Editorial: Is Telemedicine Safe? It’s Unlikely We’ll Ever Know

Orthopaedic surgeons learned more about telemedicine in 2020 than they ever wanted to know. Surgery is a contact sport, and surgeons missed having contact—both human contact in the office and the ability to help patients heal through contact in the operating room.

But when COVID-19 disrupted practices, virtual contact—telemedicine, in its various forms—was all that many surgeons had, a lifeline for desperate practices in some very rough water [11]. More importantly, it allowed our patients at least some access to nonemergency care. The conditions orthopaedic surgeons treat matter to the patients who have them, even when those conditions are not life- or limb-threatening.

Patients and providers generally were satisfied with orthopaedic telemedicine during the recent pandemic [2], though I have to imagine that those satisfaction scores were framed at least implicitly by a somewhat coercive premise: “Have it this way or have nothing at all.” Still, shorter visits and less consumption of brick-and-mortar resources almost guarantee that telemedicine will be less expensive than in-person care [1]. Telemedicine also extends the geographic reach of centrally located subspecialty care to outlying populations for whom distance may not merely be a dealbreaker but a life-ender [3]. Thoughtful observers speculate that virtual medicine is here to stay [8, 12]. I agree with them.

I also think this is generally a good thing. The vast majority of patients who’ve had care delivered telemedically (is that an adverb now?) have benefited from the experience, and although there have been some galling lapses—such as delivering end-of-life news over a blinky monitor [6]—it can be done effectively when done thoughtfully [14].

But can it be done safely? We’ll never know.

In general, ascertaining that something in medicine is safe requires the study of vast numbers of patients, and a careful search for rare (and often difficult-to-capture) endpoints [9]. Telemedicine in this context is a diagnostic tool. The only way to know with confidence that a diagnostic test is safe [7] is to perform a study that applies the gold standard of diagnosis (in this case, an in-person visit) to all patients who received the new test (here, a telemedicine visit), and determine whether important diagnoses were missed.

Imagine for a moment that attempting to make some serious diagnosis over telemedicine results in the clinician missing it, but only very occasionally. Imagine further that seeing the patient in person decreases that likelihood by a small amount, let’s say two in 1000. Increasing the risk of a fatal missed diagnosis from one in 1000 (in person) to three in 1000 (using telemedicine) is a difference we’d surely care about, given the stakes. But capturing this difference would require a study of more than 15,000 patients. And we’d need near-complete follow-up, since patients who are lost to follow-up are more likely to have experienced adverse events than are patients whom we’ve accounted for [4, 10, 13].

It’ll never happen.

This is why I question the conclusions of studies that purport to evaluate telemedicine’s cost effectiveness. If we don’t know whether something is effective, it’s unreasonable to pronounce it cost effective. Although we seldom evaluate them, rare but devastating misses must somehow factor into this calculation. A few months ago, I wrote—tongue-in-cheek, at the time—that “we’ll only learn about the safety-related soft spots of telemedicine when plaintiffs’ attorneys point them out to us” [9]. At the time, I hadn’t seen the white paper that was published by the largest physician-owned medical malpractice insurer in the United States. In it, the authors shared: “Of the telemedicine claims we have seen, the most common allegation has been missed diagnosis, and the most commonly missed diagnosis was cancer” [15]. Think of all the ways you can miss that diagnosis in your practice on a good day; now consider whether you’ve ever had a bad one, and then imagine a computer monitor between you and your patient on just such a day.

Since we can’t prove safety, we need to be circumspect in terms of how we use telemedicine after the pandemic. We’re no longer under the gun with COVID, at least not as severely as before; we now can be choosy in how we deploy this new tool. And to be clear: The reason to be choosy isn’t to avoid lawsuits, nor is it an overreaction to the many unanswered privacy- and cybersecurity-related questions that continue to swirl around the topic of virtual care [5]. The reason we should be methodical about our application of telemedicine is to make sure we don’t needlessly harm other human beings. And although telehealth may seem an appealing solution to expand access to care to populations that struggle to get it, it sometimes falls well short of our hopes in this regard [16].

Think about risk and reward. Telemedicine offers great convenience, and probably very little risk for most patients, in particular those whom we’ve examined in person recently and patients who are unlikely to have a life- or limb-threatening diagnosis. The risk-reward balance may tip slightly for patients with challenging work schedules, and patients who live far away. Post-op incision and ROM checks for patients who are doing well seem especially suitable for virtual visits, though even there we need to be mindful that one in five telemedicine-related malpractice claims may be for missed infections [15]. Visits to discuss good-news imaging studies (including fracture follow-up) and lab tests are a slam dunk, as are taking plan-of-care calls to answer patients’ questions that arise after conclusion of an in-person visit. Those indications alone would afford great convenience to many patients and efficiencies to many practices. No doubt you can think of others. Virtual care, thoughtfully applied, is a real boon.

But all of us were taught that every visit should begin with a history and physical exam, and our teachers weren’t wrong. Where a hands-on exam might turn up something important, a telemedicine visit seems like the wrong tool for the job.