Diagnostic Accuracy of Systematic Screening With Smartphone Fundoscopy and Portable Nonmydriatic Fundus Photography in Neurology Inpatients

Many patients with neurological disorders have changes in the ocular fundus, the identification of which may lead to vision-saving or even life-saving diagnoses and treatment.¹ Unfortunately, although ophthalmologists are trained to perform ophthalmoscopy by multiple techniques (eg, direct ophthalmoscopy using a hand-held ophthalmoscope, indirect ophthalmoscopy, and use of a slit-lamp biomicroscope with a hand-held 90-diopter lens) and to recognize and understand the significance of what they see with one or more of these techniques, most other physicians are not. It has been shown in multiple studies that obtaining digital photographs by both physicians and nonphysicians using a nonmydriatic camera in an emergency department setting and then having those images assessed for pathological findings by an experienced physician (usually an ophthalmologist) substantially increases the likelihood of identifying important abnormalities of pathological significance.^2-5 For this reason, many emergency departments around the world have such a system in place.

The authors of this article have brought this concept to the inpatient setting. They attempted not only to assess the prevalence of both associated and unassociated fundus pathology in a cohort of patients admitted to the neurology service at a major hospital but also to compare the diagnostic accuracy of direct ophthalmoscopy as performed by individuals with various levels of training and experience with this technique with the results and interpretation of digital images obtained with either a portable nonmydriatic fundus camera or a smartphone. Not surprisingly, they found that: 1) there was a clinically significant prevalence of fundus pathology among the patients they assessed; 2) a significant percentage of the fundus lesions identified using photography were missed by nonophthalmologists performing direct ophthalmoscopy; and 3) screening of digital images obtained with a nonmydriatic fundus camera (and to a lesser extent images obtained with a smartphone) was more sensitive than direct ophthalmoscopy in identifying a variety of both related (eg, optic disc pallor in 2 patients with MS) and unrelated (eg, diabetic retinopathy) fundus lesions.

Diagnostic tests are of two types. One type provides an objective and absolute result; for example, the concentration of sodium in the serum is “x”; the hematocrit is “y.” The second type is one for which the results require both identification and interpretation, and this, in turn, means that the sensitivity and specificity of the test are dependent on the individual performing the assessment. Unfortunately, most neurologists either have not been trained to perform ophthalmoscopy using a hand-held direct ophthalmoscope (ie, “direct ophthalmoscopy”) or have had so little training in the technique that they do not bother to perform direct ophthalmoscopy, perform it poorly, or perform it but do not have experience identifying the myriad normal variants, vision-threatening abnormalities, or even abnormalities that indicate a life-threatening neurological disease that may not be known at the time they perform the examination.

To put the above-mentioned comment in perspective with respect to the identification of fundus lesions using photographic images obtained using either a nonmydriatic fundus camera or a smartphone, there are two issues. The first—to obtain objective images of the fundus—is solved by the technique. The authors have shown, as have others, that reasonable images can be obtained using a nonmydriatic fundus camera or even a smartphone with appropriate attachments in almost any setting, including a neurology inpatient unit. The second issue—the interpretation of the imaging findings—is not so easily solved. Once the images have been obtained, there must be someone who can determine if an abnormality is present and, if so, the significance of that abnormality. It seems clear from the results of this study that, although photography of the ocular fundi of patients with neurological disorders requiring hospitalization is more sensitive than direct ophthalmoscopy in identifying both related and unrelated lesions, it will do little good for general neurologists to obtain such images unless they have the training to assess them properly or at least have an ophthalmologist available who can do so in a timely manner. In the end, there needs to be a system, not just a camera!

References

1. Thulasi P, Fraser CL, Biousse V, et al. Nonmydriatic Ocular Fundus Photography Among Headache Patients in an Emergency Department. Neurology. 2013;80(5):432-437. https://n.neurology.org/content/80/5/432.long
2. Lamirel C, Bruce BB, Wright DW, et al. Quality of Nonmydriatic Digital Fundus Photography Obtained by Nurse Practitioners in the Emergency Department: The FOTO-ED Study. Ophthalmology. 2012;119(3):617-624. https://www.aaojournal.org/article/S0161-6420(11)00848-7/fulltext
3. Bruce BB, Thulasi P, Fraser CL, et al. Diagnostic Accuracy and Use of Nonmydriatic Ocular Fundus Photography by Emergency Physicians: Phase II of the FOTO-ED Study. Ann Emerg Med. 2013;62(1):28-33. https://linkinghub.elsevier.com/retrieve/pii/S0196-0644(13)00011-5
4. Teismann N, Neilson J, Keenan J. Quality and Feasibility of Automated Digital Retinal Imaging in the Emergency Department. J Emerg Med. 2019;58(1):18-24. https://linkinghub.elsevier.com/retrieve/pii/S0736-4679(19)30740-1
5. Dunn HP, Teo KZ, Smyth JW, et al. Using Non-Mydriatic Fundus Photography to Detect Fundus Pathology in Australian Metropolitan Emergency Departments: A Prospective Prevalence and Diagnostic Accuracy Study. Emerg Med Australas. 2020;33(2):302-309. https://onlinelibrary.wiley.com/doi/10.1111/1742-6723.13619