Association between Physical Activity and Body Weight: Health Creation and Disease Prevention

Leisure time physical activity has a broad spectrum of health benefits. The objective of this study was to provide evidence and to support the association between physical activity and body weight. We used data from a probability sample of 8,128 individuals residing in Montreal (Canada). Multinomial multivariate logistic regression was used to determine the association between physical activity and body weight. Results showed no statistical association between physical activity and underweight. However, physical activity was associated with overweight, namely individuals doing less physical activity were significantly more likely to be overweight (OR=1.42; 95%CI=1.18- 1.69) than those performing intense physical activity. Additionally, the association between physical activity and obesity showed a negative incremental relationship, i.e. individuals reporting moderate physical activity were 29% more likely to be obese (OR=1.29; 95%CI=1.07- 1.55), and those reporting low physical activity were 123% more likely to be obese (OR=2.23; 95%CI=1.80-2.76) compared to intense physical activity status. Leisure time physical activity and overweight/obesity were significantly associated after accounting for several socioeconomic, lifestyle and health-related correlates. Due to the broader positive impact of physical activity on the health status of the individual, we advise health authorities to facilitate the propagation of healthy lifestyle in the community by adopting a health-creation policy in addition to the conventional disease-prevention strategy.

Keywords: Overweight; Obesity; Physical Activity; Cross Sectional Study; Logistic Regression

Body-weight, defined by body mass index (BMI), is an important health indicator. Measurements outside normal body-weight range are associated with ill-health outcomes. For instance, mental health and psychological well-being are correlated with underweight, while overweight and obesity are associated with diabetes, cardiovascular disease, kidney diseases and cancer [1-3]. According to the Canadian Community Health Survey, 63.9% (57.9% females and 69.9% males) of Canadian population, ages 18 years and older, are either underweight or overweight/obese [4]. Population based microsimulation models predict that 66% of the adult population in Canada will either be overweight or obese by year 2030 [5]. According to the 2012 edition of Montreal’s Local Health Survey Program (TOPO), 52.6% of Montreal adult population are either below or above normal body-weight range. This prevalence seems to be the second highest reported behavioral health problem in Montreal (Appendix).

Although interpersonal variability of body weight may have a genetic component, nonetheless, there are several other factors that may influence the actual body-weight of an individual [6-8]. Some of these factors are readily modifiable (e.g. education, income, physical activity, healthy eating, smoking, stress management, medical care) and some are not (e.g. sex, age, ethnicity, immigration status, chronic health conditions).

Amongst several factors associated with body weight, leisure time physical activity seems to be an effective, non-invasive, inexpensive and affordable means to manage body weight. Several studies have shown a strong correlation between obesity and physical inactivity [9-14]. However, to our knowledge, no previous study has reported the simultaneous association of gender- specific body weight and physical activity in Montreal population. The objective of this study is to show the independent association between physical activity and body weight, controlling for several socioeconomic, lifestyle and health-related variables.

This study included 8,128 subjects (3,751 men and 4,377 non-pregnant women) between ages 18 and 65 years and living in private households in the city of Montreal, Canada. Sample selection was realized by multistage stratified random probability sampling. Data of this study were extracted from the 2012 edition of Montreal’s Local Health Survey Program (TOPO). TOPO data included information on chronic diseases and their major determinants such as: social conditions (employment, education, immigration, material deprivation), lifestyle (smoking, physical activity, fruit and vegetable intake), and use of health care services. The study was approved by the Quebec Provincial Public Health Ethics Committee. Data were collected between February and November 2012 under the supervision of Le Secteur Surveillance de l’État de Santé à Montréal (Health Surveillance in Montreal, Canada). The response rate (41.4%) was computed by using the standard definition of the American Association for Public Opinion Research [15,16]. Detailed information on the survey process is available elsewhere [17,18].

The main dependent variable was the body-weight. Measurements of body-weight were based on the body mass index (BMI). BMI was computed as the ratio of body-weight (in kilograms) to body-height (in meters square). Both weight and height were self-reported. In this study, the adult body-weight was categorised into four groups according to the Canadian guidelines for body weight classification: underweight (BMI<18.5 kg/m2), normal weight (18.5 kg/m2 ≤BMI<25.0 kg/m2), overweight (25 kg/ m2≤BMI<30 kg/m2), and obese (BMI≥30 kg/m2) [19].

