Adult obesity in Manitoba: prevalence, associations, and outcomes, 3rd edition

Published: October 1, 2011
Category: Bibliography > Reports
Authors: Bailly A, Chateau D, Fransoo R, Martens P, McDougall C, McGowan KL, Prior H, Schultz J, Soodeen RA
Countries: Canada
Language: null
Types: Population Health
Settings: Government

Winnipeg, MB, Canada: Manitoba Centre for Health Policy.

Manitoba Centre for Health Policy, Winnipeg, MB, Canada

Obesity is a major public health concern in Canada and may continue to be for some time because it is influenced by a large number of factors, many of which are not easy to change. Indeed, some factors are not modifiable at all (e.g., age), and many of those that are potentially modifiable will require coordinated, long–term strategies to address. That said, this study and others show that there are factors that can be changed and have a significant impact (e.g., increased physical activity), so the ‘obesity epidemic’ should not be seen as inevitable or irreversible.

This report examines a number of aspects of obesity and its relationship with health status, health service use, and mortality. It focuses on adults (aged 18 and older) in Manitoba. While many of the results reflect findings reported in other studies, a number of important new findings have emerged.

In Manitoba, Aboriginal peoples (First Nations, Metis, and Inuit) comprise a significant proportion of the population. Unfortunately, most of the survey data used in this report excluded persons living in First Nations communities. Aboriginal residents living in all other areas of the province were included in the surveys and in this study, though their results are not reported separately. Other research (discussed in this report) has shown that the prevalence of obesity is higher in these groups, so the results in this report under–estimate the prevalence of obesity in Manitoba, especially in areas where a large proportion of residents live in First Nations communities (e.g., Burntwood).

Like many other studies, we assessed obesity using Body Mass Index (BMI) values, which are based on a person’s height and weight as collected in a number of national and provincial surveys. BMI values are an imperfect indicator of body size and composition, but remain the only indicator available for a large, representative sample of Manitobans. BMI values were calculated from direct measurements of height and weight whenever available, though that was the minority of cases. For most people, only self–reported values were available, so we ‘corrected’ the BMI values using formulae derived from a Statistics Canada study designed for this purpose. This correction provided more accurate values and ensured that we drew valid conclusions from our analyses, but it also means that the obesity prevalence values in this report cannot be directly compared to other reports using ‘uncorrected’ values (which show considerably lower prevalence of obesity).

Two key characteristics distinguish this report from others: first, it provides detailed results for Manitoba including analyses at the Regional Health Authority (RHA) and sub–RHA levels; and second, it brings together data on BMI levels with data on the rates of use of a number of key health services. These are done using the unique Population Health Research Data Repository housed at the Manitoba Centre for Health Policy (MCHP).

Like Canada overall, Manitoba has experienced a significant increase in the prevalence of obesity over time; Manitoba values have consistently been higher than national averages. This study is based on Manitoba–specific data from 1989 through 2008. Over that period, ‘corrected’ adult obesity increased from 18.4% to 28.3% among males and from 16.6% to 25.9% among females. Interestingly, however, the increase in obesity prevalence over time appears to have stopped for females, who reached 25% in 2000 and then remained stable through 2008.

Geographic analyses showed that within Manitoba, the highest obesity prevalence values are in Northern areas and the lowest are in urban areas, though increases over time were noted in all areas. However, overall obesity levels were not strongly related to population health status across RHAs. It should be noted that the data available for this analysis could not be used to validly estimate obesity prevalence for Aboriginal residents separately, though other studies have shown that their obesity levels are higher than for other Manitobans.

Our analysis of the relationship between obesity and the various risk and protective factors (23 variables plus interaction terms) revealed that socio–demographic factors (age, sex, education, and others) were the most closely related to obesity levels. The influence of age was particularly strong, with obesity prevalence being low in young adults, higher in middle–age adults, and low among older adults. Where people lived, their marital status, and employment status were also significantly associated with obesity; whereas household income was only weakly associated and frequent consumption of fruits and vegetables was not significantly related to obesity levels. All of these findings reflect the impact of each variable while controlling for all other variables in the model.

