People use symptom checkers, computer programs offered by medical schools, insurance companies, hospitals and government agencies, to try to figure out what their symptoms mean and how sick they are. You either pick your symptoms from a list or enter them freehand into the program.

Some checkers give a diagnosis of the most likely illness or illnesses, while others only offer triage advice such as whether a person should seek immediate care, seek care in the next few days or can just stay home. Some offer both a diagnosis and advice for what to do next.

So how helpful are they? While symptom checkers may not be a substitute for a doctor's visit, they can be helpful when you need to know whether or not to seek emergency care, according to a recent study of how well symptom checkers perform.

It's much more important for someone with fever, headache, stiff neck and confusion to know that they should get to the emergency room quickly than to know whether they have meningitis or encephalitis.

On average, the checkers performed about as well as telephone triage lines and better than a general Internet self-search would.

Researchers tested 23 free English language symptom checkers by inputting 45 different clinical scenarios, which they had simplified down to a core list of symptoms. In all, 15 of the scenarios required emergency care; 15 required some medical attention but not emergency care; and for the other 15, a doctor's appointment was generally deemed unnecessary.

Overall, the symptom checkers listed the correct diagnosis first only 34% of the time. The correct diagnosis appeared just over half the time in their three most likely diagnoses (51%).

The accuracy of the diagnoses is perhaps less important than the checkers' usefulness in directing people to the hospital. And here the symptom checkers did much better, correctly advising that patients call an ambulance, go to the emergency department or see a general practitioner immediately in 80% of all cases where emergency care was required.

As the researchers point out, it's much more important for someone with fever, headache, stiff neck and confusion to know that they should get to the emergency room quickly than to know whether they have meningitis or encephalitis.

Previous studies suggest that personal Internet searches of urgent symptoms will recommend emergency treatment only 64% of the time.

Overall, symptom checkers provided appropriate triage advice in 57% of the cases, a figure similar to the telephone triage services that are sometimes provided by primary care doctors to inform patients whether or not urgent care is necessary.

When symptom checkers did give incorrect advice, it was usually to tell a person to seek care when staying at home was reasonable. Depending on viewpoint, this could be described as erring on the side of caution or as promoting unnecessary care.

The bottom line seems to be that how well symptom checkers perform depends on who or what you are comparing them to. The study authors feel that the most appropriate comparison is to telephone triage lines, to which they compare favorably.

The study is published in BMJ and is freely available.