Computers are better than doctors at diagnosing serious bacterial infection – such as meningitis – in young children, according to a study published today.
Children, especially those under five years old, commonly have fever or febrile illness, but it can be difficult to ascertain the cause.
This can be vital because doctors have to be able to tell whether a child is suffering from a minor viral illness, or a serious bacterial infection, such as pneumonia, meningitis or a urinary tract infection.
Getting it wrong and failing to diagnose a serious infection and delaying treatment can be fatal – but diagnosing one where it isn’t there can result in children being prescribed antibiotics unnecessarily.
Researchers in Australia developed a computerised model to distinguish between serious bacterial infections and self-limiting non-bacterial illnesses.
The study included more than 15,000 healthy children aged under five who attended the emergency department of a large children’s hospital with a fever (defined as a body temperature of 38C or more in the previous 24 hours).
Doctors performed the usual clinical evaluation and serious bacterial infections were confirmed or otherwise by using standard tests. The signs and symptoms seen by the physicians were combined in a diagnostic model, which was compared to what actually happened in clinical practice.
The data show that of the seven per cent of children who were later confirmed as having a bacterial infection, only 70-80 per cent were prescribed antibiotics at the initial consultation. Conversely, around a fifth of those who were not subsequently confirmed as having an identified bacterial infection were probably given antibiotics when they shouldn’t have been.
In each infection – urinary tract, pneumonia and bacteraemia (bacteria in the blood) the diagnostic model out-performed or was as good as evaluation by a doctor.
Writing in the BMJ, the authors, including Jonathan Craig of the University of Sydney, point out that almost all (95%) of these children had the appropriate tests and that some doctors routinely delay giving antibiotics until test results are known, so this may help to explain the initial under-treatment. However, about two thirds of children who were not treated were subsequently prescribed antibiotics.
They conclude: “By combining routinely collected clinical information into a statistical model, we have demonstrated that a clinical diagnostic model may improve the care of children presenting with fever who have suspected serious bacterial illness.”
But Dr Alan McDevitt, of the BMA’s Scottish GP Committee, said that while computer software could be a useful tool, it was no substitute for a trained and experienced professional.
“Doctors are looking at which of the subtle signs and symptoms are important, based on years of practice and tradition.
“I think that parents would still rather trust their child’s life to professionals using everything at their disposal – and that might include computer programs.”
He is perfectly happy to use computer programmes in his practice, he says, in particular, one which calculates a patient’s cardiovascular risk. “I input the information and the programme calculates the risk factors, then we’ll make the decision, based on the results, about whether to prescribe cholesterol-lowering drugs.”
Increasingly, he added, it’s the patient who makes the final decision over whether to embark on treatment, based on their own feelings about risk and their condition. For example, having had all the potential benefits or risks of chemotherapy explained to them, a patient might decide whether to go ahead with it or not. Although a programme might be able to analyse a patient’s attitude to risk, Dr McDevitt thinks it’s probably better done in consultation between the patient and doctor.
Computer programmes do have their place, he conceded. They might, for example, point the way to a sign or symptom of a disease which was not previously thought to be important in diagnosis.
Where computer diagnostic models had been tested out, it tended to improve the diagnostic performance of doctors, he added, citing a study some years ago at the former Bangour Hospital in West Lothian, which showed that computers were better than doctors at diagnosing appendicitis. Shortly after the study took place, the doctors’ performance equalled that of the computer, he added.
Nevertheless, he cautions against putting too much reliance on software. “I wouldn’t like to think of a doctor spending so much time on a computer that they delay the treatment that is needed,” he said.