Edit: I originally made mistakes in the calculation below, have edited to correct this. See comment below by willbradshaw for details of the calculation.
Thanks! I completely agree there are other strong reasons to reduce (or eliminate) factory farming.
About your other comment – I also don’t think the situation is reassuring at all. I think it’s very plausible that the antibiotic use in agriculture could be an important driver of antibiotic resistance.
I think that we need more research on both the jumping of species barriers and on horizontal gene transfer. This paper could be interesting if you want to read more on how common horizontal gene transfer is, but I haven’t been able to find anything that gives a good assessment on how important this is for resistance (I will be very grateful for suggestions if you or someone else know good research on this!).
I know an analogous problem is that human patients often develop resistant bacteria in the normal gut microbiota when they take antibiotics, and this could also be transferred to pathogens through horizontal gene transfer – again, we don’t know how much it happens. Another thing is that some bacteria can both be living in the environment or on animals or in the normal bacterial flora, and then act as pathogens when they end op in the “wrong” place – for example bacteria that is harmless in the gut flora could cause urinary tract infection. If these develop resistance, they don’t need to do any gene transfer but simply change location to cause problems.
Personally, if I try to speculate, I would reason that it’s very unlikely that antibiotic use on animals drives resistance in human infections as much per kg used as antibiotic use on humans. So if we assume that 75% of the kg of antibiotics is used on animals, I would say that it’s unlikely to drive as much as 75% of the resistance burden on humans. It could be that it is 10% as efficient in driving resistance on humans because of the “transfer barriers”, that would mean that 7.5% (CORRECTION: should be 23%, see comments below) of the resistance burden on humans would go away if we eliminated antibiotic use in agriculture. That would still be very significant and worth pursuing. Of course, I don’t know at all if the number should be 10% or 50% or 1%, and I could also not answer if that makes it a more significant driver than for example the misuse of antibiotics for viral infections.
Apart from speculative, this reasoning is also very simplified, since there are many different classes of antibiotics that target different bacteria. It is unclear which would be the best way to measure “resistance burden”. If we would mean it as QALYs lost, it will make a big difference which bacteria is resistant and to which antibiotic(s). In some cases resistance would just mean switching over the patient to another drug and the patient recovers a couple of days later. In some cases, such as with tuberculosis, resistance could mean the alternative treatment is prolonged by years, has serious side-effects and is also much more expensive. In some cases, the patient would die. Resistance to so-called last-line antibiotics is much more severe than others. This report lists the WHO priority pathogens, I think it’s quite good in how it gives an understanding of how much worse the treatment for resistant tuberculosis is compared to “regular” tuberculosis, for example (p 21).
Even if I think we need more research to understand this better (and thereby how to design interventions and allocate resources), I don’t think that should be taken as a reason to wait with reforming agriculture to eliminate (non-therapeutic) antibiotic use. We know the mechanisms are there, and we also know that it’s possible to remove the antibiotics from production, and I don’t think we can afford to wait and see. This urgency to achieve change is also why I think it unwise to tie this cause too closely to elimination of factory farming, for example – even if that would be good, I think it would take too long time, and I think it’s possible to eliminate the antibiotic use much faster if we keep that as a (relatively) separate issue. In context where one can be nuanced I don’t think it’s wrong to bring up antibiotic use and resistance as a negative effect of factory farming, but I think as you say that there are stronger arguments to front to end factory farming.
This feels a bit petty, since I don’t really disagree with any of your conclusions, but there are some mistakes in the mathematics here.
Let’s assume a fraction p of all antibiotics used are used in animals, and a fraction 1−p are used in humans. (In your example, p=0.75.)
Let’s also assume that antibiotics use in animals is k× as effective at causing a resistance burden in humans per unit antibiotics used. (In your main example, k=0.1.)
Then the total resistance burden in humans is given by B=(use in animals)×(burden from animal use)+(use in humans)×(burden from human use), which in algebraic terms is B=p×k+(1−p)×1=1+p(k−1).
The fraction of the total burden caused by animal use is then A=pkB. If p=0.75 and k=0.1, this is A=0.0750.325≈23%. So, quite a bit more than 7.5%.
If k=0.01 (use in animals is 1% as efficient at causing a resistance burden in humans), then A=0.00750.2575≈3%. If k=0.5, A=60%.
