These days, many people turn to Internet symptom checkers to self-diagnose. Can these tools, provided by a variety of sources, be trusted with your health? According to a recent study, they're not perfect. In a great many cases, they fail to correctly identify the right illness. But are they still doing an important job?
Web-based diagnosis tools are a growing trend receiving hundreds of millions of hits every year. According to the figures, their reliability is roughly equivalent to the telephone triage lines commonly used at primary care practices, and they offer significantly better answers than general Internet-search self-diagnosis. And yet, they often get things wrong. The first wide-scale study of the accuracy of general-purpose symptom checkers has just been completed by researchers at Harvard Medical School (HMS) and published in the medical journal, BMJ.
Symptom checkers are hosted by medical schools — including Harvard Medical School — as well as hospital systems, insurance companies and government agencies — including the United Kingdom's National Health Service. A patient begins by entering his or her perceived symptoms — what and where the problem is — by using body maps, multiple-choice lists and free text entry. The tool then returns its diagnosis, a list of possible causes as well as advice on how the patient should respond. The symptom checker may advise various solutions, including seeking immediate help, booking an appointment or looking at self-care.
The researchers took 45 clinical "vignettes" — the cases of fictional patients that are used to teach and test medical students — and fed the sets of symptoms into 23 different symptom checkers. For the majority of results, the tools did not identify the correct imagined illness as their first choice. The software algorithms only listed the correct diagnosis first in 34% of cases. The correct diagnosis was included in the top three diagnoses in the list in 51% of cases and in the top 20 in 58%.
However, getting an exact diagnosis may not be as important as it seems. "It's not nearly as important for a patient with fever, headache, stiff neck and confusion to know whether they have meningitis or encephalitis as it is for them to know that they should get to an ER quickly," says Ateev Mehrotra, associate professor at HMS and Beth Israel Deaconess Medical Center and senior author of the study.
Getting the correct advice about whether or how quickly to go to the doctor may be the real strength of online symptom checkers. Twenty-three of the online tools provided correct advice in 58% of cases. But they had a much better success rate in more critical cases, correctly recommending emergency care in 80% of urgent cases. This compares with only 64% accuracy with ordinary Internet search engines.
The researchers found that the diagnosis tools chosen for this study tended to be overly cautious. They often encouraged users to seek some form of medical care for situations where staying at home might be reasonable. This is a tendency that could push up costs for healthcare providers, the researchers say, and one which would benefit from reforms of both software and healthcare systems to reduce costs.
Overall, online symptom checkers got a fairly clean bill of health. But the authors advise a sensible approach to using them. If you are seriously worried about any aspect of your health, see your doctor. "These tools may be useful in patients who are trying to decide whether they should get to a doctor quickly," says Mehrotra, "but in many cases, users should be cautious and not take the information they receive from online symptom checkers as gospel."