A Harvard study compared the accuracy of 23 online symtom checkers and found that, on average, they were correct only 34% of the time. We explain why this might be.
So, you develop an ache in your neck and take yourself off to the GP or physio. They ask you a few questions and readily provide you with a diagnosis. You feel quite chuffed until you see the notice behind them: “My diagnosis is usually correct in 34% of cases.”
Hmmm, starting to lose your confidence in your diagnosis a bit? A recent study from Harvard suggests that this is the “best case scenario” when you use a symptom checker for an online diagnosis.
In this article:
Online symptom checker accuracy
What is a symptom checker?
Why are symptom checkers so inaccurate?
Accurate online diagnosis is possible
Online symptom checker accuracy
The researchers from Harvard checked the performance of 23 symptom checkers found on the internet and the results were alarming. They found that the correct diagnosis was listed first in only 34% of cases. It gets even worse - performance varied by urgency of condition. For conditions that required emergency care, the correct diagnosis was only listed first in 24% of cases!
Some of the symptom checkers did not offer a diagnosis, but provided triage advice only. Triage advice is when someone tells you if you should visit a GP or Emergency Department or whether you can manage your condition/injury yourself at home.
The symptom checkers performed slightly better in this department with the correct triage advice being given in 57% of cases (still not going to let me sleep easy).
There was of course a big variation in accuracy across the 23 symptom checkers. The most accurate at providing the correct online diagnosis was DocResponse. They had a 50% success rate while MEDoctor only managed it in 5% of the cases.
HMS Family Health Guide gave the correct triage advice in 78% of cases (that’s looking better) while ITriage only manage it in 33% (I think they should reconsider their name!).
The following Symptom Checkers were included in the study: AskMD (USA), BetterMedicine (USA), DocResponse (USA), Doctor Diagnose (USA), Drugs.com (USA), EarlyDoc (Netherlands), Esagil (USA), Family Doctor (USA), FreeMD (USA), Harvard Medical School Family Health Guide (USA), Healthline (USA), Healthwise (USA), Healthy Children (USA), Isabel (UK), iTriage (USA), Mayo Clinic (USA), MEDoctor (USA), NHS Symptom Checkers (UK), Steps2Care (USA), Symcat (USA), Symptify (USA), Symptomate (Poland), WebMD (USA).
What is a symptom checker?
Symptom checkers are designed to provide you with an online diagnosis of your injury or ailment and/or advise you about whether or not you have to visit your GP or Emergency Department.
Symptoms are subjective signs of your injury or disease that you experience, e.g. pain, numbness or nausea, to name a few. These online diagnosis tools usually ask you to input the part of the body that you are experiencing trouble with and then ask you a few questions about this. They feed your answers into an algorithm and there you have it…your diagnosis!
I tried a few symptom checkers this morning and was told my pain in my left thigh could be anything from a blood clot to fibromyalgia. The real cause wasn’t even listed: Irritation of my lower back brought on by my terrible posture while working on my laptop (#dontjudgeme).
Why are symptom checkers so inaccurate?
The short answer is that the “symptoms” of an injury or condition is only one piece of the puzzle. The symptom checker tools simply do not ask enough questions to provide an accurate online diagnosis.
Pain in my thigh for instance can be caused by a muscle tear, a bruise, varicose veins or a parasite from the Amazon river, to name but a few.
To narrow a diagnosis down, clinicians will listen to your symptoms and then ask more questions in reaction to these. E.g. have you visited any foreign country in the last 3 months? No? Then we can take exotic parasite off the list.
How these symptoms behave, adds another important clue. Understanding the mechanism of injury, under what circumstances you experience your symptoms, how severe they are, what alleviates them etc. all helps to fine tune your diagnosis, since different injuries/conditions behave differently under the same circumstances.
The medical signs associated with your condition also tells the clinician a lot. A knee that swells up severely immediately after injury usually indicates that your injury will require a scan. On the other hand, a joint that gradually swells up over time has usually sustained a mild strain only.
Accurate online diagnosis is possible
At Sports Injury Physio we first conduct a detailed interview over video call. We then get you to perform the same movements that we would use if we examined you in a regular clinic.
This allows us to form a diagnosis and provide you with a treatment plan tailored to your specific needs. We will also communicate with your GP (with your consent) if we feel that you should have some further investigation done.
About the Author
Maryke Louw is a chartered physiotherapist with more than 15 years' experience and a Masters Degree in Sports Injury Management. Follow her on LinkedIn or ReasearchGate.