So the fraction of the human burden caused by animal use could be quite high even if the per-unit efficiency is quite low.
One comment: while all your caveats about simplified reasoning and so on are well-made and still apply, I would generally be surprised if you could substitute a number like this in your analysis with another number three times the size, without affecting anything else, such that you could make the substitution and leave the wording unchanged.
That is to say, if a contribution of 7.5% was “very significant and worth pursuing”, I’d expect a contribution of 23% to be extremely significant, and worth making a high (or near-top) priority.
Of course, that’s the result for 10%, and 10% is just a made-up number. But I think the general point stands.
Good point. I’m unsure what the best practise in editing previous comments is—I don’t want to change it so much that the subsequent comments don’t make sense to another reader. Clarified now by leaving in the original number that fits with the reasoning around it while keeping the correction in brackets.
I think it would also be worth keeping in mind how hard it is to make progress on each front. Given that there seems to be widespread non-therapeutic use of antibiotics for farmed animals, and that (I believe) most people have accepted that antibiotics should be used sparingly, I would be surprised if there were no “low-hanging fruits” there. This is not meant as a complete solution, but rather is an attempt to identify our “next best move”. I would believe that cheap in-farm diagnostic tools should now be within reach as well, or already existing.
Separately from this, I admit being confused about the nature of the question regarding the importance of dealing with over-use of antibiotics in farmed animals. My understanding is that we know that inter-species and horizontal gene transfers can occur, so is the question about knowing how much time it takes? I just don’t have a clear model of how I’m supposed to think about it. Should I think that we are in a sort of “race against bacteria” to innovate faster than they evolve? Why would a delay in transfer to humans be a crucial consideration? Is it that antibiotic innovation is mostly focused on humans? Is there such a thing as a human-taylored antibiotic vs. farmed animal antibiotic? I suppose not? I cannot wrap my head around the idea that this delay in transmission to humans is important. So I guess I’m not thinking about it right?
[Added later:] Maybe the point is that if some very resistant strain emerges in a farmed animal species, we have time to develop a counter-measure before it jumps to humans?
Edit: I originally made mistakes in the calculation below, have edited to correct this. See comment below by willbradshaw for details of the calculation.
Thanks! I completely agree there are other strong reasons to reduce (or eliminate) factory farming.
About your other comment – I also don’t think the situation is reassuring at all. I think it’s very plausible that the antibiotic use in agriculture could be an important driver of antibiotic resistance.
I think that we need more research on both the jumping of species barriers and on horizontal gene transfer. This paper could be interesting if you want to read more on how common horizontal gene transfer is, but I haven’t been able to find anything that gives a good assessment on how important this is for resistance (I will be very grateful for suggestions if you or someone else know good research on this!).
I know an analogous problem is that human patients often develop resistant bacteria in the normal gut microbiota when they take antibiotics, and this could also be transferred to pathogens through horizontal gene transfer – again, we don’t know how much it happens. Another thing is that some bacteria can both be living in the environment or on animals or in the normal bacterial flora, and then act as pathogens when they end op in the “wrong” place – for example bacteria that is harmless in the gut flora could cause urinary tract infection. If these develop resistance, they don’t need to do any gene transfer but simply change location to cause problems.
Personally, if I try to speculate, I would reason that it’s very unlikely that antibiotic use on animals drives resistance in human infections as much per kg used as antibiotic use on humans. So if we assume that 75% of the kg of antibiotics is used on animals, I would say that it’s unlikely to drive as much as 75% of the resistance burden on humans. It could be that it is 10% as efficient in driving resistance on humans because of the “transfer barriers”, that would mean that 7.5% (CORRECTION: should be 23%, see comments below) of the resistance burden on humans would go away if we eliminated antibiotic use in agriculture. That would still be very significant and worth pursuing. Of course, I don’t know at all if the number should be 10% or 50% or 1%, and I could also not answer if that makes it a more significant driver than for example the misuse of antibiotics for viral infections.
