The authors have declared that no competing interests exist.
There has been increased interest in human-animal interactions and their possible effects on human health. Some of this research has focused on human physical activity levels, mediated through increased dog walking. Much of the reported research has been cross sectional, and very few epidemiological studies have examined the association between dog ownership and mortality in populations.
We used data from the Norwegian county population-based Nord-Trøndelag HUNT Study (HUNT2, 1995–1997). Cox proportional hazards models were fitted to analyse the relationship between dog ownership and all-cause mortality. The median follow-up time was 18.5 years and the maximum follow-up time was 19.7 years.
In this population, dog owners were no more physically active than non-dog owners, both groups reporting a total of just over 3 hours/week of light and vigorous activity. Dog owners (n = 25,031, with 1,587 deaths during follow-up; 504,017 person-years of time at risk) had virtually the same hazard of dying as non-dog owners (Hazard ratio 1.00, 95% CI 0.91–1.09).
We found no evidence for an association between the presence of a dog in the household and all-cause mortality or physical activity levels in this Norwegian population. Further epidemiological research is needed to clarify this relationship, as methodological limitations and an active Norwegian population sample means that generalizable evidence is not yet clear on dog ownership and mortality.
There has been an increase in research into human-animal interaction, and its possible effect on human health. A recent trend in this literature has focused on the potential benefits of dog ownership on increased human physical activity levels, mediated through dog walking. The American Heart Association has released a statement indicating the need for more data on the influence of pet ownership on the presence and reduction of cardiovascular disease risk factors and cardiovascular disease risk [
A rapid systematic review identified the study types in this field [see Text
A rapid systematic review was undertaken to profile papers in the published literature in this field to broadly classify papers on dogs and physical activity. The Scopus data base was used from 2000 to the end of 2016, and all papers with the Title terms (dog* AND (physical activity) or walk*)) were included. Two raters (SK and AB) considered all abstracts and coded them as [
The potential health benefits of dog ownership have been described for several decades, and thought to result from dog-human interactions, companionship and interaction, especially among older adults [
In summary, most of the evidence for health effects comes from cross-sectional studies, with associations noted between mental health indicators or blood pressure levels and pet (dog) ownership [
There were no epidemiological studies of dog ownership and mortality directly [See
The present paper explores the association between dog ownership and all-cause mortality in a Norwegian population-based cohort. We hypothesised that if adults have a dog, and walk that dog in the long term, the increased physical activity levels and possibly reduced stress levels may influence mortality outcomes.
We used data from the second wave of the Nord-Trøndelag HUNT Study [
The selection of study participants is shown in
The Nord-Trøndelag HUNT Study HUNT2 (1995–1997).
Weight and height were measured at the time of participation in the HUNT2 survey, and BMI was calculated on the basis of these two measurements. All the other covariates were based on self-report. The highest level of education completed was measured using the categories “primary school”, “secondary school graduate” and “college/university graduate”. Marital state was measured using the following categories: “unmarried”, “married”, “widow, widower” and “divorced or separated”. Daily smoking was measured as a dichotomous variable (current smoker or non-smoker). We used the Hospital Anxiety and Depression Scale (HADS) to measure symptoms of anxiety and depression [
The HUNT Study was approved by the Norwegian Data Inspectorate and the Regional Committee for Medical Research Ethics (REK Central). All participants in HUNT2 provided written informed consent. The present study was approved by the Regional Committee for Medical and Health Research Ethics (REC Central) (2015/1480).
We used Cox proportional hazards models to analyse the relationship between dog ownership and survival. The start of follow-up was participation in HUNT2, and the end of follow-up was April 2015. The median follow-up time was 18.5 years and the maximum follow-up time was 19.7 years. The HUNT database is regularly updated through the National Registry on dates of death and emigration of HUNT participants. We used the date of death as the endpoint. To protect the anonymity of study participants, all study participants who died in a certain month were registered as dying on the 15th of that month. There were 12,698 deaths (23.8% of the study population) during follow-up. In total 208 study participants (0.4%) emigrated during follow-up, and they were censored at the time of emigration. We used attained age as the time scale, and adjusted for sex. We used the log-rank test and log-minus-log plots to test the proportional hazards assumption on the model.
We estimated the hazard ratio of death in dog owners (n = 10,668) compared to the reference group of people who did not own a dog (n = 42,750). We considered age and sex to be confounders in the relationship between dog ownership and mortality. There was no missing data for sex and age. We adjusted for sex as a categorical variable. We used date of birth as the time scale in the Cox proportional hazards model, which may lead to a more effective control of age than including age as a variable in the model.[
To explore the relationship between dog ownership and mortality, we adjusted for confounders as well as possible mediators in three models. However, with the exception of sex and age, a considerable part of the cohort had missing data on one or more of the covariates used in the models. In order to perform all these three analyses on the same sample, we identified a sub-cohort consisting of “complete cases” only (n = 28,746), who did not have missing on any of the covariates.
