The Study so far
The Study so far
A weighty issue solved
The limber tail results are in…
View our facts and findings to date
The size of Dogslife dogs
Dogslife helps find genetic cause of overweightness
Dogslife study findings so far
Cohort study paper
Our first results
At Dogslife, we have millions of pieces of information in our database. This information has been provided by our members over almost 10 years, which we are very grateful for. However, as our dataset grows it does become more difficult to manage. Everyone makes mistakes sometimes, and this shows in our data, particularly in the weights and heights of dogs. These mistakes can be caused by lots of reasons, including typos, faulty measurement equipment or entering the wrong unit type (e.g. pounds instead of kilograms).
Charlotte Woolley and the Dogslife Team have recently published a paper about a new ‘data cleaning’ (literally cleaning, but for data) method that identifies and corrects these errors in weights and heights. We used 7 years of Dogslife data to develop this method, which contained over 43,000 weights and 28,000 heights! We also used data from three other canine and human health studies to check that it could be used elsewhere. Our method was able to find at least 26% more errors than other methods and did not identify as many errors incorrectly.
We expect our method to be useful in lots of other research that collects similar measurement data to Dogslife and in other fields too. You can read more about Charlotte’s paper here: https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0228154.
Thanks again to all the owners who provided us with information about their dogs’ weights and heights: without your input and honest mistakes there would have been no opportunity to improve these methods!
The Dogslife team published a scientific report in May 2017, describing how dogs are affected by gastrointestinal upsets in the UK, and what factors are associated with developing them. Dr Carys Pugh analysed the records of over 6000 dogs registered with the project, detailing over 2600 separate reports of vomiting and diarrhoea.
Like children, episodes of gastrointestinal disease typically affected very young dogs most frequently, with the peak rate of diarrhoea and vomiting episodes being between 3 and 6 months of age. With the number of diarrhoea episodes per year gradually declining as dogs reached maturity before increasing again as they approached middle age (5-6 years of age).
One of the key findings was that people most often look after their dog at home when they present with these signs, rather than take them to the vet. For over 60% of diarrhoea events and over 70% of vomiting events, the dog was not taken to the vet, although this dropped to 50% when vomiting and diarrhoea occurred together. This highlights that studies based only on data collected from veterinary practice records can miss the vast majority of times that the dog has these illnesses and furthermore means that capturing information from dog owners directly is the only way we can study the true burden of these diseases as most episodes never reach veterinary records.
If you’d like to find out more details about the study see our June and July newsletters, or you can read the scientific report here: https://linkinghub.elsevier.com/retrieve/pii/S0167-5877(16)30435-4.
… and it’s more common than we thought. A lot more common. In a paper published by Dr Carys Pugh and the Dogslife Team, we found that nearly 10% of all Labrador Retrievers taking part in Dogslife have an unexplained tail flaccidity (which we call limber tail). Thanks to those of you who filled in extra questionnaires about limber tail and whom sent DNA samples to us, Carys found that swimming was a risk for dogs developing limber tail (i.e. dogs which developed limber tail were more likely to go swimming), although it was not the only factor (as a quarter of dogs did not go swimming before they developed the first episode). Dogs which were affected tended to live further north, supporting the previous anecdotal suggestions that cold weather might be a factor in dogs developing the condition.
Furthermore, dogs which were affected by limber tail were more likely to be related to each other, which suggests that there is a genetic component to the condition, which we are currently investigating. Owners generally described the condition as painful, and as having a moderate effect on their dog’s quality of life. Interestingly, the condition rarely resulted in presentation to veterinary surgeon, or indeed documentation in their records, which explains why it has been overlooked in the scientific and veterinary medical literature to date.
The search for a genetic risk for limber tail is ongoing at The Roslin Institute. You can read more about Carys’ paper here: http://veterinaryrecord.bmj.com/content/early/2016/06/27/vr.103729.full.pdf, or if you want to read more about the condition please see our previous newsletter article on it here: http://www.dogslife.ac.uk/newsletter/view/17#tail. Thanks again to all the owners who provided additional information and a sample to help with this study.
Dogslife has found out lots of amazing facts and figures about the lives of Labrador Retrievers living in the UK. Read all about it here in a poster we’ve produced which explains what we’ve achieved with participants in the project so far.
