Association of dog ownership with accelerometer-measured physical activity and daily steps in 70-year-old individuals: a population-based cross-sectional study

Physical inactivity is a strong risk factor for non-communicable diseases and premature mortality [1], and has been associated with substantial economic burden [2]. More than one third of the population in developed countries has been estimated to be physically inactive [3], and the prevalence increases with age [4]. Because the world’s population is ageing rapidly [5], targeting physical inactivity in older people is of particular importance. Given the age-related decline in levels of moderate-to-vigorous-intensity physical activity (MVPA), recent evidence suggesting that light-intensity physical activity (LPA) is associated with health benefits [6‐9] is of particular interest. Similarly, favorable associations have been observed also between daily steps and health outcomes [10, 11], which is interesting given that walking is the most preferred type of physical activity [12], as well as because step count is an intuitive metric of total physical activity. Therefore, identifying factors associated with physical activity of any intensity in older adults would be important.

Encouraging dog walking has been proposed as a strategy for population-level promotion of physical activity [13], especially in developed countries where the prevalence of dog ownership (DO) is high. Roughly 40% of households in the US [14] and Australia [15] own a dog, and 15% of households in Sweden [16]. Even though reviews have highlighted an association between DO and higher levels of walking and physical activity [13, 17], most studies have so far estimated walking and physical activity through measures of self-report [13, 17], which may suffer from bias due to recall and social desirability. This leads to an overestimation of physical activity. In contrast, device-based measures provide more accurate estimates and are better for assessment of incidental physical activity such as LPA, as well as step count [18, 19]. However, the current evidence on DO and device-measured physical activity of different intensities is limited. One study found that DO was associated with higher levels of device-measured LPA but not MVPA [20], although the study included only middle-aged individuals. Another study found that DO was associated with 2760 more steps per day [steps/d] in older adults, but only 86 participants were included [21]. Finally, a larger study found that DO was associated with up to 22% higher levels of device-measured total physical activity in older adults [22]. However, specific intensities were not investigated, and available covariates were few and self-reported [22]. For example, although adjustments were made for self-reported health and education, there were no data on marital status, income, or objectively measured health such as preexisting disease, body weight or physical function. These factors may influence either DO, physical activity, or both [23‐25].

Collectively, the association between DO and device-measured physical activity of different intensities in older adults has been insufficiently investigated. Apart from a small number of studies, limitations include lack of studies conducted in population-based samples, lack of investigation of multiple intensities of physical activity, and lack of adjustment for objectively assessed covariates which could influence DO and physical activity. In this cross-sectional study, we investigated the associations of DO with accelerometer-measured LPA, MVPA and steps/d in a population-based sample of 70-year-old women and men, while adjusting for a variety of both subjectively and objectively assessed sociodemographic and health-related factors.

In this population-based study of community-dwelling 70-year-old individuals, DO was associated with higher levels of daily LPA, MVPA, and steps, and with higher odds of meeting physical activity recommendations. Exploratory interaction analyses suggested that the associations of DO with steps and MVPA was stronger among female participants, but none of the associations differed based on BMI or time of year for examination.

One of the main results of this study was the favorable association of DO with LPA. We found that dog owners performed 19 more min/d of LPA compared to those not owning a dog. To our knowledge, there has been no previous study of the association between DO and LPA in older adults. However, our findings extend those from a study on DO and accelerometer-measured physical activity in Japanese middle-aged individuals, showing a similar estimate [20]. Emerging evidence suggest that LPA may potentially have implications for health in older adults. In a cohort study, an increase in LPA by 30 min/d was associated with 13% lower risk all-cause mortality in older men [32]. Other studies in older adults have observed beneficial associations between LPA and subjective well-being [33, 34], metabolic syndrome [26], mobility disability [35], and cardiovascular disease [8, 9].

