healthHealth and Medicine

Searching Symptoms Online? “Dr Google” Is Almost Always Wrong, Study Finds


Stephen Luntz

Stephen has a science degree with a major in physics, an arts degree with majors in English Literature and History and Philosophy of Science and a Graduate Diploma in Science Communication.

Freelance Writer

Dr Google

70,000 people search for health information online every minute, but the symptom checkers many rely on have the wrong answer as their top suggestion most of the time. Aliaksander Karankevich/Shutterstock

With people increasingly turning to the Internet for information and advice, it is no surprise symptom checkers, often casually referred to as “Dr Google”, have sprung up to diagnose your ills. Unfortunately, a test of their reliability found they don't get the correct answer in their top three suggestions almost half the time.

"We've all been guilty of being 'cyberchondriacs' and googling at the first sign of a niggle or headache," said Michella Hill of Australia's Edith Cowan University in a statement. Hill found 36 checkers, and tested each with 48 symptom sets like those used to train medical students. Some of the checkers were designed for diagnosis, others provide triage, telling the inquirer whether they need to go to hospital urgently, should see a doctor when they could, or didn't need medical attention at all.


So ubiquitous have the triage checkers become, Hill told IFLScience, the UK's National Health Service (NHS) is reportedly encouraging people who call with concerns to use them in the hope of easing strain on the system.

Most of the diagnostic checkers suggest either 3 or 10 possible conditions, although Hill told IFLScience in one case 600 possible illnesses were listed, which she called about as useful as handing out medical textbooks.

In the Medical Journal of Australia Hill reports the first result returned to her test symptoms was correct just 36 percent of the time; in 48 percent of cases the right answer was not even in the top three. Expanding to 10 answers only added the correct one in 6 percent of cases.

In some cases this is understandable. One of the symptom sets Hill used was for Ross River fever, a mosquito-borne disease rare outside Australia and a few other countries. Unsurprisingly, only the Australian checker got it right.


Others were far more alarming; one checker couldn't recognize the classic symptoms of a heart attack, and didn't recommend hospitalization – a potentially fatal error. More often, however, the error was in the other direction, over-rating the urgency of seeking medical help.

Hill told IFLScience she saw no signs the checkers are undermined by profit motives or conspiracy theories. Some of the weaknesses are probably inevitable – unlike a family doctor, the checker can't take blood pressure or examine swellings. They also don't have access to potentially important background information such as family histories.

On the other hand, many of the checkers fail to ask important follow-up questions. Few respond to reports of pain by asking about its severity, and most don't provide the opportunity to list medications the inquirer is taking.

“These websites and apps should be viewed very cautiously as they do not look at the whole picture,” Hill said


Some checkers, particularly those run by health departments, are better than others, Hill told IFLScience. Hill does think Dr Google serves some uses, though. These include providing follow-up information for people who don't absorb everything their doctor says when diagnosed in person, and alerting health authorities to outbreaks when many people in one area start searching symptoms of the same infectious disease.


healthHealth and Medicine