You probably saw all that fuss in the news a few days ago about a study in the BMJ which concluded that you shouldn’t take the full course of antibiotics – or at least, that was what was reported. I was immediately suspicious that this was one of those stories that the press spins wildly out of all proportion. So let’s see what the actual study has to say…

First off, this isn’t a “research paper” in the context of “we did some experiments and found out new data”, it’s an “analysis”, in which the authors collected existing data on antibiotic usage and analysed it. This doesn’t mean it’s not useful.

It was reported that the received wisdom that taking antibiotics for too short a time leads to resistance due to survival of remaining bacteria was wrong, and that in actual fact taking them for too long might actually promote resistance. Treating bacteria with antibiotics leads to an evolutionary effect (see figure below): those that have a mutation that makes them resistant to the antibiotic will survive when others are killed, and those resistant bacteria will multiply until all you have is resistant bacteria and a now-useless antibiotic. The reasoning behind “finishing the course” was that if you don’t take it for long enough, you leave behind pockets of probably less-sensitive bacteria that might then multiply and then you have to go back and take more antibiotic to kill the infection, which means (a) you’re not getting better and (b) more chances for the bacteria to evolve resistance. The reasoning behind the “don’t take them after you feel better” idea is that the longer you take an antibiotic, the greater the chance of resistance evolving.

From the CDC website

Does the paper actually say what the papers are reported, and we should all take antibiotics for less time, “until we feel better?” Of course, it’s not that simple. For a start, let’s explode one misunderstanding right away. When they discuss resistant bacteria, for the most part, they are not talking about the bacteria that causes the infection. Rather, they’re referring to the natural population of bacterial fauna that live on and in our bodies, some of which can cause opportunistic infections – such as when our immunity is lowered or when competing micro-organisms have been killed off by, for example, antibiotic treatment. This is called collateral selection.

However, most of the bacterial species now posing the greatest problems do not develop resistance through target selection. The clinical threat comes mainly from species such as Escherichia coli and the so called ESKAPE organisms (Enterococcus faecium, Staphylococcus aureus, Klebsiella pneumoniae, Acinetobacter spp, Pseudomonas spp, Enterobacter spp), which are all found harmlessly in us, on us, or in our environment. They can also act as “opportunistic” pathogens. When a patient takes antibiotics for any reason, antibiotic sensitive species and strains present among commensal flora on their skin or gut or in the environment are replaced by resistant species and strains ready to cause infection in the future. This collateral selection  is the predominant driver of the important forms of antibiotic resistance affecting patients today [emphasis mine]. The longer the antibiotic exposure these opportunist bacteria are subjected to, the greater the pressure to select for antibiotic resistance.

There are a few instances, notably tuberculosis, where resistance of the infections organism is a problem, but it’s these opportunistic pathogens that are the problem.

Right, what about the “complete the course” advice? Well, here again, there’s been some exaggeration by the press. For one thing, the authors note that the length of treatment courses has already been going down. They also make the perfectly valid point that in most cases studies to assess the minimum length of effective treatment time has not been assessed.

For example, pyelonephritis has historically been treated for two weeks. Trials have shown that shorter courses of quinolones are effective (seven days for ciprofloxacin and five days for levofloxacin), but no such data exist for β-lactams, which are the main antibiotic class used. Current international guidelines recommend 10-14 days’ treatment with β-lactams, based purely on absence of data for shorter courses

This is an argument for more studies before you can make an argument for less drugs. They note some infections where reducing the treatment time lengthens the duration of symptoms, such as otitis media (ear infection) in children, but even in that case it just means that the cure is slower. There are also some infections that require very long antibiotic courses (notably tuberculosis) for which there is substantial evidence. I’d be more worried about TB patients not completing their course and spreading drug-resistant TB (which is already a problem) than someone with an ear infection taking it too long, to be honest. I had my first antibiotic course in years last autumn for a nasty case of tonsillitis and the doctor was at great pains to tell me to complete the 10-day course even if I felt better because of the deep pockets of bacteria that tended to remain and re-infect. Was this backed up by evidence? Possibly not, but it may have been backed up by experience with patients coming back to her with repeated tonsillitis.

Their overall review of the evidence suggests that in most cases it’s safer to have shorter treatments, but that not nearly enough studies have been done. They also point out that in the hospital, more sophisticated diagnostics mean that it’s easier to pinpoint when an infection has been cured, but a lack of such testing in community practice means that:

..Patients might be best advised to stop treatment when they feel better, in direct contradiction of WHO advice.