We also found significant associations with leisure time activity levels and the number of hours spent in sedentary activities. These are important findings, as they support investment in interventions to improve those factors among the entire population. This pair of findings – that both leisure time activity levels and hours spent in sedentary activities were significant – is interesting. It means that both are independently related to obesity, so interventions on both factors should be considered. That is, initiatives to decrease the number of hours Manitobans spend in sedentary activities may be beneficial for all, including among those who are already active in their leisure time. The results also suggested that those who reported more hours of sleep were less likely to be obese, though this relationship did not reach statistical significance. Finally, while this analysis included many individual characteristics and risk and protective factors, their combined influence explained only a small proportion of the total variation in obesity, reminding us that many other factors are also important in understanding obesity. Hereditary effects and food consumption were likely the most important influences for which detailed data were not available for this study.

Relationships between obesity and a number of chronic diseases have been shown in many previous studies; similar analyses in this study largely mirrored those findings, though some findings did not reach statistical significance. The strongest and likely most important associations found here were between obesity and the incidence and prevalence of both hypertension and diabetes, which were dramatically higher among those with higher BMI levels. These are particularly important indicators because their impact often has a domino effect: hypertension and diabetes both cause a substantial burden of morbidity (illness) and mortality directly. They are also related to the development of other serious health problems, most notably heart disease and stroke, which are leading causes of death. We also analysed the incidence of the most common types of cancer, but found no significant relationships, likely owing to the relatively small sample size available. These findings for hypertension, diabetes, and cancer held for both sexes; other diseases showed some differences among BMI groups by sex.

We also investigated the relationship between obesity and the use of health care services including physician visits, hospital use, prescription drugs, home care, and personal care homes. This section, which capitalizes on the uniquely powerful health data system (the Repository) housed at MCHP, provides the most important contributions from this study. Overall, the results revealed that while the Obese group almost always had the highest rates of health service use, the differences between it and the Normal and Overweight groups were relatively small. That is, the health care system is not being overwhelmed by the demand for health services related to obesity. This finding is particularly important because no previous studies have been able to provide this kind of analysis on a large representative sample with such comprehensive data on health service use.

Furthermore, for a number of indicators, the higher rates were only evident for those at particularly high BMI values. For example, the Obese group had more physician visits per year than others, but only about 15% more overall; moreover, the rise in rates only occurred above a BMI of 32 for females and 35 for males. Prescription drug costs were highest above a BMI of 35 for females and above 37 for males. Hospitalization rates were higher for the Obese group in both sexes, but only at BMIs of 33 or higher.

Causal modelling of health service use rates indicated that illness level was by far the strongest predictor of health service use, followed by sex, and then other factors including BMI, age, and socioeconomic status.

The ‘reasons for’ physician visits and inpatient hospitalizations were spread over many causes, though the visit category, which includes diabetes, was more prominent among the Obese group. Also, an interesting trend emerged to suggest that the Obese group used hospital services more frequently for conditions beyond the top 10 causes of hospitalization.

The final chapter is dedicated to the analysis of mortality (death rates). Long–term follow–up analyses show the highest mortality rates are among the Obese group, followed by the Overweight group, and finally the Normal group. However, further analysis revealed these effects to be indirect, as BMI group was not a significant predictor of mortality when age, sex, and other variables were also accounted for in the analysis. These findings suggest that obesity may not be a direct cause of mortality, but remains important because it is related to the development of a number of diseases, which are in turn related to mortality. The story may be different for the Overweight group, as our results and those from a number of other recent studies show that they face no higher mortality risk than the Normal group; indeed, some studies show the Overweight group is at lower risk of death than the Normal group.

Taken together, the results from several chapters in this report and other studies suggest a re–examination of our understanding of the concept of ‘Overweight’ and the risks with which it may be associated. Many of the findings from this study and others show that outcomes for the Overweight group are similar to, or even better than, those for the Normal group; so being overweight may not carry the level of risk previously thought. However, the likely ‘catch’ is the connection with age: since most adults gain weight over time (at least until about age 60), being in the Overweight group at a young age may mean a higher likelihood of reaching the Obese level at some point. And the results of this study and others clearly show that the Obese group experiences significantly higher morbidity and mortality.

Despite the vast body of work done to date, significant further research is needed to answer the many remaining questions. Ideally, future studies should use a longitudinal design and incorporate direct effects of BMI on health and mortality as well as indirect effects via related chronic diseases/events. Longitudinal analyses may also reveal that optimal health outcomes might be related to systematic changes in BMI level over the life course. And ideally, detailed food consumption data should also be included.

Age,Population Markers,Mortality Prediction,Morbidity Prediction,Canada

Please log in/register to access.

Log in/Register

LinkedIn Facebook Twitter

© The Johns Hopkins University, The Johns Hopkins Hospital, and Johns Hopkins Health System.
All rights reserved. Terms of Use Privacy Statement

Back to top