Model 1 was adjusted for sex and age only, as previously described in the total cohort analysis section.
Model 2 was adjusted for the variable in Model 1 in addition to education and marital status, which we considered to be possible confounders in the relationship between dog ownership and mortality.
In Model 3, we adjusted for the variables in Model 2, as well as a set of variables related to health and lifestyle factors. These variables were: High levels of symptoms of anxiety and depression, BMI, smoking and physical activity. Several of the variables in Model 3 may be mediators in the relationship between dog ownership and mortality. Low intensity and high intensity physical activity per week in the last year were entered into the model as two separate categorical variables. Daily smoking was entered as a dichotomous variable (smoker/non-smoker). As there was evidence that the relation between weight and mortality was not linear, BMI was entered as a categorical variable with five categories (<18.5, 18.5–24.9, 25–29.9, 30–34.9, and ≥35). A Kaplan-Meier plot suggested that there might be a very slight difference in survival in younger versus older dog owners. We therefore stratified by age (below and above the age of 65) and repeated the analyses on the complete cases cohort. We also stratified by sex. Analyses were performed using Stata 13.1.
General characteristics of the study population are showed in
Dog owners | Not dog owners | ||||||||
---|---|---|---|---|---|---|---|---|---|
n | % | Mean | SD | n | % | Mean | SD | ||
N | 10,668 | 42,750 | |||||||
Age | 10,688 | 46.7 | 14.1 | 42,750 | 51.2 | 17.7 | |||
Females | 5,633 | 52.8 | 23,306 | 54.5 | |||||
Deaths during follow-up | 1,586 | 14.9 | 11,112 | 26.0 | |||||
Emigration during follow-up | 45 | 0.4 | 161 | 0.4 | |||||
Marital status | 2,294 | 21.5 | 10,193 | 23.9 | |||||
7,226 | 67.9 | 25,485 | 59.8 | ||||||
697 | 6.5 | 2,876 | 6.7 | ||||||
432 | 4.1 | 4,096 | 9.6 | ||||||
Education | 3,454 | 33.4 | 14,848 | 36.6 | |||||
4,870 | 47.1 | 17,224 | 42.4 | ||||||
2,016 | 19.5 | 8,524 | 21.0 | ||||||
Self-reported health "poor" or "not so good" |
2,625 | 24.8 | 11,735 | 27.7 | |||||
Low quality of life |
306 | 2.91 | 1032 | 2.46 | |||||
Mean anxiety score (HADS-A) | 9,272 | 4.3 | 3.3 | 35,468 | 4.1 | 3.2 | |||
Mean depression score (HADS-D) | 9,922 | 3.5 | 3.1 | 38,850 | 3.4 | 3.0 | |||
Daily smoker | 3,270 | 32.1 | 10,738 | 26.5 | |||||
Possible alcohol problem |
748 | 8.4 | 2,553 | 8.5 | |||||
Hours of low intensity physical activity per week in the last year | 9,206 | 2.1 | 0.9 | 35,888 | 2.0 | 0.9 | |||
Hours of vigorous physical activity per week in the last year | 7,789 | 1.2 | 1.1 | 29,972 | 1.1 | 1.0 | |||
BMI | 10,638 | 26.4 | 4.1 | 42,455 | 26.3 | 4.1 |
The Nord-Trøndelag Health Study (HUNT2), 1995–97.
a As reported on the stand-alone question “How is your health at the moment?”. The other possible response categories were “good” and “very good”.
b Reporting that they were "somewhat dissatisfied", "dissatisfied" or "very dissatisfied" with life on a stand-alone question on quality of life (“Thinking about your life at the moment, would you say that you by and large are satisfied with life, or are you mostly dissatisfied?”)
c CAGE (Cut down, Annoyed, Guilty, Eye-opener) score > = 2 (Ewing JA. Detecting alcoholism. The CAGE questionnaire. JAMA. 1984 12;252(14):1905–7)
We estimated the hazard ratio (HR) of death in the total sample (n = 53,418). Using people who did not own a dog as the reference group, dog owners had a hazard ratio (HR) for all-cause mortality of 0.98 (95% confidence interval (CI): 0.93–1.03), adjusted for sex and age. There were 12,698 deaths during follow-up, and the person-time at risk was 892,985 person-years.
