You often hear that an incomplete course of antibiotics can promote the emergence of drug-resistant bacteria. Patients are told not to stop taking antibiotics just because you feel better but instead to always complete the full dosage. This is obviously partly for the health of the patient. But it is also said to guard against drug-resistance.
I have never been able to understand the logic. So I sat down to try to figure it out. I came up with one possible answer but it seems second-order to me so I am not sure if its the primary reason.
One thing that seems obvious is that every time you ingest antibiotics you create natural selection pressure within your body that favors any drug-resistant mutants that are there. This is why the headline cause of super-bugs is overuse of antibiotics. For example antibiotics are often prescribed and taken for non-bacterial illnesses like the common cold. Also, antibiotics are given to livestock that people eventually eat.
But this doesn’t get us to the recommendation to complete your course of antibiotics. Since the problem is the antibiotics, it tells us to use them as little as possible. How could it be that the first dose helps the super-bugs, but the last dose hurts them? They are drug-resistant after all, right?
I used to think that it must be that the selection pressure increases the rate of mutation. This may be true but I don’t think it’s enough to support the complete-course policy. Suppose I am halfway done with the recommended dosage and I stop. My usage so far has encouraged mutations and I may be infected now with a drug-resistant strain. Now the mutation-inducing effect is irrelevant because the superbugs are already there. Their natural growth rate is going to dwarf any additional growth due to mutations.
More generally, since every prescription of antibiotics is a public health cost, then every time I continue to take my doses I am creating the same externality.
I finally hit on one idea which has to do with competition within the body among bugs of differing levels of resistance. Start with this observation. If I have a super-bug in me and I continue to take anti-biotics, I kill off all the wimpy-bugs leaving the super-bugs to have free reign over my body. Presumably they grow faster without the competition. OK. But that seems again to suggest that, at least from a public health perspective, I should stop taking the drug so that the wimpy-bugs can outcompete the super-bugs.
So we add one twist to the model. Suppose that my body has a baseline system of defenses that can fight off any bad guy, super- or otherwise, as long as there are sufficiently few of them. Then, if I stop taking the drug too early, my defenses may still overwhelmed by the sheer numbers. All the bugs start growing again and if I started with just a few super-bugs in me, then when I start to show symptoms again I now have lots of them. However, had I continued to completion, all of the wimpy-bugs would be gone and my body’s natural defenses could have mopped up the few super-bugs that were left hanging around.
It’s a coherent theory. But I am not sure I believe that it is quantitatively important. So I still find the advice a little mysterious. Any of you know better?
Cheap Talk 
11 comments
Comments feed for this article
February 9, 2010 at 7:32 am
Leigh Caldwell
I think it’s your last model, with the addition of another tweak: drug-resistance is not perfect.
Antibiotic doses are chosen in order to destroy bacteria with overwhelming force – if you take the full course, virtually all bugs will be eliminated, even the majority of “drug-resistant” mutations. Your body has enough firepower to sort out the last few resistant specimens on its own.
But if you stop while the regular bugs are still there, your body is less likely to eliminate the drug-resistant ones before they escape into the wider world.
A couple of other thoughts – the term ‘superbug’ is a bit misleading as these bugs have no additional potency or generalised resilience compared to ordinary bacteria; the only difference is their specific interaction with one particular drug.
The question about whether this is quantitatively significant highlights one key point: the significant quantity is ONE organism. Only a single bacterium has to escape to have the potential to establish itself in the outside world. Of course in any given case, that probably won’t happen, but enough people not finishing their drugs and it will occur eventually.
A simpler, more abstract description: Until a population is _fully_ eliminated, it will always tend to grow to fill its ecological niche. Therefore any selective, elimination of one type of organism tends to, in the long run, increase the absolute numbers of any other. The short-run rise in relative numbers becomes a long-run rise in absolute total. Only total eradication can prevent this from happening.
The same population composition effect occurs on a larger scale, so that the world’s total population of bacteria becomes resistant to more and more drugs over time.
A final analogy: go around your (rat and cockroach infested) house with a hammer, killing all the rats and any cockroaches that you can manage to reach. If you stop before getting every single cockroach, you’ll soon find that there are more of them than before. Voila – your hammer is much less effective at eliminating vermin than it used to be.
February 9, 2010 at 7:37 am
Leigh Caldwell
p.s. Needless to say I don’t know any more about this than you do! Here’s an article which I’m sure will educate both of us:
http://en.wikipedia.org/wiki/Antibiotic_resistance
February 9, 2010 at 7:56 am
Travis
Replace “drug resistance” with “incomplete drug resistance” and you’ve got it.
It’s not a binary relationship. A big enough dose (continuing until your final dose) and they’re all dead. Giving them just enough so the ones with a little resistance live, and you have selection.
