As we start World Antimicrobial Awareness Week, I’m going to  try to write a few posts about various aspects of antimicrobial use and resistance in animals. This first post is a bit long, but it covers an important topic: antimicrobial use targets.

I get asked a lot about whether antimicrobial use targets are need in veterinary medicine. My typical answer is, “Yes, we need them, but we can’t set good targets – yet – in most situations.”

In Canada, we have no way to set proper targets because we don’t have enough good data on current antimicrobial use and efficacy, and we don’t know what targets would be useful. Targets need to be more than a political or marketing statement. If we were to say “Yay, we reduced antibiotic use in animals by X%,” but we have no idea if it was useful in terms of the animal health impact or reduction of antimicrobial resistance, so what?

In concept, antimicrobial use targets are simple: we set a level of use that we want to be below, at a national, sector (e.g. beef cattle), prescriber or farm level. However, it’s a lot more complicated than it appears.

  • Targets can be helpful, when they improve antimicrobial use, but “improve use” is poorly defined. It’s typically used synonymously with “reduce use” but they’re not always the same. How and when we use antimicrobials, how long we use them and which antimicrobials we choose are as, or more, important than the total amount.
  • Targets can also be bad, when they compromise animal health and welfare, and/or result in shifts in drug use that reduce the total amount of antimicrobials used, but actually worsen how and what antimicrobials used.

Most often, the focus of antimicrobial use targets is the total mass of drugs (kgs, tonnes) used. Why?

Total mass of drugs sold / prescribed / dispensed is usually the easiest number to get, and it’s often a big, scary number that gets attention. At the crudest level, it can be reporting of the total mass of antibiotics used in animals, overall. That’s pretty useless. We can get more information by looking at use in specific populations or animals.

My personal opinion is that mass-based metrics (the main way used to measure and set targets for antimicrobial use) largely suck. I use them, but that’s because it’s typically all we have, but they have major limitations.

What are some of the limitations of mass-based metrics for measuring antimicrobial use?

Mass just tells us how much antibiotic went into the whole animal population. It doesn’t tell us why, when or how.  Mass data of all antibiotics combined is particularly useless, because we use different dosages for different antibiotics, we treat different species differently, and there are large differences in sizes of different animals, from grams (e.g. chicks) to hundreds of kilograms (e.g. cattle).

Differences in drug potency also make it extremely difficult to interpret such a measure. For example, if I change from a drug that is dosed at 20 mg/kg twice a day to one that is dosed at 2 mg/kg once a day, I’ve reduced the mass of drug used by 95% (40 mg/kg/day to 2 mg/kg/day). That sounds great at first glance; however, it could actually be the last thing we want, since higher potency drugs (that are used at a lower dosage) are usually newer, broad spectrum, higher-tier drugs that are more important in human medicine as well. So, if a target was set to reduce antimicrobial use in Canada by 50%, we could do that quickly by switching to higher potency drugs, but it would actually increase the risk of emergence of serious antimicrobial resistance to more important drugs.

The table below shows an example of antimicobial use in dogs in two different scenarios. I took some of our published data about treatment of dogs for bacterial cystitis (Weese et al. JVIM 2021) and calculated the total antimicrobial use for 1000 dogs based on the study data, versus following the recommended ISCAID guidelines for antimicrobial treatment of dogs with cystits instead.

There’s a big difference in total antimicrobial use between the two scenarios, but the totally actually increases when the guidelines are followed! Why? Because use is shifted away from the newer, broad spectrum, higher-tier drugs that are dosed at lower mg/kg  than older drugs that are similarly effective. So, this change would probably be beneficial from a resistance standpoint, but would look terrible if we only focused on total drug mass used. I added a scenario at the bottom of the table where our intervention also dropped the treatment duration from 10 days to 5 days, to highlight how important duration of treatment is.  This should be a big focus for companion animal interventions since we still tend to use excessively long treatment courses.

Overall drug mass data can be useful if we’re comparing apples to apples, such as looking at year-to-year total kgs of antibiotics used in a certain species if the population hasn’t changed and if relative use of different drugs hasn’t changed. Mass-based measures of antimicrobial use have been used successfully to reduce use in livestock in Denmark, through their “Yellow Card” system. So, I don’t discount the utility of more refined mass-based approaches, but we can do better.  Overall mass of drugs used can be misleading in some situations, and doesn’t guide optimal stewardship activities. Ultimately, it’s a crude measure that can be hard to use outside of situations where there’s a very narrow and well defined scope of use.

