Trends in weight change among Canadian adults: Evidence from the 1996/1997 to 2004/2005 National Population Health Survey

by Heather M. Orpana, Mark S. Tremblay and Philippe Finès

Introduction
Rate of gain slowing
Amount gained varies
Smaller percentage of Canadians gaining
Men and women who gain, gain more than before
Weight change a dynamic process
Cumulative effects
Concluding remarks
Acknowledgements
References

Data source
Analytical techniques
Definitions
Limitations

Introduction

Consistent with trends in other countries (Flegal et al 1998; Odgen et al. 2006), the prevalence of obesity has been rising in Canada. From 1978/1979 to 2004, the percentage of Canadian adults who were obese rose from 14% to 23% (Tjepkema 2006). The increasing prevalence of obesity is a major public health concern, as excess weight has been associated with type II diabetes, cardiovascular disease, psychosocial difficulties, osteoarthritis, and premature mortality (National Institutes of Health 1998).

Valuable as it is, cross-sectional information about the prevalence of obesity cannot provide information about patterns of weight change among individuals. Longitudinal data are needed for insight about patterns and rates of change that are behind the increase in obesity in Canada. For instance, a recent longitudinal study showed that almost a third of Canadians whose weight was in the acceptable range in 1994/1995 became overweight in the following eight years, and about a quarter of those who had been overweight became obese (Le Petit and Berthelot 2006). Only 2% of acceptable weight individuals became obese over the same period.

As an extension to this, an understanding of obesity requires information about the rate at which individuals are gaining (or losing) weight. Longitudinal studies of American adults have demonstrated that, in general, they gain weight up to 55 or 60 years of age, after which they start to lose weight (Williamson 1993; Sheehan et al. 2003; Truong and Sturm 2005). Few studies have examined rates of weight change in a representative sample of Canadians. Analyses of data from the 1981 Canada Fitness Survey and the 1988 follow-up, the Campbell Survey of the Well-being of Canadians demonstrated that body mass index (BMI) was relatively stable over the period between the surveys, but the researchers did not estimate the rate of change (Katzmarzyk et al. 1999). Another study based on the same data estimated that in families of at least two persons, the weight change from 1981 to 1988 was a gain of 2.9 kg for fathers and a gain of 3.5 kg for mothers (Hunt et al. 2002). However, both these studies examined only two points in time, and so could not determine whether rates of weight gain were changing or remaining stable.

The purpose of this analysis, which is based on longitudinal data from the National Population Health Survey (NPHS), is to examine two-year changes in the self-reported weight of the Canadian adult household population from 1996/1997 to 2004/2005, and to determine if the rate of change is accelerating, slowing down, or remaining the same (see Analytical techniques, Definitions and Limitations).

Top of Page

Rate of gain slowing

Canadians continue to gain weight, but indications are that the pace at which they are gaining has slowed down. The average self-reported weight of people aged 18 to 64 rose 0.96 kg for men and 0.86 kg for women over the two years from 1996/1997 to 1998/1999 (Chart 1). During the 2000/2001 to 2002/2003 interval, average gains were higher: 1.12 kg for men and 1.02 kg for women. Over the next two years (2002/2003 to 2004/2005), Canadians’ weight continued to rise, but the average amount gained was lower: 0.74 kg for men and 0.57 kg for women. Regression results (Table 1) indicate that this pattern of weight gain is statistically significant. Thus, overall Canadian adults were still gaining weight, but significantly less than in the earlier periods.

Top of Page

Amount gained varies

Changes in weight were significantly associated with sex, age group and body mass index (BMI) (Table 1).

Over the eight years from 1996/1997 to 2004/2005, the average self-reported weight of men and women in all age groups increased. However, in each two-year interval, younger people aged 18 to 33 experienced significantly greater average gains than did 34 to 49 year-olds; older adults aged 50 to 64 experienced significantly smaller gains (Chart 2 and Chart 3).

The general trend of a decline in the amount of weight gained in the 2002/2003 to 2004/2005 interval applied to men and women in most age groups. The exception was men aged 18 to 33 whose average weight gain in the last interval was greater than that in the previous one.