The main independent variable was the leisure time physical activity. The latter was measured by the short version of the International Physical Activity Questionnaire (IPAQ) [20]. Leisure time physical activity data referred to the time spent being physically active during the past 7 days. Responses were grouped into low, moderate and high categories based on the Metabolic Equivalent Task minutes per week (MET-min/wk) [20]. All other socioeconomic, lifestyle and health-related variables were used to account for their independent effect on the physical activity. These variables were: sex, age, education (less than secondary/ secondary/university), language spoken at home (French only/English only/other), residency status (immigrant/non-immigrant), household income ($20,000 intervals), material deprivation index (a measure of socioeconomic conditions at the neighborhood level grouped into quintiles), smoking status (current smoker/past smoker/non-smoker), daily intake of fruits and vegetables (less than five servings/five servings and more), alcohol consumption (non-excessive/excessive, i.e. five or more drinks (six for men) on the same occasion at least 12 times in the last year), stress in life (yes/no), having a family physician (yes/no), and presence of physical or mental problems (the respondent’s self-rated overall health as well as illnesses diagnosed by a health professional) [21]. Diagnosed physical illnesses were asthma, fibromyalgia, arthritis, back pain, high blood pressure, chronic bronchitis, emphysema or chronic obstructive pulmonary disease, diabetes, heart disease or cardiac problems, and cancer. Diagnosed mental health illnesses were mood disorder (depression, bipolar disorder, or dysthymia) and anxiety disorder (phobia, obsessive-compulsive disorder, or panic disorder).

The initial analyses were stratified by sex. However, since results did not differ by sex except for annual household income variable, we created a sex-by-household income interaction variable and combined men and women data to gain power. First, we tabulated the frequency distribution of socioeconomic, lifestyle and health-related characteristics of the study population. Second, we used the Rao-Scott χ2 test to analyze bivariate associations between explanatory and outcome variables; significance was set at P < 0.05. For the multivariable analysis, we used multinomial logistic regression because our outcome variable had four nominal values. We compared each of the underweight, overweight and obese categories to the normal-weight group. Accordingly, we reported adjusted odds ratio (OR) estimates and their 95% confidence intervals (CIs); significance was affirmed if the 95% CI of an OR estimate did not include unity. All analyses were performed using SAS, version 9.1.3 (SAS Institute, Inc.).

Table 1 show that our sample has a rather homogeneous sex and age distribution. On the other hand, more than half of the sample report speaking only French at home and having university education. Additionally, more than 60% are born in Canada and have annual household income equal to or higher than CAN$40,000. Table 1 also provides the prevalence of some lifestyle habits in the population such as low physical activity (21%); currently smoking (22%); insufficient intake of fruits and vegetables (60%); and excessive use of alcohol (17%), as well as some health related variables such as overweight/obesity (48%), high stress in life (33%), physical and mental health problems (58% and 86% respectively) and access to family physician (59%). All these study variables are significantly associated with body weight except for language spoken at home and stress in life (Table 2).

In the multivariate analysis, physical activity was not statistically associated with underweight (Table 3, Column A). Nonetheless, older age (OR=0.72; 95%CI=0.62-0.83), having a family physician (OR=0.69; 95%CI=0.48-0.98), and males with high annual household income (OR=0.23; 95%CI=0.10-0.55) respectively showed a protective effect, while suffering from physical health problems was a risk factor (OR=1.48; 95%CI=1.04-2.12).

Physical activity was statistically associated with overweight (Table 3, Column B). Namely, those with low physical activity were significantly more likely to be overweight (OR=1.42; 95%CI=1.18-1.69) than those with intense physical activity. This association was adjusted for all other variables including daily fruit and vegetable intake. Furthermore, males with high annual household income were significantly more likely to be overweight than their low-income counterparts (OR=1.60; 95%CI=1.25-2.05). However, there was no significant association between body-weight and household income for female population.