There are additional metrics that can be used to provide more refined information, such as mg/PCU (population correction unit), defined daily dose (DDDvet), defined course dose (DCDvet) and animal daily dose (ADD). These all attempt to provide more context to the results, but each has its own limitations. They’re still better than just looking at total mass of drug, but even with these metrics we’re still left with a lot of gaps in understanding how antibiotics are used.

Ok, so I’ve whined enough about how mass-based metrics suck (or at least aren’t good enough). What can we do?

There’s not a single, one-size-fits-all measure that works best for all species, situations and goals, but we do have options. While everyone wants a single number on which to focus, that’s usually not be possible. We need to use combinations of metrics, including mass-based and dose-based metrics, and ideally also look more at how we use antibiotics and areas that are more amenable to action-based targets.  Here are some examples of things we could assess:

Percentage of animals treated

This can be a very practical and useful approach, looking at either an entire animal population or a more targeted subpopulation (e.g. a farm or production group). For example, we could look at the percentage of feedlot cattle that get antimicrobials for respiratory disease, or the percentage of dairy cattle that get antimicrobials for mastitis during a lactation. In companion animals we could look at things like the percentage of cats with lower urinary tract disease that get antimicrobials (which should be very low, because most don’t have infections), the percentage of dogs with acute diarrhea that get antibiotics (ditto) or the percentage of animals undergoing surgery that get peri-operative antimicrobials. This measure can tell us more about specific use and provide useful comparisons. If a feedlot treats 95% of their cattle for pneumonia, but a similar sized feedlot in the same province only treats 50% of their animals, that’s a relevant difference. If the farms are truly similar, it tells the farm treating 95% of cattle that they can likely reduce their use a lot through management changes or being more selective about which animals are treated (and they could look to the other farm for examples of what they could change). It’s directly actionable information that doesn’t necessarily aim to penalize the first farm, but can ideally spark some action.

Appropriateness of use

This is a big one for me. If we can measure how often antimicrobials are used appropriately (i.e. not just what and how much, but how they are used and, importantly, whether use fits with standard practices), we have a great metric to track. For example, knowing that 100 animals in a group were treated is somewhat useful, but knowing that 80 animals were treated according to guidelines and 20 were not tells me a lot more, and lets us focus our efforts on the situations where treatment wasn’t consistent with guidelines.

Who defines “appropriate” antimicrobial use and how that’s determined then become the big questions. Ideally, we’d look at what percentage of antimicrobial use follows accepted guidelines, but that only works if we have good guidelines. We have guidelines for use in some situations, but for others, so it’s not an instant fix.  We’ll never have guidelines that cover everything, but we should still strive to look at the appropriateness of use in common situations.

What do we need to have useful antimicrobial use targets of any sort?

  • A robust, accurate, secure and user-friendly data collection system that lets us efficiently measure antimicrobial use at the animal/veterinary/farm level
  • Political will to act (in Canada, our federated system means we require buy-in from all provinces, territories and the federal government)
  • Adequate and sustained funding
  • Effective targets that will actually have an impact on antimicrobial resistance (and not just another metric)

In the agriculture sector, we also need (well, maybe don’t actually need but really want) buy-in from industry groups. There will always be some hesitation to participate from groups that don’t want external interference or monitoring, but proper surveillance and targeting can be useful to them as well.  Ideally targets aren’t punitive, they’re informative. Knowing how a farm uses antimicrobials and how that compares to similar farms can be useful for assessing a farm’s practices. If one farm uses 50% more antibiotics than a similar farm, they need to think about why.

  • Are there management issues that can be improved so antimicrobials aren’t needed as often?
  • Are antimicrobials being used when they aren’t necessary due to misinformation or practices that simply need to be updated?

Having data can help figure out what’s happening, which can lead to action and ultimately improve production and save a lot of money on antimicrobials (and other veterinary costs). There will also likely be more scrutiny from consumers and trade partners about how antimicrobials are used. The goal of antimicrobial use targets doesn’t have to be restrictive. Yes, there could ultimately be an approach that penalizes people for being above the target level, but that’s not what targets are really about. They are meant to help us understand how we are using antimicrobials and identify ways to improve and monitor how we’re doing.

Back to the initial question: Do we need veterinary antimicrobial use targets?

  • Yes.

Can we set them now?

  • Not in Canada.

What do we need?

  • Development of a system to efficiently monitor antimicrobial use, and political will to put the time and effort into creating that system. Then we can think about how to use those data to improve things.