An individual’s body mass index was associated with how much his or her self-reported weight changed in each two-year interval (Chart 4 and Chart 5). On average, overweight people gained 0.8 kg less, and obese individuals, 1.9 kg less than did people whose weight was in the acceptable BMI range (Table 1). In fact, during most two-year intervals, people who were obese experienced a mean loss in self-reported weight.

The overall decline in the average amount of weight gained in the last two-year interval could reflect several processes: an increase in the number of people losing weight, an increase in the amount of weight lost, a decrease in the number of people gaining weight, a decrease in the amount of weight gained, or a combination of these factors. Further analyses were undertaken to examine how the slowdown in the amount of weight gained was related to these possibilities.

Top of Page

Smaller percentage of Canadians gaining

During each of the first three intervals, almost half of adults experienced a weight gain, but from 2002/2003 to 2004/2005, only 44% of men and 46% of women did so (Table 2). For men, but not women, this was a significantly lower proportion than in the previous intervals.

As well, 32% of men experienced a loss in weight from 2002/2003 to 2004/2005, a significantly higher percentage than in the first two intervals (28% and 27% respectively). Among women, the proportion losing weight did not differ significantly from one interval to another, ranging from 29% in the first three intervals to 32% in the last.

Top of Page

Men and women who gain, gain more than before

For the men who gained weight, the average amount gained rose over time from 4.56 kg in the first interval to 4.99 kg in the last, a statistically significant increasing trend (Table 2). The average gain among the women who gained weight varied, ranging from 4.50 kg to 4.78 kg. A statistically significant trend of an increasing amount of weight gained among women was also observed.

Among the men who lost weight, there was no statistically significant trend in the amount lost, with the average in the 4.42 kg to 4.68 kg range. By contrast, among the women who lost weight, the average loss rose significantly from 4.35 kg in the first interval to 4.91 kg in the last.

Thus, the overall decrease in the average change in weight in the last interval (2002/03 to 2004/05) appears to be driven by a combination of factors—a smaller proportion of men gaining weight and greater losses among the women who lose weight.

Top of Page

Weight change a dynamic process

An important consideration in examining trends in weight change is that the same people did not gain, lose or maintain their weight over all two-year intervals. For example, of the women who lost weight from 1996/1997 to 1998/1999, almost 64% gained weight over the subsequent interval from 1998/1999 to 2000/2001. Conversely, of the women who gained weight in the first interval, approximately 38% lost weight in the following interval, while approximately 39% gained weight. The pattern was similar for men and across subsequent intervals. Almost two-thirds of the people who lost weight in a given interval gained weight in the next.

Top of Page

Cumulative effects

The average two-year weight changes among adults from 1996/1997 to 2004/2005 were gains of 0.5 kg to 1 kg, and the overall change in weight during the entire eight years was a gain of 4.01 kg for men and a gain of 3.44 kg for women. While these amounts may appear relatively small, such changes are cumulative, resulting in a further shift of the distribution of an already predominantly overweight and obese population toward unhealthy weights (Tjepkema 2006). And even a small shift in the population distribution toward excess weight can have important consequences for the incidence of weight-related diseases (Rose 1985).

Top of Page

Concluding remarks

The results of this analysis describe the pattern of weight change among Canadian adults from 1996/1997 to 2004/2005, but they do not explain it. While it is known that poor nutrition and lack of physical activity are primary contributors to weight gain, alternative explanations such as environmental factors may be important and should be considered (Keith et al. 2006).

These data indicate that patterns of weight gain, loss and stability are dynamic, and warrant further research to identify the correlates and causes of the slowdown in the average amount of weight gained and the increase in the number of people losing weight. The increase in the amount of weight gained by those people who gained weight also deserves more investigation. Longitudinal analysis, which provides information about how individuals move from one weight category to another, is needed for public health strategies aimed at addressing the problem of obesity in Canada. Analysis of subsequent cycles of the NPHS will make it possible to determine if the decrease in the rate of weight gain continues in the future.

