The main findings of this study were that participants with four or more exercise facilities within 1,000-meter road network buffer zones surrounding their residences spent more time in objectively assessed moderate to vigorous physical activity, and were more likely to meet the physical activity recommendations, compared to participants with no exercise facilities within their buffer zones. This association was independent of sex, age, income, marital status and time of year.
Our findings are in accordance with the results of a previous study, which showed a significant association between objectively assessed density of exercise facilities within circular buffer zones and self-reported frequency of exercise [18
]. Another study from the U.S. that investigated the association between density of exercise facilities within circular buffer zones of different sizes and a range of self-reported physical activities [13
] reported similar results, although the association for the smallest buffer zones (radius 0.5 miles/805 meters) was not statistically significant. In contrast, a Spanish study found no association between numbers of exercise facilities per 10,000 inhabitants and self-reported physical activity [17
]. That study measured, however, the availability of exercise facilities at the province level, and the large geographic areas used may explain the lack of association. A further study from the U.S. found no association between objectively assessed availability of exercise facilities and leisure time physical activity, as assessed using the International Physical Activity Questionnaire [34
]. That study was based on relatively small circular buffer zones (radius 400 meters) and a dichotomized measure of availability of exercise facilities (yes/no).
In contrast to some previous findings [13
], none of the socio-demographic variables included in this study (sex, age, income or marital status) modified the association between availability of exercise facilities and physical activity. In a Swedish urban setting, where differences in socioeconomic status may be less pronounced than in, for example, the U.S., individuals with different incomes seem to benefit to the same extent from exercise facilities.
Several studies have reported seasonal differences in physical activity, with higher levels of physical activity during spring and summer and a decline in activity during the colder months [20
]. A review of the effect of season on physical activity from 2007 concluded that availability of exercise facilities could increase the opportunities to be physically active all year round in cold and wet climates [23
]. However, we found no significant interaction between time of year and availability of exercise facilities in any of our analyses, suggesting that availability of exercise facilities is of equal importance for physical activity throughout the year.
The present study has some limitations that should be considered. It is a cross-sectional study and causality cannot therefore be determined. In addition, there may be unmeasured confounders for which we did not control for in the present study (i.e., residual confounding may exist). We cannot exclude the possibility that gyms and other exercise facilities may be established in neighborhoods where physically active people live, or that people who like to exercise move to neighborhoods with good availability of exercise facilities. This, together with the fact that our sample was recruited from a large urban region, may to some extent affect the generalizability of our results. It is also important to recognize that the physical activity recommendations are based on evidence from studies of self-reported physical activity and health outcomes. It is possible that misclassification occurred when assessing by accelerometry whether the physical activity recommendations were met. Another limitation is that we only measured the availability of exercise facilities around participants’ residences and not around their workplaces or their route to and from work, where they may spend a considerable amount of time [35
]. Accelerometers may also underestimate the intensity of some physical activities performed at exercise facilities (e.g. resistance training, spinning and swimming) due to lack of mid-bodily movement and the device being non-water resistant. Compared to another population-based Swedish sample [4
], our sample spent more time in moderate to vigorous physical activity (median time 42 versus 31
min/day). The other study was conducted in 2001 and its sample also included rural participants. In contrast, our sample was exclusively urban and was recruited in the capital of Sweden. However, our non-response analysis showed small or no differences in socio-demographic factors between participants and non-participants, which means that any selection bias was most likely non-differential.
The present study also has several strengths. We were able to use detailed road network data including not only roads, but also cycle paths and footpaths. There were considerable differences when visually comparing the road network alone and the road network combined with cycle paths and footpaths. The use of these detailed network data to produce line-based buffer zones around participants’ residences likely gave a good picture of the areas that are actually accessible to participants. By using objective data on availability of exercise facilities we were able to exclude the possibility of same-source bias (i.e., physically active persons reporting a higher availability of exercise facilities compared to their less active counterparts). Furthermore, accelerometers, unlike self-report, do not suffer from bias due to social desirability and recall problems [37
], although it is possible that accelerometers may create some reactivity to wearing the device. However, any such bias is most likely non-differential, i.e., equal in all types of neighborhoods.
The association between availability of exercise facilities and physical activity that was identified in this study could be explained by a number of possible mechanisms. Having a large number of exercise facilities near one’s home may increase the chance of finding a mode of exercise that is attractive in terms of type of activity, cost and social atmosphere. This may explain why participants with ≥4 exercise facilities within their buffer zones were more physically active compared to those with no facilities, while participants with 1-3 facilities were not. The mere presence of exercise facilities could, by putting physical activity in the minds of passers-by, also increase the overall levels of physical activity and not just exercise performed at these facilities. In agreement with this hypothesis, Sallis et al. showed that the presence of exercise facilities close to the individuals’ homes did not seem to be associated with participation in the specific activities offered at those facilities, but rather with an increased overall exercise frequency [18