The hazard ratios adjusted for possible confounders are shown in
Model 1 | Model 2 | Model 3 | |||||
---|---|---|---|---|---|---|---|
HR | 95% CI | HR | 95% CI | HR | 95% CI | ||
1 | · | 1 | · | 1 | · | ||
Others (reference) | |||||||
Dog in the household | 0.99 | 0.91–1.09 | 1 | 0.91–1.09 | 1 | 0.91–1.09 | |
1 | · | 1 | · | 1 | · | ||
Others (reference) | |||||||
Dog in the household | 0.99 | 0.88–1.12 | 1 | 0.89–1.13 | 1 | 0.88–1.12 | |
1 | · | 1 | · | 1 | · | ||
Others (reference) | |||||||
Dog in the household | 1.01 | 0.89–1.15 | 1.01 | 0.89–1.15 | 1.02 | 0.89–1.16 | |
Others (reference) | 1 | · | 1 | · | 1 | · | |
Dog in the household | 0.96 | 0.86–1.08 | 0.96 | 0.86–1.07 | 0.98 | 0.88–1.09 | |
1 | · | 1 | · | 1 | · | ||
Others (reference) | |||||||
Dog in the household | 1.04 | 0.90–1.21 | 1.05 | 0.91–1.22 | 1.02 | 0.88–1.18 |
The Nord-Trøndelag Health Study (HUNT2, 1995–97). Follow-up from the time of participation in HUNT2 until April 2015.Model 1: Adjusted for sex and age. Model 2: Model 1 + adjusted for education and marital statusModel 3: Adjusting for all above variables and anxiety and depression scores, BMI, physical activity levels and smoking.
Results of survival analyses stratified by age (>65 years and ≤65 years) are shown in the middle panel of
Results of analyses stratified by sex are shown in the lower panel of
While dog ownership is widely believed to confer health benefits, there is a paucity of long-term prospective studies. Our study is one of the few epidemiological studies that examined the long-term outcomes of dog ownership with mortality risk in a large population-based cohort. We found that living in a house with a dog was not associated with mortality, and this finding persisted in sex and age sub-group analyses. Further, participants in this study living in households with dogs did not appear to be more physically active than participants who did not.
Although the epidemiological literature on dog presence and mortality is limited, findings are consistent. Our results are in agreement with US studies that utilized NHANES data that failed to show any significant association between dog ownership and mortality risk [
There are at least two possible explanations for the consistent, but limited association between dog ownership and mortality. Owning a dog may have health benefits [
Regardless of the likely explanation of the null findings, it is possible that the limited literature on this question is a sign of publication bias, i.e. other researchers have carried out similar analyses and because they have also found null associations, the data were never published. This also indicates the limitations of the plethora of cross-sectional evidence, which provides an optimistic potential for dog ownership and health, that needs to be confirmed by repeated longitudinal epidemiological evidence.
The findings in this study were somewhat surprising, given the reasonably consistent evidence from systematic reviews that dog owners are more physically active than non-dog owners [
Further, characteristics of the HUNT cohort may have contributed to the results we found in our material. The Nord-Trøndelag region is mostly rural, and a sizeable proportion of household dogs may have been “working dogs” used for hunting or dog sledding, not as family pets promoting regular dog walking. Moreover, the study population is characterized by a relative high level of physical activity, so dog ownership may not have any additional effect on the level of activity. By contrast, dog ownership may increase physical activity in cities and urban areas where opportunities for other outdoor activities are more limited.
This study of dog ownership and mortality has the longest follow-up to date although this could be seen as both a strength and a weakness due to relatively short lifespan of dogs and the absence of information on subsequent dog ownership, as discussed above. Our analyses were statistically powerful and utilized a large cohort with a high response rate and over 12,000 events. We used an objective registry data to measure the endpoint, and the only source of loss to follow-up was emigration, which was very low at 0.4%. Like the remaining studies in the field [
We found no evidence for an association between the presence of a dog in the household and all-cause mortality or physical activity in this large regional cohort from Norway. The methodological limitations of our study are endemic in this field of research and as such our work can neither confirm nor refute that dog ownership is beneficial for mortality risk, especially not in urban areas. This question needs to be answered in future studies that, ideally, are designed for the purpose and collect additional information on dog ownership, dog walking and dog survival. Until we have such data, dog ownership can be promoted only as a modality that may improve quality of life and aspects of health other than reduced mortality risk.
The HUNT study is a collaboration between HUNT Research Center [Faculty of Medicine, Norwegian University of Science and Technology (NTNU)], NordTrøndelag County Council, Central Norway Health Authority, and the Norwegian Institute of Public Health.