Apart from speculative, this reasoning is also very simplified, since there are many different classes of antibiotics that target different bacteria. It is unclear which would be the best way to measure “resistance burden”. If we would mean it as QALYs lost, it will make a big difference which bacteria is resistant and to which antibiotic(s). In some cases resistance would just mean switching over the patient to another drug and the patient recovers a couple of days later. In some cases, such as with tuberculosis, resistance could mean the alternative treatment is prolonged by years, has serious side-effects and is also much more expensive. In some cases, the patient would die. Resistance to so-called last-line antibiotics is much more severe than others. This report lists the WHO priority pathogens, I think it’s quite good in how it gives an understanding of how much worse the treatment for resistant tuberculosis is compared to “regular” tuberculosis, for example (p 21).
Even if I think we need more research to understand this better (and thereby how to design interventions and allocate resources), I don’t think that should be taken as a reason to wait with reforming agriculture to eliminate (non-therapeutic) antibiotic use. We know the mechanisms are there, and we also know that it’s possible to remove the antibiotics from production, and I don’t think we can afford to wait and see. This urgency to achieve change is also why I think it unwise to tie this cause too closely to elimination of factory farming, for example – even if that would be good, I think it would take too long time, and I think it’s possible to eliminate the antibiotic use much faster if we keep that as a (relatively) separate issue. In context where one can be nuanced I don’t think it’s wrong to bring up antibiotic use and resistance as a negative effect of factory farming, but I think as you say that there are stronger arguments to front to end factory farming.
This feels a bit petty, since I don’t really disagree with any of your conclusions, but there are some mistakes in the mathematics here.
Let’s assume a fraction p of all antibiotics used are used in animals, and a fraction 1−p are used in humans. (In your example, p=0.75.)
Let’s also assume that antibiotics use in animals is k× as effective at causing a resistance burden in humans per unit antibiotics used. (In your main example, k=0.1.)
Then the total resistance burden in humans is given by B=(use in animals)×(burden from animal use)+(use in humans)×(burden from human use), which in algebraic terms is B=p×k+(1−p)×1=1+p(k−1).
The fraction of the total burden caused by animal use is then A=pkB. If p=0.75 and k=0.1, this is A=0.0750.325≈23%. So, quite a bit more than 7.5%.
If k=0.01 (use in animals is 1% as efficient at causing a resistance burden in humans), then A=0.00750.2575≈3%. If k=0.5, A=60%.
So the fraction of the human burden caused by animal use could be quite high even if the per-unit efficiency is quite low.
That was sloppy of me, thanks a lot for the correction! Edited in the comment.
Cool, thanks.
One comment: while all your caveats about simplified reasoning and so on are well-made and still apply, I would generally be surprised if you could substitute a number like this in your analysis with another number three times the size, without affecting anything else, such that you could make the substitution and leave the wording unchanged.
That is to say, if a contribution of 7.5% was “very significant and worth pursuing”, I’d expect a contribution of 23% to be extremely significant, and worth making a high (or near-top) priority.
Of course, that’s the result for 10%, and 10% is just a made-up number. But I think the general point stands.
Good point. I’m unsure what the best practise in editing previous comments is—I don’t want to change it so much that the subsequent comments don’t make sense to another reader. Clarified now by leaving in the original number that fits with the reasoning around it while keeping the correction in brackets.
I think it would also be worth keeping in mind how hard it is to make progress on each front. Given that there seems to be widespread non-therapeutic use of antibiotics for farmed animals, and that (I believe) most people have accepted that antibiotics should be used sparingly, I would be surprised if there were no “low-hanging fruits” there. This is not meant as a complete solution, but rather is an attempt to identify our “next best move”. I would believe that cheap in-farm diagnostic tools should now be within reach as well, or already existing.
Separately from this, I admit being confused about the nature of the question regarding the importance of dealing with over-use of antibiotics in farmed animals. My understanding is that we know that inter-species and horizontal gene transfers can occur, so is the question about knowing how much time it takes? I just don’t have a clear model of how I’m supposed to think about it. Should I think that we are in a sort of “race against bacteria” to innovate faster than they evolve? Why would a delay in transfer to humans be a crucial consideration? Is it that antibiotic innovation is mostly focused on humans? Is there such a thing as a human-taylored antibiotic vs. farmed animal antibiotic? I suppose not? I cannot wrap my head around the idea that this delay in transmission to humans is important. So I guess I’m not thinking about it right?
[Added later:] Maybe the point is that if some very resistant strain emerges in a farmed animal species, we have time to develop a counter-measure before it jumps to humans?