Effectiveness of Back up/Delayed Prescribing

The problem about all trial data is it’s the average effect. So we always want to know, what about my particular patient? So really, you have to use a bit of clinical judgement to target antibiotic use, and I think use delayed antibiotics flexibly. Most people do not need an antibiotic. So the question is, who are the people who are most likely to need an antibiotic? So let’s look at otitis media first. The individual patient data, the IPD meta-analysis in The Lancet, the children who are most likely to benefit are those under two, those with pus, or those with bilateral disease, both ears affected. And here you’re talking about a number needed to treat of around 4 to 5.

You have to remember that the benefit that that analysis looked at was at days 3 to 7, actually when symptoms are much milder. So just remember the slides that I showed you about otitis media. Most of the really bad stuff, the pain, is in the first 24-48 hours. So you’ve got a number needed to treat there of 4 to 5, for by and large milder symptoms. From our trial, if you like, we looked at the children who are most likely to benefit, and it was those with a higher temperature or vomiting. A number needed to treat of 3 to 5. So the picture here is more floridly unwell children and younger children are likely to benefit a little bit more.

Does that mean that all of them have to have an immediate antibiotic? Again, I would use your clinical nous. No, I think it’s perfectly fine for such children to have a delayed antibiotic still. But if they’ve got more florid symptoms and are more unwell, than certain you could even halve the waiting time. So the normal waiting time from 72 hours could be reduced to 36 hours or even 24 hours. So I would clinically use things flexibly according to how unwell people are. Sinusitis, a similar type of thing. Another individual patient data meta-analysis in The Lancet. A number needed to treat of 15 for the average patient we see.

So you have to treat 15 patients, roughly, for 1 to benefit in sinusitis. But if you have pus visible in the pharynx, a number needed to treat of around 8. So I think the message from these two, otitis media and sinusitis, is unwell patients still settle. But if you’re going to use a delayed prescription, I would possibly shorten the waiting time, halve the waiting time from the averages that we discussed earlier. In chest infections, there’s some nice, simple stratification that you can do clinically. But basically, there are six symptoms and signs. As you can see, two on the history, two chest signs, two vital signs.

And if you have none of those six, about 1% will have consolidation on an X-ray. One to two of those, around 5%, 1 in 20 will have. And most people who come to see us are in that category. If you have three or more, 20%. So if I see somebody with three or more of those, I’d be tempted to either have an immediate antibiotic or a delayed antibiotic with a reduced time course. One to two, delayed antibiotics are pretty reasonable. Or ask people to come back– a good safety netting. And if they don’t have any of those, then I certainly wouldn’t be prescribing antibiotics.

So for chest infections, I would use a stratified approach and target the delayed prescription to those who have a middling risk, if you like. So that will be dealt with, as I said, in more detail. But I think we could say probably reasonably that if you use better diagnosis and targeting of delayed prescriptions, it’s likely to lead to better outcomes. You certainly don’t need to use a delayed prescription in everybody.