Point-of-Care Ultrasonography in the Diagnosis of Retinal Detachment, Vitreous Hemorrhage, and Vitreous Detachment in the Emergency Department

IMPORTANCE Ocular symptoms represent approximately 2% to 3% of all emergency department (ED) visits. These disease processes may progress to permanent vision loss if not diagnosed and treated quickly. Use of ocular point-of-care ultrasonography (POCUS) may be effective for early and accurate detection of ocular disease. OBJECTIVE To perform a large-scale, multicenter study to determine the utility of POCUS for diagnosing retinal detachment, vitreous hemorrhage, and vitreous detachment in the ED. DESIGN, SETTING, AND PARTICIPANTS A prospective, observational diagnostic/prognostic study was conducted at 2 academic EDs and 2 county hospital EDs from February 3, 2016, to April 30, 2018. Patients who were eligible for inclusion were older than 18 years; were English- or Spanish-speaking; presented to the ED with ocular symptoms with concern for retinal detachment, vitreous hemorrhage, or vitreous detachment; and underwent an ophthalmologic consultation that included POCUS. Patients with ocular trauma or suspicion for globe rupture were excluded. The accuracy of the ultrasonographic diagnosis was compared with the criterion standard of the final diagnosis of an ophthalmologist who was masked to the POCUS findings. Seventy-five unique emergency medicine attending physicians, resident physicians, and physician assistants performed ocular ultrasonography. EXPOSURE Point-of-care ultrasonography performed by an emergency medicine attending physician, resident physician, or physician assistant. MAIN OUTCOMES Main outcome measures the sensitivity and specificity of presenting to the ED with ocular symptoms. detachment. Point-of-care ultrasonography is not intended to replace the role of the ophthalmologist for definitive diagnosis of these conditions, but it may serve as an adjunct to help emergency medicine practitioners improve care for patients with ocular symptoms.


Introduction
Ocular symptoms are commonly evaluated in the emergency department (ED) and comprise approximately 2% to 3% of all ED visits. 1 These presentations can be benign or can result in permanent vision loss if not quickly identified, diagnosed, and treated. Three common diagnoses encountered in the ED are retinal detachment (RD), vitreous hemorrhage (VH), and vitreous detachment (VD). Of these 3, RD is considered a true ophthalmologic emergency that requires immediate diagnosis and treatment. 2 Patients with RD may complain of sudden, painless, monocular vision loss as well as flashes and floaters in the visual field. Similar to RD, symptoms of VH and VD may include vision loss, blurry vision, and visual floaters. Distinguishing between these 3 conditions is clinically important, because patients with VH and VD can often be discharged with close outpatient follow-up, whereas patients with RD may need emergency evaluation by an ophthalmologist.
Currently, patients with ophthalmologic symptoms undergo initial testing that includes visual acuity, direct ophthalmoscopy, slitlamp examination, and tonometry. 3 However, the criterion standard for the establishment of a diagnosis of ocular diseases such as RD is an ophthalmologic evaluation. The diagnosis of ocular disease by an ophthalmologist may entail procedures such as a dilated ophthalmoscopic examination, optical coherence tomography, or ophthalmic ultrasonography. 4,5 These procedures are used to evaluate the posterior chamber of the eye and clearly visualize the distinct layers of the retina. Ultrasonography has been used by ophthalmologists for decades to evaluate ocular symptoms but has gained favor by emergency medicine practitioners. 6 Previous studies have shown that emergency medicine physicians are able to use ocular point-of-care ultrasonography (POCUS) to identify RD in the ED. 7-10 However, these studies had limitations, including small sample size, highly trained sonographers, and large CIs. The largest study thus far is a retrospective study that included 142 patients, 34 of whom were found to have RD. 11 Another large retrospective study included 115 patients, but only 16 received a diagnosis of RD. 12 To date, no large-scale, prospective, multicenter trials have been performed to evaluate the ability of emergency medicine practitioners to diagnose RD, VH, or VD using POCUS.
Our objective was to perform a large-scale, prospective, multicenter study to determine the accuracy of ocular POCUS in the evaluation of RD, VH, and VD. We compared the emergency medicine practitioners' POCUS diagnosis with the criterion standard of the attending ophthalmologists' final diagnosis.