Although Canadian adults are still, on average, gaining weight, public health messages promoting healthy eating and physical activity have proliferated, and it is possible that without these interventions, the current rate of weight gain might have been higher.

Top of Page

Acknowledgements

Jean-Marie Berthelot generated the idea for this article. Georgia Roberts contributed invaluable statistical and methodological assistance, and Kathy White and Christel Le Petit provided helpful comments on successive drafts.

Top of Page

Data source

This analysis is based on data from five cycles (cycles 2 to 6) of the National Population Health Survey (NPHS), conducted by Statistics Canada from 1996/1997 to 2004/2005. Every two years since 1994/1995, the NPHS collects data about health status, health behaviours, and other determinants of health. This survey is representative of the household residents in all provinces in 1994/1995, excluding Territories, Indian reserves, Crown Lands, health institutions and members of the Canadian Forces bases and some remote areas in Ontario and Quebec. Although the NPHS also has an institutional component covering residents of health institutions such as nursing homes, that sample was not analysed in this article.

In 1994/1995, 20,095 household residents were selected to be members of the NPHS longitudinal panel. Of these, 86.0% agreed to participate, yielding a sample of 17,276. Response rates in subsequent cycles were 92.8% in 1996/1997; 88.3% in 1998/1999; 84.8% in 2000/2002; 80.5% in 2002/2003; and 77.4% in 2004/2005. More detailed descriptions of the NPHS design, sample and interview procedures are available in other papers and reports (Tambay and Catlin 1999).

Data were collected primarily through computer-assisted personal interviews in 1994/1995 and primarily through computer assisted-telephone interviews thereafter. To rule out the potential impact of collection mode on the results, only data from 1996/1997 to 2004/2005 were analysed. Telephone interviews comprised over 96% of all interviews in 1996/1997 and 1998/1999; over 98% in 2000/2001 and 2002/2003; and more than 99% in 2004/2005.

Top of Page

Analytical techniques

The sample for this analysis consisted of people who were aged 10 to 60 in 1994/1995. As they aged through 1996/1997 to 2004/2005, individuals were included in the analysis if they had reached age 18 by the beginning of a given interval, and were excluded if they were 65 or older at the end of a given interval. For instance, a respondent who was 62 in 1996 would have been included in the 1996/1997 to 1998/1999 interval, but excluded thereafter. Records for women who were pregnant at the beginning or the end of an interval were excluded for that interval.

For the first interval, 1996/1997 to 1998/1999, there were 9,387 respondents aged 18 to 64 at the beginning and end of the interval; 318 cases were missing data on weight; and 203 women were excluded because they were pregnant at the beginning or end of the interval. Thus, the sample size for the first interval was 8,866 cases. For the second, third and fourth intervals, there were 8,689, 8,098, and 7,517 respondents aged 18 to 64 at the beginning and end of the interval. Of these, 220, 160, and 139 cases, respectively, were missing data on weight, and 156, 152, and 139 women were excluded because of pregnancy. Additionally, for analyses using body mass index (BMI), cases with missing height were excluded: 7, 6, 12 and 8 cases for the first, second, third and fourth intervals, respectively.

Average weight change for the four two-year intervals from 1996/1997 to 2004/2005 was calculated for men and women. To reduce the effect of outliers, individuals who gained or lost more than 25 kg were given a value of 25 kg. For the first interval, 28 cases of weight loss and 23 cases of weight gain were truncated at 25 kg; for the second interval, 32 cases of weight loss and 31 cases of weight gain were truncated; for the third interval, 27 cases of weight loss and 42 cases of weight gain were truncated; and in the fourth interval, 30 cases each of weight loss and weight gain were truncated.

