Robot sitting in an office

‘Dr Google’ wrong most of the time, study finds

Image credit: Dreamstime

Australian researchers have assessed free online symptom checkers, finding that they produce an accurate diagnosis around a third of the time and struggle with uncommon conditions.

Previous studies have estimated that approximately 80 per cent of Australians search for health information online, and 40 per cent look for self-treatment advice online.

Many of these people may turn to symptom checkers: programs which suggest potential diagnoses and triage advice based on symptoms listed by the user.

Some concerns have been raised about online symptom checkers: they are often not affiliated with reputable organisations; can encourage users to seek unnecessary care; and raise questions about data security, privacy, and lack of regulation. Notably, there are also concerns that they provide inaccurate health information.

“We’ve all been guilty of being “cyberchondriacs” and googling at the first sign of a niggle or headache,” said lead author and postgraduate student Michella Hill. “But the reality is these websites and apps should be viewed very cautiously as they do not look at the whole picture – they don’t know your medical history of other symptoms.”

“For people who lack health knowledge, they may think the advice they’re given is accurate or that their condition is not serious when it may be.”

Hill was among a team of researchers at Australia’s Edith Cowan University who decided to put the most popular symptom checkers to the test. They picked the top free symptom checkers suggested by search engines and app stores, creating a sample set of 36 web-based and mobile symptom checkers.

They assessed the symptom checkers using 30 existing patient vignettes and 18 new symptom-based scenarios, with the correct diagnoses confirmed by three doctors with 87 years of combined clinical experience.

The researchers found that the symptom checkers produced the correct diagnosis just 36 per cent of the time, reflecting the findings of similar studies. The correct diagnosis was within the top three results 52 per cent of the time and within the top ten results 58 per cent of the time. The checkers performed less effectively when diagnosing uncommon conditions, with some checkers not including illnesses that exist in Australia such as Hendra virus and Ross River fever.

All of the symptom checkers warned users that their service is not a substitute for consulting with a doctor.

The symptom checkers provided accurate advice on when and where to seek care 49 per cent of the time. The triage advice tends toward risk aversion, which could lead to some users seeking urgent treatment unnecessarily. The advice for seeking medical attention for emergency and urgent care cases was appropriate most of the time (60 per cent) with accuracy falling to 30-40 per cent for non-emergencies.

Hill said that – despite their shortcomings – symptom checkers have a limited role to play in healthcare: “These sites are not a replacement for going to the doctor, but they can be useful in providing more information once you do have an official diagnosis.”

The study has been published in the Medical Journal of Australia.

Sign up to the E&T News e-mail to get great stories like this delivered to your inbox every day.

Recent articles