Thanks to all the information our members have kindly provided on their dog’s height and weight we can show you the average heights for dogs taking part in the study for the first three years. Don’t worry if your dog is above or below the average in either of the graphs below, because just like us, Labrador Retrievers come in all shapes and sizes! Thanks again for all the data entered to the study website.
Thanks to the incredible dedication of owners taking part in the Dogslife study, the project has helped scientists discover why some Labrador Retrievers become overweight. Using DNA samples sent to us by hundreds of Dogslife owners in 2013, scientists were able to identify and confirm that a genetic mutation causing an increase in appetite was present in over a fifth of all Labrador Retrievers. Many owners also provided details about their dog’s appetite by filling in an additional questionnaire posted to them, which helped the scientists to measure the “motivation” of each dog to eat.
The mutation was discovered in a gene called pro-opiomelanocortin (POMC), which is known to be involved in the signalling pathways in the brain which recognise hunger and the feeling of being full after a meal. In some dogs, the gene was found to have a deletion in it (a mutation where part of the gene is missing). Dogs who carried a copy of the mutated gene had higher body conditions scores, higher body weights, and higher food motivation scores than dogs who had normal copies of the gene. The mutation also had an “additive” effect, so dogs carrying two copies of the gene had higher body condition scores, body weights and food motivation score when compared to dogs carrying a single copy of the mutation, or no copies of the mutation.
Assistance dogs (Guide Dogs) carried proportionally higher numbers of copies of the mutated gene when compared to general populations of Labradors Retrievers (such as Dogslife dogs). It was hypothesised that this may be an effect of selecting the assistance dogs with temperament and trainability by using food as a positive reward. The mutation was also found in Flat-coated Retrievers.
If you want to read the scientific paper detailing the findings of this study, you can find it here: http://www.sciencedirect.com/science/article/pii/S1550413116301632. Thanks again to all the Dogslife owners who provided samples, and those who also completed the additional questionnaire on appetite, to enable us to uncover these scientific discoveries. The information for this study was anonymised so we are not able to report to owners whether their dog carries the mutation, or what their dog’s food motivation score is, although some owners may be able to guess! This work highlights again how Dogslife is enabling scientific discoveries which allow us to understand more about the health of Labrador Retrievers.
In July 2015 we published a scientific paper about the height, weight and lifestyle of dogs participating in Dogslife, up to their fourth year of age. We were surprised to find that, on average, adult male Dogslife dogs were 59cm or 23.3 inches in height, which is 2-3cm (about 1 inch) taller than the Kennel Club breed standard (which is 56-57cm or 22-22.4 inches). Only 13% of dogs whose height was reported to the study after they reached one year of age actually met the Kennel Club breed standard. The female Dogslife dogs were, on average, 55.1cm or 21.7 inches tall which meets their breed standard (55-56cm or 21.6-22 inches). Again, there was a lot of variation between dogs, and only one fifth of individual female heights after one year fell within the breed standard range.
In terms of weight, Labrador Retrievers are one of the dog breeds that are thought to be more at risk of developing obesity. We found that at two years of age, on average, female dogs weighed 26.8kg and males weighed 31.6kg. We also found that dogs kept putting on weight at nearly 0.9kg per year between one and four years of age. Working dogs weighed more than 2kg less than pet dogs, and having another dog in the household was associated with dogs being lighter (0.48kg) when compared to households which had no other dogs. The big surprise was finding that chocolate Labrador Retrievers weighed nearly 1.4kg more than yellow or black dogs (on average). We hope to find out whether there is something about being chocolate makes the dogs heavier, or whether it is simply that in breeding for chocolate Labrador Retrievers, we have intentionally or unintentionally selected for heavier dogs.
The types of food reported to Dogslife were dried only (80%), a mixture of dried and wet (13%), raw (1.9%), home prepared (1.1%), wet only (1.1%) and other (3.0%). The majority of dogs (72%) did not have varying diet types with 66% eating a consistent diet of dried food. The plot shows that daily feeding frequency decreased as the dogs aged and settled at twice daily for most dogs at between six and nine months.
