were a number of technical issues that have not been previously reported in the literature. Addressing the technical challenges may increase the utility of this tool in the ED setting for both PC and resident education.

Disclosure Statement

M.V.B. receives is a paid consultant with Carl Zeiss Meditec. All other authors have no financial disclosures to report.

Funding Information

This study was supported by an unrestricted departmental grant to the Wilmer Eye Institute from Research to Prevent Blindness and a Digital Education & Learning Technology Acceleration (DELTA) grant from the Johns Hopkins University Office of the Provost.

Supplementary Material

Supplementary Figure S1

Supplementary Table A