To determine whether the rate of change in body weight was increasing or decreasing over time, multiple linear regression was conducted using a person-period dataset, predicting two-year weight differences from time, time squared, sex, age group and BMI category at the beginning of the interval. Age group and BMI category were time-varying covariates. Records for individuals missing data on any variable for a given interval were excluded for that interval only. This accounted for less than 4% of records for any interval. Thus, for the regression model, the sample was 8,866 records for the 1996/1997 to 1998/1999 interval; 8,313 for the 1998/1999 to 2000/2001 interval; 7,786 for the 2000/2001 to 2002/2003 interval; and 7,239 records for the 2002/2003 to 2004/2005 interval.

To clarify patterns underlying observed differences in weight change, the proportion of people gaining weight, losing weight, or remaining stable (no change in self-reported weight), as well as mean weight gain among those who gained weight and mean weight loss among those who lost weight, were analysed for each two-year interval. Confidence intervals of the sex-specific proportions of respondents who gained weight, lost weight or remained stable were calculated, and compared to determine if they differed significantly over time. To test for the association between time and weight gain or weight loss, a linear regression was performed only on records where an individual experienced a weight gain or loss.

In order to take the complex survey design of the NPHS into account, the bootstrap method was used to generate confidence intervals of the estimates for all analyses (Rao et al. 1992; Rust et al. 1996; Yeo et al. 1999). Bootstrap weights for individuals were applied to each record for an individual (Fitzmaurice et al. 2004). Significance was set at p <0.05 (i.e., probability less than 0.05), and the weights for the longitudinal square file were used to weight the records to reflect the Canadian household population in 1994/1995. All analyses were conducted in SAS 9.

Top of Page

Definitions

Weight was self-reported and converted to the nearest kilogram for respondents answering in pounds. Similarly, height was self-reported and converted into metres for respondents answering in feet and inches.

Body mass index (BMI) was calculated by dividing weight in kilograms by height in metres squared. According to Health Canada guidelines (Health Canada 2003), individuals whose BMI was less than 18.5 kg/m² were considered underweight; those whose BMI ranged from 18.5 to 24.9 kg/m² were considered to be an acceptable (normal) weight; those whose BMI ranged from 25.0 to 29.9 kg/m², overweight; and those whose BMI was 30 kg/m² or more, obese.

Three adult age groups were identified: 18 to 33 years, 34 to 49 years, and 50 to 64 years.

Top of Page

Limitations

The data in these analyses were obtained primarily by computer-assisted telephone interviews and were self-or proxy-reported. Self-reported data may be affected by response biases such as social desirability. Self-reported weight is generally an underestimate of measured weight (Tjepkema 2006). If this reporting bias changed within individuals over time, it could affect the results. It is possible that increased media attention on obesity in recent years may have changed the magnitude of this bias. However, analyses using American data indicate no significant change in the magnitude of the bias associated with self-reporting weight and height during the periods 1988 to 1994 and 1999 to 2002 (Ezzati et al. 2006).

As in all surveys, non-response may introduce bias into the survey results. While the 1994/1995 longitudinal square weights adjust for non-response at the initial measurement, they do not adjust for subsequent non-response. Differential non-response may have affected the results. However, because regression using the person-period dataset does not require a respondent to answer at each cycle in order to include them in the analysis, this bias is somewhat attenuated. Future analyses should take non-response patterns into account, to investigate the possibility that selective attrition is affecting the results.

Because the data can be conceptualized as observations nested within individuals, a growth curve model would be an appropriate approach to analysing the data (Singer and Willett 2003). Initial analyses were conducted using a growth curve model in SAS; however, the estimates of SAS PROC MIXED have been reported to be biased when survey weights are used in the estimation (Asparouhov 2005), and the bootstrapping procedure was not available to estimate the variance of the growth curve model approach. Thus, an alternative approach using a person-period dataset was adopted. While it is less efficient than a growth curve model, it is unbiased and allowed for variance estimation using the bootstrap procedure.

You need to use the free Adobe Reader to view PDF documents. To view (open) these files, simply click on the link. To download (save) them, right-click on the link. Note that if you are using Internet Explorer or AOL, PDF documents sometimes do not open properly. See Troubleshooting PDFs. PDF documents may not be accessible by some devices. For more information, visit the Adobe website or contact us for assistance.