After the Dogslife dogs reached one year of age, 56% of females and 47% of males were reported to have been neutered. This increased to 64% and 52% for dogs over two years of age. The median* age of neutering was 282 days for males and 297 days for females – both lying between 9 and 10 months of age. [* Median is a type of average that looks at the middle of a distribution. In this case, half of the Dogslife dogs that were neutered were neutered before the median and half were neutered after.]
There were a lot of interesting findings about the lifestyle data and, in particular, there were some geographic differences. For instance, we found that dogs in Scotland and Wales spent more time each day being exercised than dogs in England. Overall, the dogs were exercised for an average of over two hours each day.
Most of the Dogslife dogs slept indoors alone (55%) but large numbers also slept indoors with a person (and possibly another pet) (21%), or indoors with another pet only (19%). Just 4.3% slept outside, but they did not typically sleep outside all of the time. There were yearly peaks in dogs sleeping outside in August 2011 and 2012 and July 2013. Dogs that slept outside at least once were more likely to be from Northern Ireland than elsewhere in the UK and much more likely to be working dogs.
If you would like to read the full version of the paper containing our findings from the study so far, please visit http://www.sciencedirect.com/science/article/pii/S0167587715002329. Many thanks again to all our members for the valuable information provided to the Dogslife project on their Labrador Retrievers.
In May 2014, Carys Pugh (who was then in the third year of a PhD on the Dogslife project) had a paper published in the on-line journal, Canine Genetics and Epidemiology. This paper sought to answer the question of what we can learn from studies like Dogslife. Carys reviewed the literature published about cohort studies of dogs (i.e. studies which look at groups of dogs for periods of time) to examine the benefits of wider use of such studies in the future.
Cohort studies, like Dogslife, offer the investigators the opportunity to assess the relationship between risk factors (such as exercise or feeding) and diseases over time. Historically, cohort studies of people initially investigated specific exposures, for example the impact of smoking on the risk of developing lung cancer, but there has been a movement in recent years to more broadly assess the impact of complex lifestyles on health. These studies generate new theories about the impact of environmental and genetic risk factors on disease.
Carys found that although cohort studies have been few and far between in dogs, they have made a valuable contribution to the understanding of dog health in areas such as infectious disease, for example in the control of the infectious disease Lieshmaniasis. The role of specific effects, such as the effects of neutering and dietary restriction on dog health, have also been directly investigated. Following the trends in human health, large cohort studies of dogs have been set up to assess the wider impact of dog lifestyle on their health.
The biggest cohort study of dogs to be set up to date is Dogslife. Thank you to everyone who supports the study and provides vital information about their dog’s health and lifestyle to help us identify the factors that affect dogs’ health over their lifetime. You can read Carys’ paper in full here: http://www.cgejournal.org/content/1/1/5
The Dogslife Project relies on Dogslife members for all the data collected and the study is an example of ‘citizen science’. One of the crucial aspects of using this type of data collection is working out how well the data recorded reflects what is really happening to the dogs taking part in the study. It is important that the data accurately represents the health and lifestyle of the dogs participating. Wherever possible, we seek to improve how questions are asked to make it easier for contributors to give us accurate data. This process is known as data validation, and it has been one of our main priorities since starting the study. Data validation ensures that the data collected is as accurate as possible so that any findings from the study will be true.
As part of her PhD, Carys Pugh visited a number of Dogslife participants and reviewed over 100 veterinary records from participating dogs. She found that members occasionally forgot when their dog had been ill but that when they remembered, the information they gave was extremely accurate, and completely consistent with the veterinary records. This is a really positive finding because it means we can investigate the risks for dogs developing illnesses in the cohort using owner definitions, without having to go back to their vets to confirm the diagnosis. Owner feedback also led us to change some questions in the questionnaire and we hope we have made it a little less onerous for owners to complete. Her work also suggested that the longer the time between the owner visiting the website, the less likely they were to remember an illness.
