JAVMA paper linking Lasix use to sudden death in racehorses

The paper notes “horses racing while on furosemide medication were at 62% increased odds of sudden death” and calls for further in-depth research.

Yawn. Since most horses race on Lasix, I’m not sure that news is worth the headline.

So a 62% increase odds of sudden death is worth a yawn?


The paper did call for further in-depth research, which, in the context of racehorses dropping dead, is never a bad idea.



The actual paper is called “Fifteen risk factors associated with sudden death in Thoroughbred racehorses in North America (2009–2021)”.

15! Which include such things as distance of the race and length of time since the last race.

It doesn’t surprise me that everyone has jumped on the Lasix bandwagon and made it the headline–but it is far from the whole story that the article portrayed.


What I’m curious about is, does lasik contribute to the sudden deaths or the pre-existing conditions that warrant lasik. It’s entirely possible I missed that in the article.

1 Like

Is Lasix used for pre-existing conditions or is Lasix used in case a condition occurs?

It has to be administered by a veterinarian at the track in most, if not all, states. I do not know if there has to be a dirty scope in order to “qualify” a horse for Lasix; if that’s not a legal requirement then it certainly ought to be an ethical one. I’ll let those with actual hands-on experience answer that.

It is a fact that we have more incidents of bleeding in the US than in Europe/UK, simply because of the way our racehorses live. They’re stabled at the track 24/7, unless they get time off at a farm for injury or just some R&R. They train on dirt for the most part, which makes sense because that’s the same surface most of them compete on, but there’s more potential for irritating particles to get into the airway.

Considering up until a few years ago, like 99% of horses in the US ran on lasix, I’m feeling instant skepticism at the frequency of these lasix-linked sudden deaths.

The Bloodhorse article does a terrible job at communicating numbers paramount for understanding what kind of risk they are talking about.

How can we have appropriate and productive conversations about lasix when even industry publications sensationalize the topic?


It’s too bad the paper is behind a paywall. I’d like to see how the modelled this.

As always, it’s an association. (Firetrucks don’t cause fires!) Determining causality is a different kind of study.

Yes, I would like to read it, too- though not so badly I’ll pay or go crazy hunting it down.

I just don’t know how you can have a large enough sample size of sudden deaths in horses not on furosemide to draw any conclusions. They say the incidence of sudden death is 0.13 in 1000 starts. That’s not a big subset to draw data from, especially when you figure nearly all were probably on lasix given the fact that nearly all horses here run on lasix period.

1 Like

Let me see what I can do.

@jlm0305 really came in clutch!

Some interesting numbers:

The data followed 4,198,073 starts made by 284,387 different horses between 2009-2021.

There were a total of 536 sudden death incidents; 518 of those horses were on lasix and 18 were not.

Of the total sample, only about 5% of starts (233,276) were made without lasix.

I’ve forgotten most of what I learned in my stats classes, but I’m going to push back on the idea that we can extrapolate a lot about sudden death from such a small population of horses without furosemide. The authors acknowledge this themselves in the discussion.

This is not a pro-lasix post. This is a “let’s not fan the flames with sensational headlines without context” post.

The 14 other factors were nothing surprising. Season, surface, purse, age, sex, state-bred race v not, distance, number of starts in 30 days, number of starts between 90-180 days, claiming status, lifetime distance, lifetime wins, time since layoff, and on vets list recently. They started with 49 risk factors they analyzed, but those were the 15 shown to have statistical significance.

1 Like

Yeah, it’s a challenging dataset.

The authors acknowledge many of the issues and offer some feasible hypotheses about what they see. This clearly an exploratory modeling study–like data mining. Run it again and again and sixteen different ways and see what you can glean about building a theory. There’s nothing wrong with that, methodologically, but it does reduce the interpretive value of p values (significance).

The authors also acknowledge the instability of the model. Variables apparently change significance based on which variables are left in and which are removed from a particular model run. That just screams multicollinearity (that variables are not independent so significance can be questionable). Not at all unexpected. They don’t report details on the diagnostics (perhaps because of article length limits–nothing nefarious), but I bet it’s pretty messy. Many models of complex phenomena are. Probably need a different model form. If it were MY study, I’d be digging around in the residual errors. The unexplained part. That’s were the juice will be.

What I think is more interesting are the findings related to a horse ever having been on the vet list and days since the horse was on the vet list.

If you wanna be data driven and think banning lasix is important you need to take a hard look at listed horses and the time from when they’re listed to when they raced (and died). That may well be the ‘signal’ and race day lasix is actually ‘noise’.

Lots more work to be done. Very cool that the data are available to do it. Thanks @jlm0305!

1 Like

I agree.

There is an issue with that subset IMO that further limits takeaways from this particular examination of data:

There were 233,276 starts without lasix (starts, not total horses) within the time parameters of the study. In 2020, regulations on lasix for 2 yos changed, with most (all?) jurisdictions prohibiting lasix in 2yo races. In 2021, those regulations expanded further to prohibit lasix in all triple crown prep races and most stakes races. Though 2020 and 2021 were atypical racing years thanks to covid, those changes certainly altered the profile of what a horse running without lasix looks like. The expanded regulations and changing attitudes also likely increased the total number of starts without lasix rapidly in a short period of time.

Going forward, we might actually have enough horses running without lasix in the US to draw some conclusions about whether lasix is a signal or noise for sudden death.

1 Like

Agree. The data are complex. Lots going on in there.

Here’s an article written in response to the JAVMA paper, criticizing its methodology and its results.

Letter to the Editor: Doug Daniels, DVM, National HBPA President (thoroughbreddailynews.com)


Not surprising to see blowback. I’d rather have statisticians critique the method than vets, though. :face_with_monocle:

The critique about bias and sponsorship is interesting. I’d say that JAVMA needs to assess that. They had to know this paper would get some attention.

1 Like

the minute we start viewing the current practices as the “norm” for the foreseeable future, is the day we stop making the sport better for the horse. I was expecting blowback on the article. It may be a widely used drug and a widely accepted one. That does not mean it doesn’t come with inherent risk or long term health risks. It also doesn’t mean that potential risk factors should never be investigated or deemed significant enough to potentially stop use of a drug. This type of scientific investigation is important and it needs to continue.

No one said it risk factors shouldn’t be investigated or deemed insignificant.

Framing the results as furosemide leads to 62% increase in sudden death without further context doesn’t accurately represent how often this is occurring. That increase was derived from a tiny subset of horses that was predominately comprised of first time starters for the first ten years of the data collection.

These conversations are important, but they will only be divisive if the data set is sensationalized.