The association between physical activity and obesity (Table 3, Column C) was rather interesting because it showed a negative incremental relationship, i.e. the lower the physical activity, the higher the odd of being obese. Hence, individuals reporting moderate physical activity were 29% more likely to be obese (OR=1.29; 95%CI=1.07-1.55), while those reporting low physical activity were 123% more likely to be obese (OR=2.23; 95%CI=1.80-2.76) compared to their intense physical activity counterparts and controlling for all other relevant variables in the model. The association of some correlates were similar in both overweight and obese models, such as, age, education, material deprivation index, present of physical health problem and sex-specific annual household income. However, other correlates were model specific, for instance, recent immigrants were 36% less likely to be obese (OR=0.64; 95%CI=0.48-0.84) and non-excessive alcohol consumers were 38% more likely to be obese. It is worthy of note the association of smoking status with obesity, namely, current smokers were 28% less likely to be obese (OR=0.72; 95%CI=0.57-0.90), while past smokers were 30% more likely to be obese (OR=1.30; 95%CI=1.06-1.59) compared to non-smokers.

We studied the independent association between physical activity and body weight among the adult Canadian population in Montreal city. Results of our multivariable regression analysis, concurrently comparing underweight, overweight and obese groups with normal-weight adults, revealed that physical activity is significantly associated with overweight and obesity after controlling for relevant socioeconomic, lifestyle and health-related variables. Moreover, this association revealed a “dose-response” relationship; namely, the odds of being overweight or obese were significantly increased with reduced leisure time physical activity. Our findings confirm similar associations between body weight and physical activity reported in the literature [11,12,22,23].

Being overweight or obesity is an important risk for many chronic diseases and premature death [3,24]. This study provided evidence that leisure time physical activity is a significant protective factor to reduce body weight independent of the recommended intake of fruits and vegetables. Interestingly, leisure time physical activity is an inexpensive, non-invasive, effective and easy to implement method to manage body weight with no known side effects. Moreover, the benefits of physical activity are observed over a large spectrum of diseases and disability, as well as within different groups of the population such as the elderly, the adolescents, adults, and pregnant women [3,25-36]. In fact, beyond its impact on body weight, physical activity has also been reported to improve physical appearance and enhances mood and self-esteem [37]. In terms of prevention, Lee et al. (2017) suggest that increasing physical activity could yield billions of dollars in saving that overweigh the cost of intervention [38]. Given the large spectrum of its benefits and the ease of its implementation, shouldn’t daily practice of physical activity constitute a vital recommendation of every health professional? For example, Institute of Medicine suggests a 60min/day of moderate-intensity activity to prevent overweight or obesity [39]. Our study supports the latter argument and advocates the propagation of leisure time physical activity as an effective means towards health creation in addition to disease prevention.

The predominant health problems of the adult population in Montreal are related to behavioral and lifestyle habits (Appendix). In another words, most prevalent diseases are preventable in this population. Hence, wouldn’t it be more effective if public health departments explore and encourage health creation, in addition to disease prevention, to attain an optimum population health status? We think providing the population with better opportunities for leisure time physical activity is an excellent avenue towards an ultimate attainment of healthy body, mind and spirit. Indeed, some studies confirm the positive impact of the neighborhood walkability/design-for-biking interventions on the reduction of overweight and obesity [40,41]. Hence, we urge local governments to design neighborhoods that facilitate the propagation of healthy lifestyle in the community.

Our research showed an independent and statistically significant negative association between the practice of leisure time physical activity and overweight/obesity after accounting for several other socioeconomic, lifestyle and health-related correlates. Furthermore, due to the broader positive impact of physical activity on the health status of the individual, this study urged health authorities to facilitate the propagation of healthy lifestyle in the community by adopting a health-creation policy besides the conventional disease-prevention strategy.

This study has some limitations inherent to its design. As a population survey, the data were based on self-reported information which may introduce misclassification bias should the information be wrongly reported. For instance, it is suggested that the self- reported measurement of body-weight be regarded as underestimate [42]. Moreover, the low response rate could lead to selection bias should a particular characteristic of the sample be misrepresented. However, neither of these limitations presented a threat to the quality of our data, because distributions of several sociodemographic variables were comparable to those of the census data. Additionally, the cross-sectional nature of the study design does not let us infer any causal relationship between body-weight and physical activity variables. Finally, this study did not consider genetic and sedentary lifestyle determinants of body-weight because data were not available.

We are grateful to the local and regional advisors for their participation in the planning stages of this study. Funding for this study was provided by Direction régionale de santé publique du Centre intégré universitaire de santé et de services sociaux du Centre- Sud-de-l’Île-de-Montréal.