The validation work has been published as a series of guidelines for how to undertake validation. The full paper is available here: http://www.jmir.org/2015/4/e96/
When dogs reach their first and third birthdays, we will invite owners to upload four specific photos of their dog to allow us to assess their dog’s mouth, ear and body. We ask for these photos because they allow us to evaluate the usefulness of digital images for the assessment canine health, with reference to the information in the questionnaire owners have already kindly detailed. For example, we can assess tartar build up on the teeth, in relation to the diet fed, and the relative size of individuals in relation to the quantity of food fed. We hope to use the photos to see how dogs’ oral and aural health and bodily appearance changes over time. To find out what we’ve identified using the photos, take a look at February 2013’s newsletter article. For more information on how to take the photos and upload the images, please login and go to: www.dogslife.ac.uk/owner_upload/manage
Between June 2013 and June 2015, 137 Dogslife members attached an accelerometer to their dog’s collar for a week. The accelerometers measure exactly how much movement occurs every minute, and this is recorded as a number or “count”, which increases proportional to how much the accelerometer moves. We already know the threshold activity values in dogs (measured as “counts per minute”) which differentiate sedentary activity (such as sleeping), mild activity (such as walking) and moderate to vigorous activity (such as running or playing). The counts from one of the participating dogs are shown below and Monday seems to have been a particularly active day that week.
As part of her PhD, Carys worked with the data from the 137 dogs and tried to see how the accelerometer readings compared with answers to the Dogslife exercise questionnaire. On average, the dogs spent 85% of their time being sedentary, 11% of their time engaging in light activity and 4% being moderately to vigorously active each day. The dogs that spent the most time exercising off lead spent the least amount of time being sedentary and the most amount of time being vigorously active. Interestingly, fetching chasing and retrieving, and time spent on ‘other’ exercise, both seem to be positively associated with time spent on mild activity.
We really appreciate the efforts of everyone who put the accelerometer on their dog’s collar. We hope to publish the work so that we can describe the exercise of the cohort and start linking the exercise reports with health in the cohort.
In January 2013, the first detailed report from the Dogslife project was published in the scientific journal BMC Veterinary Research. The publication details how the study was set up, how participants were recruited, and some of the information participants kindly provided to the project in its first year. There is no directly comparable study to Dogslife, either of animals or man, where participants have been recruited to participate in an “on-line” study of their pet’s health, and the publication serves as a valuable guide to show others how to achieve the best results. Its importance is highlighted by the fact that over 1000 people accessed the paper to read in its first 30 days of publication, making it the second most ”read” paper in the journal over this period.
Given that Dogslife is the first internet-based longitudinal study of pet health, a number of important conclusions were drawn from the data collected in the project. Even simple questions, such as what is the most efficient way to recruit participants into the project, were not known before the study was developed. As you might expect, owners were most likely to participate in the project when we had permission from the Kennel Club to contact them directly by e-mail and/or post. As you can see from the figure below, after an initial flurry of visits to the website, the number of people joining the project was closely related to both the number of people registering their dog with the Kennel Club and the number of people visiting the website (Figure 1).
The majority of dogs registered with the study were living in households containing a family or an adult couple. When asked their reason for participating in the study, the majority of owners reported that they were doing so for altruistic reasons (to help with the research). Over 90% of members elected to receive the monthly Dogslife newsletter, and nearly 50% of members used the on-line scrapbook to store photos of their dogs.
Comparisons of the gender and coat colour of dogs enrolled in the project, and their recorded address (postcode) with that of all the dogs registered with the Kennel Club over the same period showed that recruited dogs are broadly the same as (and therefore representative of) the general Labrador Retriever population. This is important as we hope that any conclusions we draw from the project will be applied to the whole Labrador population in the future.
We estimated that approximately 80% of dogs participating in the project developed an illness in their first year of life (Figure 2), with the median age at which dogs were first reported to have developed an illness being just over 4 months. Approximately half of dogs were estimated to present to their vet for an illness in the same time period (Figure 3), with the median age at which dogs first presented to their vet being just over 10 months of age. We are currently in the process of analysing the veterinary records of part of the cohort, which will allow us to report much more accurately the frequency of different diseases recorded by the project.
You can read the full manuscript here if you wish. In line with the aims and ethos underpinning the study, we have paid to publish the results in an “open access” scientific journal which means that anybody can read our report if they wish without having to pay a fee or subscription. This is important, as it means we can disseminate the results of our project quickly and to everybody who wants to see them.
Thanks again to everyone who supports the project and helps us find out what keeps dogs healthy!