Study Design
This study followed the Standards for Reporting of Diagnostic Accuracy (STARD) reporting guideline.
We conducted a multicenter, prospective, observational diagnostic/prognostic study using a convenience sample of patients between February 3, 2016, and April 30, 2018, who presented to the ED with ocular symptoms for which RD, VH, or VD was suspected and who underwent emergent ophthalmologic consultation. Ocular symptoms included blurry vision, flashers and floaters, and vision loss. Four different EDs were used to collect data and enroll patients. Patient enrollment began at different dates owing to site institutional review board approval. The study was approved by all institutional review boards at each of the participating hospitals. Both written and oral consent were obtained from each patient prior to enrollment in the study. The University of California, Irvine, UCLA (University of California, Los Angeles), University of Southern California, and Loma Linda University institutional review boards approved the study for their respective sites.

Study Setting
Of the 4 sites, 2 were academic EDs and 2 were county hospital EDs with academic emergency medicine attending physicians present. All 4 sites support an emergency medicine residency, ophthalmology residency, and emergency ultrasonography fellowship. The combined annual ED census of all 4 sites is greater than 300 000 patient visits per year with a culturally and economically diverse patient population. Twenty-four-hour ophthalmologic consultation was available at all 4 sites.
Seventy-five unique practitioners evaluated patients with ocular symptoms in the ED, including emergency medicine attending physicians, resident physicians, and supervised physician assistants.
These practitioners had variable POCUS experience and training. Each site provided annual POCUS training and independent credentialing for all providers. Before enrollment, we gave all practitioners a 30-minute lecture followed by 30 minutes of hands-on scanning of healthy volunteer models. The training introduced the practitioner to ocular POCUS and outlined the key sonographic features that distinguish RD, VH, and VD.

Selection of Participants
Any patient was eligible for enrollment in the study who presented to the ED with ocular symptoms; with a concern for RD, VH, or VD; and undergoing an ED ophthalmologic consultation.
Undergraduate research assistants present throughout the various EDs between 8 AM and midnight monitored the ED tracking board for eligible patients. Practitioners were approached and asked if the patient had concern for RD, VH, or VD. Patients who met the study criteria and were undergoing an ophthalmologic consultation were approached for enrollment in the study by the research team.
We excluded persons younger than 18 years, non-English or non-Spanish speakers, those who declined to be enrolled in the study, and those with ocular trauma or suspicion for globe rupture.

Study Protocol
All enrolled patients underwent a POCUS performed by the treating practitioner. To ensure that the practitioners were not influenced by the ophthalmologic examination results, POCUS was performed before the patient's ophthalmologic consultation. The ophthalmologist who examined the patient was masked to the results of the POCUS. We performed ocular POCUS using the following ultrasound machines: Mindray TE7 (Mindray North America) and Sonosite M-Turbo (FUJIFILM Sonosite). All POCUS machines were equipped with a linear, high-frequency probe at 7.5 MHz with a dedicated ophthalmologic setting. This setting produced a thermal index less than 1.0 and a mechanical index less than 0.23.
Patients were placed in an upright or supine position based on practitioner preference. We then applied ultrasound gel to the upper eyelid and placed the linear ultrasound transducer over the patient's closed eyelid. We obtained both sagittal and transverse views of the affected eye. In the transverse orientation, we aimed the probe marker to the patient's right; in the sagittal orientation, we aimed it cephalad. Using ultrasonography, we inspected the posterior chamber of the globe for the presence of an RD, VH, or VD. Depth and gain were set at the discretion of the treating practitioner. Practitioners performed both static and kinetic examinations to aid in distinguishing among the 3 conditions. During a static examination, the patient held the eye still and the sonographer fanned through the globe. During a kinetic examination, the sonographer held the probe steady and the patient was instructed to look left and right.
The entire orbit was scanned by the practitioner in a fanning motion. B-mode ultrasonography was used to visualize the patient's vitreous body and posterior chamber. An RD was confirmed by the presence of a bright, echogenic membrane tethered to the optic disc but separated from the choroid ( Figure 1A). A posterior VD was defined by the presence of a detached, thin, mobile membrane at the interface between the vitreous and the retina ( Figure 1B). These 2 abnormalities were differentiated based on the visual appearance of the membrane and whether the membrane was tethered to the optic nerve. A VH was defined by the presence of a fluid collection of variable echogenicity in the posterior chamber that rotated with kinetic examination ( Figure 1C). These findings were recorded immediately on a standardized data collection sheet by research personnel at bedside following POCUS. The ultrasonographic diagnoses of the emergency medicine practitioners were compared with the criterion standard of the ophthalmologists' final diagnoses after their evaluation. For several patients, more than 1 diagnosis was recorded.