Another key result was that dog owners took roughly 1700 more steps/d compared to those not owning a dog. This is about 1000 steps/d lower than what was found in a previous longitudinal case-controlled study in older adults [21]. However, that study was likely not conducted in a representative sample, because it included a very small number of volunteers. Our study sample may better represent a general population. When the estimates in the present study were expressed in relative terms, dog owners took 24.4% more steps/d compared to those not owning a dog. This finding is supported by the results from a population-based study from the UK, which showed that DO was associated with up to 22% higher levels of accelerometer-measured total physical activity in older adults [22]. Moreover, we found that the association between DO and steps/d was stronger in females compared to in males. We are not aware of any similar results reported in the literature, and a previous systematic review found mixed evidence regarding whether sex correlates with dog walking [24]. Therefore, we may only speculate about potential explanations for this result. One explanation could be perceived feelings of neighborhood safety, which has been shown to correlate with dog walking among females [24]. However, it should be noted that our interaction analyses were exploratory, hence these findings are preliminary and warrants further investigation. Regardless, there is evidence to suggest that such an increment in daily steps observed in both male and female may potentially be clinically relevant. For instance, cohort studies including both older men and women showed that every 500 steps/d and every 1000 steps/d was associated with about 10% lower risk of cardiovascular disease [36] and about 15% lower risk of mortality [32, 37]. Another cohort study found beneficial associations also with incident falls [38]. Findings related to step volume may also have practical implications because it represents an intuitive metric that is easy to understand for the general population, and because steps can be easily monitored using devices that have become increasingly common.

Interestingly, we also found that dog owners accumulated 11 more minutes/d of MVPA. In a previous study from Japan, DO was associated with about 7 more minutes/d of accelerometer-measured MVPA in middle-aged adults [20]. However, that estimate was not statistically significant, perhaps in part due to the smaller study sample. Another possible explanation for the different result in that study compared to our study could be demographic factors. The study from Japan included individuals in working age who have more limited time for leisure time physical activity such as MVPA. In contrast, our study population comprised older individuals above retirement age who would have more time to recreationally walk their dog. Another factor that could contribute to the beneficial association in the present study may have to do with environmental factors, which have been shown to correlate with dog walking [24]. The municipality of Umeå, wherein the present study was conducted, is characterized by a dog-supportive physical environment, including walking areas in proximity and high access to public spaces and a variety of green- and outdoor areas. Importantly, measures of the physical environment have been proposed as key [24] when it comes to promoting dog walking in the community and in turn increase population levels of physical activity. Altogether, the favorable associations found in the present study between DO and higher levels of physical of all intensities therefore suggest that efforts to promote dog walking in the community may help increase population levels of physical activity in older adults. However, given the lack of similar studies conducted in older adults, further population-based studies are warranted, including also older adults of a wider age span and from other regions.

This study has some important strengths, including the relatively large population-based sample with a high participation rate (84% of all invited) and a high adherence to wearing the accelerometer (92.4%). Together, this reduces selection bias and increases the possibility to generalize the findings to other cohorts of community-dwelling 70-year-old individuals. We were also able to adjust our analyses for several objectively assessed covariates, either from clinical examinations or though national registries with nationwide coverage, which minimizes information bias. Also, the assessment of physical activity using accelerometers is a notable strength. In a previous study based on the same population, self-reported physical activity was largely overestimated as compared with accelerometer-measured physical activity [39].

Some limitations should also be noted. First, the cross-sectional design prevents causal inferences from being drawn, although there is some prospective data supporting a link between DO, dog acquisition and increased physical activity [40, 41]. Moreover, although we adjusted our analyses for several covariates, there is the potential of residual and unmeasured confounding. For example, data on dog characteristics was lacking, which may be thought to potentially influence the associations. However, a systematic review found mixed evidence for an association between dog characteristics such as dog breed, and dog walking [24]. Also, because socioeconomic data were obtained from age 65 for all participants, t socioeconomic factors may have changed between age 65 and study participation at age 70. For example, if some participants retired during this period this would affect annual income. However, we deem the risk of bias due to changes in annual income to be low given that official data show that about 80% of people in Sweden born during 1947–1949 (the same year as participants in the present study) had retired by the age of 65 [42]. Another factor to consider is that the accelerometer wear protocol was changed during the study, which may have introduced bias. However, a sensitivity analysis excluding participants wearing the accelerometer during nighttime confirmed the results of the primary analysis. Next, the accelerometer cut-points used to classify physical activity intensities were based on absolute intensity and originally validated in younger people [27]. In older adults, this could potentially result in misclassification between LPA and MVPA. However, by using these cut-points we have previously shown associations of LPA with cardiovascular disease and mortality [8]. Also, given the relatively high mean physical activity in this population as compared with other populations of similar age [33, 43], these cut-points appear to be plausible for use in the present population. Regarding step count, the accelerometer used in the present study may slightly underestimate the total number of steps, yet it correlates strongly with other commercial step count devices [44].

To summarize, this cross-sectional study conducted in a population-based sample of 70-year-old individuals showed that DO was associated with higher levels of daily LPA, MVPA, and steps. This adds knowledge regarding how DO may play a beneficial role in promoting physical activity in the older population. The findings could support the development and evaluation of targeted interventions seeking to promote dog-friendly environments and facilitate dog walking in the community.