I met my newly admitted patient in the quiet of his private room. He was frail, elderly, and coughing up gobs of green phlegm. His nasal cannula had stepped its way across his cheek during his paroxsysms and was pointed at his right eye. Although the room was uncomfortably warm, he was shivering and asking for more blankets. I could hear his chest rattling across the room.
The young hospitalist dutifully ordered a chest X-Ray (which showed nothing of particular interest) and reported to me that the patient was fine as he was afebrile and his radiology studies were unremarkable. He would stop by and check in on him in the morning.
I shook my head in wonderment. One look at this man and you could tell he was teetering on the verge of sepsis, with a dangerous and rather nasty pneumonia on physical exam, complicated by dehydration. I started antibiotics at once, oxygen via face mask, IV fluids and drew labs to follow his white count and renal function. He perked up nicely as we averted catastrophe overnight. By the time the hospitalist arrived the next day, the patient was looking significantly better. The hospitalist left a note in the EMR about a chest cold and zipped off to see his other new consults.
Similar scenarios have played out in countless cases that I’ve encountered. Take, for example, the man whose MRI was “normal” but who had new onset hemiparesis, ataxia, and sensory loss on physical exam… The team assumed that because the MRI did not show a stroke, the patient must not have had one. He was treated for a series of dubious alternative diagnoses, became delirious on medications, and was reassessed only when a family member put her foot down about his ability to go home without being able to walk. A later MRI showed the stroke.
A woman with gastrointestinal complaints was sent to a psychiatrist for evaluation after a colonoscopy and endoscopy were normal. After further blood tests were unremarkable, she was provided counseling and an anti-depressant. A year later, a rare metastatic cancer was discovered on liver ultrasound.
Physicians have access to an ever-growing array of tests and studies, but they often forget that the results may be less sensitive or specific than their own eyes and ears. And when the two are in conflict (i.e. the patient looks terrible but the test is normal), they often default to trusting the tests.
My plea to physicians is this: Listen to your patients, trust what they are saying, then verify their complaints with your own exam, and use labs and imaging sparingly to confirm or rule out your diagnosis. Understand the limitations of each study, and do not dismiss patient complaints too easily. Keep probing and asking questions. Learn more about their concerns – open your mind to the possibility that they are on to something. Do not blame the patient because your tests aren’t picking up their problem.
And above all else – trust yourself. If a patient doesn’t look well – obey your instincts and do not walk away because the tests are “reassuring.” Cancer, strokes, and infections will get their dirty tendrils all over your patient before that follow up study catches them red handed. And by then, it could be too late.
The short answer, in my opinion, is yes.
The long answer is slightly more nuanced. As it turns out, studies suggest that one’s relative risk of death is increased in teaching hospitals by about 4-12% in July. That likely represents a small, but significant uptick in avoidable errors. It has been very difficult to quantify and document error rates related to inexperience. Intuitively we all know that professionals get better at what they do with time and practice… but how bad are doctors when they start out? Probably not equally so… and just as time is the best teacher, it is also the best weeder. Young doctors with book smarts but no clinical acumen may drop out of clinical medicine after a short course of doctoring. But before they do, they may take care of you or your loved ones.
It has been argued that young trainees “don’t practice in a vacuum” but are monitored by senior physicians, pharmacists, and nurses and therefore errors are unlikely. While I agree that this oversight is necessary and worthwhile, it is ultimately insufficient. Let me provide an illustrative example.
When I was a new intern I was assigned to a patient with curious eyelids. He was a mildly obese, middle aged man with a beard who spoke in hushed tones. What struck me the most was that he had voluminous upper eyelids. They were so strange that I couldn’t stop staring at them. He didn’t have any hives or red blotches on his skin, and his eyeballs were clear and white. There was no pus or discharge of any kind. I was so perplexed that I began to search through his medical record for answers before I embarrassed myself by asking for a consult. After many hours of digging, I discovered the smoking gun.
