App MDs Are Likely To Give The Wrong Diagnoses
LYNN NEARY, HOST:
When you're feeling under the weather, one of the first stops is Doctor Google, or you can find a symptom checker online and plug in your ailments. There are dozens of programs to choose from - the Mayo Clinic and WebMD each have one, and they're designed to provide you with a diagnosis. It turns out these symptom checkers don't do such a good job of it. That's according to a new study by researchers at Harvard Medical School. Scott Hensley hosts NPR's Shots blog, and he's with us to explain. He's in the studio with me now. Thanks, Scott.
SCOTT HENSLEY, BYLINE: Hi, Lynn.
NEARY: And you have a perfect guinea pig right here because I've been sick all week. You may hear that in my voice a little bit. So we can put in my symptoms and see what deadly disease I have.
HENSLEY: Let's try it.
NEARY: OK, let's do it.
HENSLEY: All right, so we're at the Mayo Clinic Symptom Checker. Let's find out what your symptoms might be. There's a long list here that we need to check.
NEARY: All right, cough, difficulty or painful swallowing, yeah.
HENSLEY: Hoarse or muffled voice.
NEARY: (Laughing) Yes.
HENSLEY: Wow, you've get - you're checking off a lot of boxes here.
NEARY: Sneezing, watery eyes.
HENSLEY: OK, let's click Find Causes.
NEARY: OK, let's read what's on this list.
HENSLEY: Hay fever.
HENSLEY: Acute sinusitis.
NEARY: GERD. Epiglottis.
HENSLEY: Epiglottitis, I think.
NEARY: Epiglottitis, OK. Laryngitis, which I might be heading towards.
HENSLEY: Strep throat.
NEARY: But top of the list is common cold, which is probably what I have.
NEARY: OK, all right. Let's talk about the studies. So they - the Harvard researchers tested 23 online symptom checkers. What did they find out?
HENSLEY: Yeah, so they did a pretty interesting test. They fed in 45 different conditions - some rare, some common - with standardized kinds of symptoms and saw what the symptom checkers spit out. And when it came to the top suggestion for what was wrong with people, they were right about a third of the time. If you scrolled down the list, the way that we did, and went as far as 20 on the list, then they were right about half the time.
NEARY: So is there anything that these kinds of apps - these symptom checkers - is there anything that they do well?
HENSLEY: Quite a few of them went the next step and gave you advice about should you stay at home and take care of this yourself, go see your doctor or nurse practitioner for a regular appointment or get yourself to the emergency room right away - they did better at that. So in the case of the emergency care, in particular, they were right 80 percent of the time. The thing that the researchers point out, which is an important caveat, is that the checkers tended to be a little risk averse. So they tended to send people for care more often than they needed to. So in the case of things that the researchers thought you should be able to take care of yourself, they only said that that was right about a third of the time.
NEARY: Yeah, and I would think if you're in a situation where you're ready to go to the emergency room, you might be really getting yourself to the emergency room without going to the computer. You'd know, in other words.
HENSLEY: You would, but I think, for instance, in the case of a heart attack, sometimes the symptoms are pain in the arm and things like that, where the person might not connect the dots. But I think you're right. If it's something that's really a full-on emergency, it might be that you're not going to take time to boot up the computer or turn on your phone.
NEARY: So when all is said and done, how useful are these tools?
HENSLEY: I think they're not too bad as a first step, but I think new models are going to be a lot better. So as the technology gets better, I think they will improve.
NEARY: Scott Hensley hosts NPR's Shots blog. Thanks so much, Scott.
HENSLEY: You bet.
NPR transcripts are created on a rush deadline by Verb8tm, Inc., an NPR contractor, and produced using a proprietary transcription process developed with NPR. This text may not be in its final form and may be updated or revised in the future. Accuracy and availability may vary. The authoritative record of NPR’s programming is the audio record.