February 9, 2010 at 10:26 am
Noah Yetter
I would guess that bacteria persisting in the face of an antibiotics-fueled immune system resemble radioactive decay, i.e. it can be modeled as a half-life. If that’s true then the reason for finishing the course of treatment should be obvious.
February 9, 2010 at 11:14 am
samson
Timely! I just had this discussion with a microbiologist friend, for I too was puzzled, unable to escape the conclusions you’ve documented. I think a Leigh Caldwell has the model my friend mentioned. The two ingredients are imperfect resistance and baseline immune response. Importantly, the immune system has some basic ability to deal with an infection, but needs external aid in the form of antibiotics when facing a large invading force. Once the antibiotics take care of the vast majority of the bugs, which requires them being hit repeatedly over X days, the immune system deals with the remainder. Importantly, an antibiotic has a specific mechanism for dealing with the bugs, usually targeting reproduction of the the bug. The immune system has multiple ways of dealing with bugs, and so an antibiotic resistant strain is not immune-system resistant too.
I think this model also explains the occasional prescription of multiple classes of antibiotics for particularly resistant strains. Each dose of a particular antibiotic will kill probabilistically, and the goal is to reduce the numbers below some threshold so that the immune system has both the time and the ability to respond.
February 9, 2010 at 11:42 am
Robert Wiblin
If you pour very dilute bleach on a bacteria covered bench, bacteria that have a little bit of bleach-resistance are left. If you pour concentrated bleach on the same bench, no partially bleach resistant bacteria are left because no bacteria are left. No opportunity for resistance to accumulate.
February 9, 2010 at 11:53 am
Bobby Wallace
I think we’re getting at part of the reason, but could it be simpler? If you don’t finish the full course, you increase the chance that the infection will flare-up again and in turn increase the opportunities for mutation and the likelihood you will require even more antibiotics.
February 10, 2010 at 12:13 pm
Karl Katzke
I think you’re looking at the problem too narrowly, Jeff. There’s a few more things you’re missing in your simple analysis.
1) The generations we’re dealing with are short in the term of days, and the population can multiply exponentially.
2) Classes exist within the bacterial world. One antibiotic will kill one type of bacteria, another will kill another.
3) Beneficial competitive bacteria exist that may be killed — friendly fire, so to speak, paving the way for secondary infections or further infections.
4) Certain drugs will cause problems within the host’s body. Stronger classes of antibiotics have serious neurological side effects. (Just as an example, NEVER ingest neosporin.) Many antibiotics will kill the fauna in the host’s stomach.
5) We’re seeking a balance point, not an absolute. Manual feedback upsets balance points. The key is to do it for the correct amount of time. It’s like a see-saw; Tip it too far, and you get worse side effects. Don’t tip it enough, and your butt hits the ground unexpectedly.
There are far too many variables for your simple quantification above to make sense … but maybe I’m missing why it doesn’t make sense to you.
February 19, 2010 at 12:17 am
Daniel R Hawes
I’ll side with Travis. Think of your initial state of having zero truly “resistant” bacteria. You don’t complete your dose, so you’re left with a few bacteria; namely those with the highest level of resistance (but they too would have died, had you continued your meds).
Now, when these left over bugs start multiplying again, this is when you’ve created a window for a high-resistant (and if mutation occurs possibly entirely resistant) population to arise. There is a chance, that this new population indeed becomes entirely immune to you original meds…
February 22, 2010 at 12:54 am
Dan
I’m not sure if anyone will be reading the comments on such an old post, but I am a biochemist at a pharmaceutical company so I thought I would weigh in.
Lets say a patient has taken 5 days of a 10 day prescription and is deciding whether to take the final 5 days of drugs.
If I understand correctly you are confused because either 1) the patient has the super bug at this point, and the damage is done or 2) the patient does not have the super bug and stopping treatment is no more dangerous from a public health perspective than the original infection.
The “added twist” and many of the comments above explain much of why stopping early is a bad idea. I would add that experiments have shown when resistance develops it goes something like this: in the first 5 days the antibiotic kills most of the bacteria, leaving a few that are resistant, but also sickly. If the patient stops treatment these few multiply eventually giving rise to the superbug who is resistant and not sickly.
A little more detail for anyone who is interested: The first bacteria to become drug resistant pay a large price for that resistance because the drugs target fundamental pathways and are not easily avoided. These bacteria are not particularly good at surviving and reproducing anywhere. Given enough time under “favorable” conditions a second mutation will occur that maintains the resistance, but makes it much less costly. “Favorable” conditions here are 1)drug free and 2) not too many other bacteria around. Ie exactly the conditions in a person who stops taking their meds after 5 days. Drug free is part of the favorable conditions because resistance is not perfect.
After 5 days you do not contain a superbug, but you contain bacteria that have made one of the two evolutionary steps to superbug-dom. So finish your prescription and do not let this beast finish its evolution.
February 22, 2010 at 11:01 am
jeff
thank you very much!