Are health diagnoses in video calls accurate?

How does a doctor's diagnosis over a video call compare to a diagnosis in person?
07 September 2022

Interview with 

Bart Demaerschalk, Mayo Clinic


A doctor coming out of a computer screen


Newspapers at the moment are full of letters to editors from patients disgruntled that they can't get face to face appointments with their GP. This is partly because medical consultations have moved heavily towards the use of video and telephone calls. This is partly down to Covid-19, but also is becoming increasingly common as pressed practitioners, that have seen their list sizes climb by 20% in some cases without any corresponding increase in doctor numbers, find that they can assess more patients more quickly if they use these new digital approaches. Some, particularly younger people, welcome this as more convenient. Others are less enamoured. But the crucial question must be, is it safe? Bart Demaerschalk is a neurologist and a digital care specialist at the Mayo Clinic where they've recently done a study to find out...

Bart - Our principal objective was to determine the diagnostic accuracy of the provisional diagnoses that clinicians make when they assess a patient by video telemedicine in the patient's home, compared to diagnoses established in a traditional in-person clinic evaluation. And we reason that the true scenario to test were patients presenting with a brand new clinical problem.

Chris - I guess the only way to do that then is to do a telemedicine diagnosis and then see the patient for real. And see if you change your mind, having seen them face to face versus seeing them on a Zoom call or similar

Bart - That's exactly what we did. As the World Health Organization declared the COVID 19 pandemic, and we anticipated that there would be a tremendous rise in the utilization of video telemedicine, we launched this study. This was a review of patients at Mayo clinic presenting with a new clinical problem by video telemedicine to their home, and then were subsequently seen in a traditional in person clinic environment. And the principal result was that in 87% of the patients, the provisional diagnosis following the video telemedicine consultation, matched the diagnosis established at the traditional in-person visit.

Chris - It's really interesting that you say that because, when I went to medical school, one of the first things that we were taught was more than 85% of the time, you should be able to make a diagnosis just on the basis of the history, what the patient tells you. So your numbers line up perfectly with someone else's statistics. But what that does mean is that 15% of the time it didn't. So what did you discover about those 15% of diagnoses where you had to change your mind

Bart - Precisely. So for clinical conditions that required a traditional in-person physical examination, and for those conditions where the diagnosis rested upon the result of a diagnostic study, the diagnostic accordance of the video telemedicine visit was less than the average. We also discovered that not surprisingly clinical conditions like psychiatry, psychology had high diagnostic concordance and other conditions like ear nose and throat, ophthalmology, dermatology had a lower diagnostic concordance.

Chris - Are you advocating then that maybe what we do is some kind of hybrid approach where, rather than just see everyone initially on a video call, that there are some aspects of medicine, which are really very well practiced via this sort of route, but other things we really do need to see people. And maybe we could have some kind of dichotomy there where we see face to face complaints that do look like they really do need to be seen in person, but there are some things that can vary safely be handled remotely?

Bart - That's correct. The main message is that given that this study was launched at the height of the COVID 19 pandemic, the main result is quite reassuring to our patients. The vast majority of their clinical concerns and indications they brought forward were sufficiently diagnosed by video telemedicine. But the learning yielded the fact that, for some conditions, the diagnostic concordance was high and for others low, and that this might be one of several ways in which to tailor the video telemedicine examination to those indications and clinical problems and diagnoses where it's, where it lends itself best. Over time, some clinical specialties have improved upon the video examination. For example, there are colleagues at Mayo clinic that have published their work with improving the musculoskeletal examination by video telemedicine, teaching their patients how to help the clinician over video perform a better physical examination in the absence of being able to lay hands. So this is clearly an evolving field.

Chris - Were any of the things where you had to change your diagnosis or management, subsequent to seeing the person face to face, tantamount to a clinical incident in the sense that were you not to have seen that person face to face it would've turned into a life threatening or harmful situation for that patient?

Bart - Yes, we did study morbidity and mortality. All the patients in the cohort we followed for six months after the data collection and 31 died, but in only one of those was the cause of death determined to be possibly related to a diagnostic error. So although the incidence was low, the study indicated that it's conceivable.

Chris - So what's your reading of all this then? And what would you say should be the model then based on your learning?

Bart - Almost certainly there are gonna be a multitude of digital healthcare interventions or interactions between clinicians and their patients. And we will be practicing in a hybrid model. I think in instances of primary care video, telemedicine is more likely to be successful for established patients. And for those patients that are presenting for the first time to a primary care provider with a new clinical condition, a video first approach is quite reasonable. And then escalating those cases where the clinical problem either requires a diagnostic test, a physical examination, or is in one of the categories that we've learned that has a traditionally lower diagnostic concordance to be expeditiously elevated to an in-person examination.


Add a comment