NO, it did NOT!
Even though the paper said that, it’s not what they actually found. Unfortunately, a lot of people are reading that headline, or worse, they’re reading “…yada yada… dogs spreading SARS-CoV-2… yada yada.”
What did the study really find?
Let’s break down some important aspects of the paper on which this headline is based. The study, entitled “The spread of SARS-CoV-2 in Spain: Hygiene habits, sociodemographic profile, mobility patterns and comorbidities“(Rodriguez-Barranco et al. Enviro Res 2021), reported that people who walked their pets were 78% more likely to have reported having had or maybe had COVID-19.
- They didn’t investigate that any further, and it’s not clear what “walked their pets” entailed (e.g. walking a dog once outside the house vs walking a dog on a regular basis).
- The question, strangely, asks about walking “pets” not “dogs.” It’s reasonable to assume that the pets were at least mostly dogs, but they didn’t actually specify that.
- How the authors analyzed the data is also unclear. The answer options for the pet-walking question were yes / no / I don’t have a pet. However, for analysis, they combined pet owners who didn’t walk their pets and non-pet owners (further demonstrating that they did not look at the risk of living with a pet).
- They said that 6.9% of people who walked their pet had been infected, vs 4.2% of those who did not take their pet for a walk, despite the fact that most of the latter group actually didn’t own a pet at all.
So what they really looked at was leaving the house with a pet (vs living with a pet), and that raises some serious questions about how clearly they thought about the results. The focus should be on the “going for a walk” component, since that’s what they actually studied. Unfortunately, they didn’t also ask if people went for walks without pets.
Some of the other study results also raise more questions than answers:
- They reported that people who used home delivery for food were almost twice as likely to have had COVID-19. Does that mean they were getting infected by delivery people and would have been safer shopping? Presumably not. My concern is that there was some reason that people were more likely to order food, and that was also a risk factor for COVID-19. For example, if they knew they had been exposed or were in some other high risk situation, that might lead people to avoid stores and also be at higher risk of being infected.
- Another big issue is the fact that people with COVID-19 were presumably more likely to order delivery after being diagnosed. The survey doesn’t ask what they did BEFORE getting infected (if they had COVID-19), just what they did during Spain’s period of restrictions. So, finding increased risk from home delivery might actually be because people who were more likely to use home delivery were otherwise higher risk or already had COVID-19.
Another concern is who they surveyed. The study population is critical for any study like this. You need to understand the study population to understand the results and any potential bias. You can get really misleading information or not understand your results if you don’t understand the group of people that provided them and how they compare to the general population.
Why is the study population so important? Here’s an over-the-top example to illustrate: Let’s say a study said “pet owners were significantly more likely to say their dog was an important part of the family compared to cat owners,” but the study only enrolled people through websites like www.ilovemydog.com and www.mycatisademonicpsychopath.org – you can see how we might end up with a biased understanding of the situation.
In the discussion of the Spanish study, the authors mention that most respondents were graduate or post-graduate students, which is a pretty specific group. We have to consider how well they represent the general population. The farther away they are from average, the less confidence we have in extrapolating results to anyone other than graduate and post-graduate students. I’m not saying there’s a problem using this study population. What I’m saying is we just don’t have enough information to know what it means. That’s one of the (many) things I’d flag reviewing a paper like this.
Survey response rate (and response bias) is also an issue, but I deleted my detailed comments on that since this post is already getting quite long (and probably a bit dry).
The way the authors wrapped things up is a big issue for me.
In the discussion they say, “These results point to living with dogs as a strong risk factor for COVID-19 infection.” Their concluding statement was, “The results of this study demonstrate that living with dogs… have been the main routes of transmission of SARS-CoV-2 during the most restrictive period of confinement in Spain.”
Neither of those is true.
Pet ownership was not associated with increased risk of COVID-19. Their statistical analysis of pet ownership did not identify any risk. Walking a pet was a potential risk factor, not owning or living with a pet. There’s a long paragraph in the discussion talking about risk from dogs, despite the fact the paper didn’t actually look at that, and they did not find a risk from pet ownership.
So, what does this study tell us about pets and risk of COVID-19?
It’s hard to say… probably nothing.
This study raises some interesting questions but doesn’t provide many answers. It certainly doesn’t provide answers about risks associated with pets. A more rigourous peer review could have helped catch and address some of those issues.
The study does NOT show pet ownership was a risk factor for COVID-19.
- If the pet walking risk factor is real, I suspect the critical factor is more “walking” than “pet.”
I’ll stick with the same messaging I’ve had for months about animals and COVID-19:
- If you have or might have COVID-19, stay away from animals (human and non-human).
- If your pets have been exposed to someone with COVID-19, keep them away from others (just like you would if you or your kids were exposed).