Statistical Analysis
Data were collected by research assistants using portable electronic devices at bedside and  14.2(StataCorp) and the robust vce option. For a combination of RD, VH, or VD, we expected that ultrasonography would be at least 80% sensitive based on previous data. Thus, we calculated a sample size of 225 patients using an estimated 15% incidence of RD, VH, or VD in our population.
Statistical significance was also calculated in Stata, version 10 using a 2-tailed test. Significance was determined as P < .05.

Results
We approached 252 patients for enrollment in the study and excluded 27 patients from the final data analysis for the following reasons: 13 patients declined to be enrolled, 8 patients had incomplete data collection, 4 patients did not receive ophthalmologic consultation in the ED, and 2 patients requested to be removed from the study following enrollment. Two hundred twenty-five patients were included in the final data analysis. One hundred thirty-five (60.0%) of the patients were men and 90  (Figure 2) The pooled sensitivities, specificities, positive predictive values, and negative predictive values for the 3 disease processes are listed in the Table. Discussion Ocular POCUS is a diagnostic modality that may aid emergency medicine practitioners in identifying vision-threatening ocular disease processes. 13 Point-of-care ultrasonography is ideal for the ED setting because of its portability, lack of radiation exposure, and time efficiency. Utilizing POCUS to evaluate ocular pathology is promising because the eye is superficial and fluid filled. The available literature has shown that emergency medicine practitioners can detect ocular anomalies using ocular POCUS. [7][8][9][10][11]14 Blaivas et al 7 prospectively enrolled 61 participants to assess the accuracy of POCUS for evaluating general ocular disease processes and found a sensitivity of 100% and specificity of 97.2%.
A 2017 study by Baker et al 10 showed that emergency medicine practitioners are able to differentiate between RD and VD with moderate accuracy; in that study, the sensitivity for retinal detachment was 74.6% and the sensitivity for posterior vitreous detachment was 85.7%. Because RD may result in irreversible vision loss, the ability to detect it promptly may be useful in improving transition of care from emergency medicine to ophthalmology, substantiating the need for these patients to receive emergency consultation. Emergency medicine practitioner-performed POCUS was not as sensitive in identifying VD and was only modestly accurate at diagnosing VH. However, the higher specificities for these 2 pathologies indicate that emergency medicine practitioners are better at successfully ruling in these conditions. The lower sensitivities may have been associated with the fact that most emergency medicine practitioners are more focused on finding RD, or are more comfortable identifying RD than VD and VH. These 2 disease processes, unlike RD, are not considered true ophthalmologic emergencies, and these patients may be referred to an ophthalmologist for prompt outpatient follow-up. The ability to accurately differentiate between these ocular disease processes may be useful for determining the urgency with which patients would need to be examined by an ophthalmologist.
We believe that, given the results of our data, POCUS can be used by emergency medicine practitioners to quickly identify RD, VH, and VD in the ED. The addition of POCUS to the history and physical examination provides a useful adjunct method to confer additional information to the ophthalmologist.

Limitations
There are several limitations to this study. Our study was conducted within 4 different EDs. Two of the sites were academic EDs and 2 were county EDs; thus, it is unclear whether the findings will translate to patient populations in different settings. Point-of-care ultrasonography is also operatordependent, and our sonographers had varying levels of ultrasonography experience and proficiency.
The amount of training required for proficiency in ocular POCUS was not addressed in this study.
Interrater reliability was not evaluated in this study but should be considered in future studies. An additional limitation of our study was convenience sampling because our research team was able to enroll patients daily only from 8 AM to midnight despite the availability of 24-hour ophthalmologic services. Although our study primarily evaluated the use of POCUS to diagnose RD, we did not ask our sonographers to specifically distinguish macula-on from macula-off detachments or retinal tears.
Diagnoses other than RD, VH, and VD were not evaluated using POCUS and should be considered in patients with ocular symptoms presenting to the ED. Patients with globe rupture and possible traumatic RD were excluded from the study; therefore, our results may not be generalizable to this population. Finally, physicians performing POCUS were not masked to the patient's history or physical examination results; thus, the independent contribution of POCUS is unknown.

Conclusions
Our findings suggest that emergency medicine practitioners are capable of accurately identifying and differentiating among RD, VH, and VD. Point-of-care ultrasonography is not intended to replace the role of the ophthalmologist for definitive diagnosis of these conditions; it serves as an adjunct method to help emergency medicine practitioners improve care for patients with ocular symptoms.
This diagnosis method may be of particular benefit to EDs where around-the-clock ophthalmologic consultation may not be accessible.