Apparently, he had been given repeat boluses of 1 Liter of IV normal saline by dutiful interns and residents who had not communicated with one another about who would write the order. So they all did. This man was so fluid overloaded that his eyes were literally bugging out of his head. No one had noticed the edema because of his size, and because (thank God) his heart and kidneys were young and healthy enough to handle the load without going into outright failure. Also, normal saline is such an innocuous medication that it didn’t flag any concerns by the nurses (who were also rotating through the service and busy swatting the more obvious mistakes being made by the fresh crop of interns).
If this poor patient had congestive heart failure or kidney disease, he could have been killed by well-meaning, diligent interns with salt water. Fortunately for him, he made a full recovery – and because there was technically “no harm done” I don’t even think this case was discussed in M&M (morbidity and mortality) conference, and I also doubt that anyone was reprimanded. Sounds crazy, but there are bigger fish to fry in July.
So my point is this: rookie mistakes are not always tracked, documented, addressed, or perhaps even noted. But they are real. They are scary. And they are lurking at every teaching hospital in this country. We must all remain on high alert – and question everything. Because even eyelids offer important clues, and water can kill.
In a recent Forbes editorial, conservative commentator John Goodman argues that the Texas Medical Board is sending the state back to “the middle ages” because they are trying to limit the practice of medicine in the absence of a face-to-face, doctor-patient relationship. He believes that telemedicine should have an unfettered role in healthcare – diagnosis and treatment should be available to anyone who wishes to share their medical record with a physician via phone. This improves access, saves money, and is the way of the future, he argues.
He is right that it costs less to call a stranger and receive a prescription via phone than it does to be examined by a physician in an office setting. But he is wrong that this represents quality healthcare. As I wrote in my last blog post, much is learned during the physical exam that you simply cannot ascertain without an in-person encounter. Moreover, if you’ve never met the patient before, it is even more likely that you do not understand the full context of a patient’s complaint. Access to their medical records can be helpful, but only so much as the records are thorough and easy to navigate. As the saying goes: garbage in, garbage out. And with EMRs these days, auto-populated data and carry-forward errors may form the bulk of the “narrative.”
Telemedicine works beautifully as an extension of a previously established relationship. Expanding a physician’s ability to connect with his/her patients remotely, saves money and improves access. But bypassing the personal knowledge piece assures lower quality care.
I currently see patients in the hospital setting. I run a busy consult service in several hospital systems, and I have access to a large number of medical records, test results, and expert analyses for each patient I meet. Out of curiosity, I’ve been tracking how my treatment plans change before and after I meet the patient. I read as much as possible in the medical record prior to my encounter, and ask myself what I expect to find and what I plan to do. When medical students are with me, we discuss this together – so that our time with the patient is focused on filling in our knowledge gaps.
After years of pre and post meeting analysis, I would say that 25% of my encounters result in a major treatment plan change, and 33% result in small but significant changes. Nearly 100% result in record clarifications or tweaks to my orders. That means that in roughly 1 in 4 cases, the patient’s chief complaint or diagnosis wasn’t what I expected, based on the medical record and consult request that I received from my peers.
If my educated presumptions (in an ideal setting for minimizing error) are wrong 25% of the time, what does this mean for telemedicine? The patient may believe that they need a simple renewal of their dizziness medicine, for example, but in reality they may be having heart problems, internal bleeding, or a dangerous infection. Let’s say for the sake of argument that the patient is correct about their needs up to 75% of the time. Are we comfortable with a >25% error rate in healthcare practiced between strangers?
Goodman’s cynical view of the Texas Medical Board’s blocking of telemedicine businesses for the sake of preserving member income does not tell the whole story. I myself have no dog in this fight, but would side with Texas on this one – because patients’ lives matter. We must find ways to expand physician reach without eroding the personal relationship that makes diagnosis and treatment more customized and accurate. Texas is not returning healthcare “to the middle ages” but bringing it forward to the modern age of personalized medicine. Telemedicine is the right platform for connecting known parties, but if the two are strangers – it’s like using Facebook without access to friends and family. An unsatisfying, and occasionally dangerous, proposition.