A wide range of small studies have lately been suggesting that children who own or interact with a pet have better physical and psychological health. But the largest study of its kind to date now disproves this hypothesis.
The new study was carried out by researchers at RAND corporation, a nonprofit think tank and part of RAND Health, which is an independent healthy policy research program.
The new research brings advanced statistical tools such as double robust regression analyses to the study of this topic, which the scientists used to account for other factors that may influence a child’s health rather than pet ownership, such as family income.
Additionally, to the authors’ knowledge, this is the largest statistical study to investigate the link between children’s health and family pet ownership.
The first author of the study is Jeremy N. V. Miles, and Layla Parast, a statistician at RAND, is the corresponding author for this research.
Numerous small studies – referenced by Miles and colleagues – have suggested that owning a pet may improve the health and psychological well-being of children.
However, most of these studies, say the researchers, have been subject to two main flaws: firstly, they did not properly account for the so-called selection bias or the issue of confounding – that is, factors such as family income that may bias the results.
Statistically, a solution to this problem is applying “propensity scores” – an approach that is typically used to allow researchers to calculate the probability that a person, for instance, might be treated differently based on bias-inducing traits such as age or gender.
But, the researchers said, few of the studies analyzing pets’ effect on the health of children have used propensity scores.
Miles and colleagues analyzed data from 2,236 households that owned either a dog or a cat and compared them with 2,955 households that did not have a pet.
The researchers obtained the data from the 2003 California Health Interview Survey – a large, population-based, random-digit dialed survey of families.
The survey gathered information about the health status and health-related and psychological behaviors of the families interviewed. Despite the fact that the survey was carried out in more recent years, the 2003 survey was the only one that contained a question about cat and dog ownership.
Miles and colleagues narrowed down their research to families that had at least one child between 5 and 11 years old.
Questions assessed by the researchers included inquiries about the overall health and well-being of the child, whether the child had received an attention deficit hyperactivity disorder (ADHD) diagnosis, and questions about whether or not the parent had any concerns about the mood, feelings, and behavior of the child.
In terms of the statistical analysis, the researchers used “survey-weighted linear and logistic regression analyses” with pet ownership as the main variable. Survey weighting is often used when statisticians have to estimate regression models based on survey data.
The authors explain that as opposed to most statistical studies, which use most of the available control variables to adjust for possible confounding factors, the current study used a more advanced statistical tool called
This approach used propensity scores and weighted the regression models so that “those with a pet were comparable with those without a pet on all available confounding factors in the data.”
Parast explained to Medical News Today what “double robust regression” means, saying, “The approach is robust in the sense that sometimes, when one ‘adjusts’ for these kinds of factors, it is done by simply adding these factors to a specific regression model.”
“But for this to be appropriate, your assumed model has to be correct. For example, the model may be assuming something like a linear relationship between income and the likelihood of owning a pet. Our approach makes fewer modeling assumptions and is thus more robust to incorrect model specification. “
Overall, the scientists accounted for more than 100 confounding factors that could influence the results, including income, language skills, and the type of housing they lived in.
The study found that, as expected, children in families that owned a pet were in better health and tended to be more physically active than children in families without a pet.
Furthermore, children in pet-owning families were more likely to have ADHD, but their parents were less likely to be concerned about their mood, feelings, behavior, and ability to learn.
However, after the researchers adjusted the findings using the double-robust approach and including propensity scores, the link between pet ownership and children’s health was no longer statistically significant.
These results are more reliable than those of previous research, the scientists say, because their study is the largest of its kind to date.
“We could not find evidence that children from families with dogs or cats are better off either in terms of their mental well-being or their physical health […] Everyone on the research team was surprised – we all have or grew up with dogs and cats. We had essentially assumed from our own personal experiences that there was a connection.”
Layla Parast
Parast also spoke to MNT about the limitations of her and her team’s study, saying, “Our main limitations are that (1) we do not have information about how long the family owned the pet or how much interaction the child had with the pet, and (2) we do not have information on long-term health outcomes.”
Therefore, she added, “we would love to see future work examining this association where more detailed information about the pet ownership […] and long-term health and social outcomes could be measured.”
The most accurate test for whether or not pet ownership improves children’s health, the authors say, would be a trial wherein families are randomly assigned a pet and control families are not. Such a randomized trial would have to follow-up on the family’s health for 10 to 15 years, the authors say, but this is not financially feasible.