Note the “might” in that sentence; they cite one study on pneumonia as supporting evidence. They even say that you need more research to determine if “stop when you feel better” is an appropriate simple slogan for advice. The fact that the press printed this as if it were the solid conclusion of the study is extremely irresponsible. Perhaps they only read the title. It’s not as if it’s even a new idea: this has been reported on before in almost exactly the same way.


This isn’t to say that antibiotic resistance is a severe problem. We’ve already got effectively untreatable gonorrhoea, and multi-drug resistant TB and MRSA. There are far worse causes than of it than this though: outside of the EU, the use of antibiotics as growth promoters in livestock is a major problem, and all antibiotics should be made prescription-only. In much of the developing world, they can be bought freely at shops, without even consulting a pharmacist, and people even feed them to their children like vitamin pills to stop them getting ill, a fundamental misunderstanding of their purpose. This graphic from the CDC indicates that over-prescribing is just one part of the problem:

From the CDC website

Why were the authors keen to promote this idea? Their take-home message is:

The key argument for changing how we discuss antibiotic courses with patients is that shorter treatment is clearly better for individual patients. Not only does an individual patient’s risk of resistant infection depend on their previous antibiotic exposure but reducing that exposure by shorter treatment is associated with reduced risk of resistant infection and better clinical outcome.

I think what they are thinking of here is really personalised medicine. Hence “individual patients”. But which individuals? Currently, if you go to your doctor with a common infection, they usually won’t even test to see what bacteria is causing it: there’s a list of usual suspects and, for example, they might be able to guess that it’s Staph. aureus by the patches on your throat. You’ll probably be prescribed a broad-spectrum antibiotic that kills everything. This isn’t ideal, but until we have quick, cheap diagnostic tests to more specifically work out what your infection is, then you’re not going to get hospital-level diagnostics. In short, in order to overcome the first problem on that graphic, over-prescribing, you’re going to need to overcome the last one; lack of rapid tests.

The medical profession sounds similarly exasperated by the reporting, which leads me onto my final point: the messages and advice that the public receive.

Commenting on the study, Professor Helen Stokes-Lampard, leader of the Royal College of General Practitioners said:

It is important that patients have clear messages, and the mantra to always take the full course of antibiotics is well-known – changing this will simply confuse people

I understand where she’s coming from, and I also understand that I’m coming from the perspective of a scientist, trained to weigh complex evidence on a case-by-case basis if necessary, but I really dislike the “it will just confuse people” message that’s used to justify a blanket, over-simplified one-size-fits-all rule. For one thing, it sounds patronising, and if people feel patronised they don’t generally listen. For another, what really confuses people is when the messages keep changing – just look at the mixed messages we get over whether fat or sugar is more “bad” for you.

We’ve already seen another example of this: the UK advice for the “safe” amount of alcohol that could be consumed during pregnancy was recently revised from, effectively, “very little”, to “none.” This wasn’t even done in response to any new evidence – very little is perfectly fine – but because it was thought to be, you guessed it, “too confusing.” Now that is patronising. Worse, thanks to advice on the same websites that alcohol can cause fetal malformations, women who have drunk some alcohol before they knew they were pregnant, panicked and got abortions because they thought they’d irreparably damaged their fetus. (This really only happens when you drink excessively and regularly).

In fairness, the Professor also points out that you shouldn’t change widespread behaviour on the basis of one study, and that the length of most antibiotic courses are anyway tailored to the specific infection. So why can’t their message just be:

“How long you need to take antibiotics for depends on what your infection is and how bad it is: follow your doctor’s instructions” [ideally also printed on the medicine label]

That’s more universally true, and it’s less confusing. The authors disagree with me; they take the view that if the medical profession just explained that the evidence had changed, then people would accept this, and point to how the public have been successfully educated that antibiotics aren’t suitable to treat viral infections are like colds, and that anyway, the natural instinct is to stop taking medicine when you feel well. I take their point, but they’ve already said that more studies need to be done, and anyway, it just risks substituting one overly simplified message for another. There are cases when you do have to finish the course long after you feel better. Also, “feel better” is very subjective – completely better, or just not completely awful anymore? How is a small child supposed to be able to describe this? How are their parents supposed to know? This just isn’t helpful. I think the best advice would be: “Use as directed by a qualified medical practitioner.” That way, they won’t have to change the advice when more individually tailored treatments become available!



Llewelyn, MJ. et al, The antibiotic course has had its day. BMJ 2017; 358 doi:


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s