The tragic case of baby William Mead reported all over the media this week demands a response. As in so many failures, there were several points at which a different decision might have avoided the patient’s death, but the spotlight has been on advice from NHS 111.
NHS 111 will be “fundamentally” overhauled, the Health Secretary has announced, following a string of scandals. (Telegraph). “We need to improve the simplicity of the system – so that when you get 111 you aren’t asked a barrage of questions, some of which seem quite meaningless” says Jeremy Hunt. I might ask, who will be conducting the fundamental overhaul? The establishment figures who have so consistently defended NHS 111, or those like myself who opposed the concept from the outset? I think we all know the answer.
Hunt is an intelligent man and can surely understand the principles of sensitivity (measures the rate of false negatives, as in this baby case) and specificity (measures the rate of false positives, a bigger problem for 111 given that 17% of callers are sent ambulances or told to go to A&E). The problem is, even if the algorithm is perfect, to increase sensitivity means asking more questions, some of which will seem meaningless, all of which will take time, and false positives will increase.
A simple thought experiment shows that it’s logically impossible to achieve 100% sensitivity, as it requires an infinite number of questions. “Cough: Do you have a red flagpole embedded in your chest?” Just checking.
Our competitor webGP/eConsult has mired itself in the red flag mess. Start with acne and you’ll be asked whether you’ve had a fever and been travelling in a malarial area, both common enough. Now you are directed to go to A&E. But you only went in with acne??? And how is that supposed to relieve pressure on the NHS?
Why don’t they ask whether you’ve taken more than 20 paracetamol in the last day? UK deaths in 2013 from paracetamol, 226 (ONS). Deaths in the same year from malaria, 7. They don’t even touch on a problem which has 30 times the death rate. By the way, if you’re reading this Arvind, that’s not a suggestion. There is no end to it. You can’t say “We’ll only ask the obvious ones” or you’ll have Jeremy Hunt at the despatch box again, offering condolences.
Safety lies in the system as a whole, not its components in isolation. Humans can make good clinicians, but they can’t make computer algorithms as good as good clinicians.
Our view with askmyGP is that the professional clinician should have total control of decision making, assisted by what the computer does best in collecting useful data from the patient. Usually they will look at notes and ask more questions. Crucially that makes the system more efficient, so more demand can be responded to faster, and dealt with in the most appropriate way, both clinically and economically.
“Efficiency is a moral imperative” says Don Berwick, and it is essential for safety. The least safe interaction of a patient with general practice is to be turned away by a receptionist, and that happens in the UK over 100,000 times a day. We are building a better, safer system.
Our next webinars are filling up: “Unlearning the appointment system” – tackles three beliefs which trap us in an unsustainable present.
Thanks for reading, and if you have any reactions then please comment below.