Dog saliva probably passes you by unnoticed — until it winds up on your hands or clothes — but for people studying dog well-being, dog spit is a highlight of their day. This is because saliva is easy and informative: it’s easy to collect, easy to store, and easy to test for its concentration of cortisol — a measure widely used as an indicator of stress and welfare. But even though researchers are up to their (metaphorical) elbows in dog spit, there’s still a lot researchers don’t know about the relationship between salivary cortisol and dog well-being.

Enter Mia Cobb of the Anthrozoology Research Group at Monash University in Australia and Nancy Dreschel from the Department of Animal Science at Penn State University. Along with their co-authors, they embarked on a meta-analysis to establish a reference range for dog salivary cortisol and to explore the relationship between cortisol and a variety of dog, experimental, and environmental parameters. The study was recently published in Domestic Animal Endocrinology and is available open access through August 1, 2016. Download the PDF now and then come back to this post. I’ll wait.

A meta-analysis is just a fancy way of pulling together and analyzing the data from many different studies on a similar topic. Studies measuring salivary cortisol have explored a wide range of topics, so a meta-analysis of these studies is useful because it’s a “formal, quantitative statistical technique for combining the results from multiple independent studies to systematically derive conclusions about that body of research.” This meta-analysis aggregated raw data from 31 different studies investigating a multitude of scientific questions pertaining to thunderstorm phobia in dogs, dog-dog interactions at the park, play groups at an animal shelter, dogs with long- or short-stays at shelters, working dogs living in kennels, dogs participating in animal-assisted interventions, and healthy pet dogs visiting an animal hospital, among other topics. Here are some of their findings:

More than a number

Salivary cortisol is not like cholesterol (if you too cross your fingers when finding out your cholesterol, maybe you know where I’m going with this). Cholesterol has the following guidelines for humans: under 200 mg/dL is desirable and 240 mg/dL and above is high. Cobb and colleagues found that in dogs, cortisol readings are susceptible to both intra- (within) and inter- (between) individual variability. A high reading is not universally ‘bad’ or an indicator of ‘bad welfare,’ and a lower reading is not necessarily ‘good,’ as people might initially think.

“Although the idea of an easy test for stress and welfare in dogs is incredibly appealing to researchers, working dog industries and pet owners alike,” Cobb explains over email, “there is no perfect test that gives a magic score to determine good welfare or distress in dogs.” Because of this individual variability, they highlight that “care should be taken when using cortisol studies for dogs at a group or population level.”

The meta-analysis also identified other factors that could influence salivary cortisol levels, in a way that smaller studies using cortisol could not establish. Dogs whose owners or primary handlers were present had significantly lower salivary cortisol concentrations than those whose primary people were absent at the time of testing. This highlights an environmental and experimental factor that researchers should consider.

“We're still learning about the influence of things like individual personality on how animals express stress, behaviourally and physiologically, so it's important to look at the whole picture when assessing dog well-being, using multiple measures,” explains Cobb. She adds, “It's crucial researchers report on all the factors this study has shown to significantly affect cortisol concentration, both to aid interpretation of their study and to help future researchers compare their own findings with other studies.” For example, they were surprised to find that intact females had higher salivary cortisol concentrations than spayed females, intact males and neutered males, highlighting the need for studies to take sex and neuter status into account.

The next finding is probably going to throw some for a loop: dogs living in animal shelters for more than two weeks had significantly lower salivary cortisol concentrations than dogs in private homes or working dog facilities. Animal shelters are typically not great experiences for dogs because shelters are marked by new and noisy everything, a general lack of control, and even the loss of a human social partner. Lower salivary cortisol could indicate ‘dysregulation’ of the stress response, the researchers suggest, in accordance with what has been observed in other studies. Lower concentrations in dogs at shelters could indicate that they’ve fatigued their hormonal stress response to the point that it is not functioning normally. Cobb and colleagues explain that something similar is observed in humans who are chronically stressed or unable to sleep.

What’s next? 

The meta-anlaysis highlights a need for “interdisciplinary dialogue and to read research outside of our own species of interest,” Dreschel adds over email. In the meta-analysis, they discuss that “both high and low cortisol concentrations can be problematic or advantageous to humans, depending on a multitude of individual factors such as personality, timing, context, prior stressors, and life history.”

“Just as human researchers have depended on lab animal research for insight into human physiology and psychology, we need to look to non-canine models, including rodents, humans, and other species, to better understand our canine companions,” explains Dreschel.

Salivary cortisol needs a context, adds Cobb: “Beware the headline that declares all dogs love or hate ‘X’, especially if they’ve only used one measure, like salivary cortisol. Good research will use multiple measures of physical and behavioural well-being. If the study you read about in the media has only used one kind of measure (like salivary cortisol), then the findings relating to stress or welfare should possibly be considered preliminary, and probably shouldn't be generalised outside of the study population.” 

Additional reading

Cobb, M. 2013. Throw another dog in the (data) pool. Do You Believe in Dog? blog

Hekman, J. 2009. Spit Girl. Dog Zombie blog

Hekman, J. 2014. Designing stress studies, part 2: how do you get your sample? Dog Zombie blog


Cobb, M. L., Iskandarani, K., Chinchilli, V.M., and Dreschel, N. A. 2016. A systematic review and meta-analysis of salivary cortisol measurement in domestic canines. Domestic Animal Endocrinology 57, 31—42. 

Thanks to Mia Cobb, my Do You Believe in Dog? colleague, for discussing her research with me and contributing to this post.