The purpose of this study was to examine and compare the dietary habits and lifestyle behaviors of self-defined vegetarians and non-vegetarians from a population-based representative sample of BC adults. Approximately 6% of the sample, weighted to reflect the BC population, reported being vegetarian. The findings of this study suggest that the dietary habits, lifestyle behaviors, and food-choice motivations of self-defined vegetarians differ from those of non-vegetarians, and that there may be variation between men and women which has not previously been examined in population-based studies.
Several aspects of our results warrant additional consideration, one of which is the small proportion of self-identified vegetarians who adhered rigidly to diets free from animal flesh. Occasional use of seafood, poultry, or meat by a majority of those who consider themselves to be vegetarian has also been reported in other studies [9, 15]. If a strict definition of vegetarianism had been used, the prevalence in our study would be less than 1.5% rather than close to 6%. Despite basing our analysis on respondents' self-definition, we still observed a number of differences in nutrient intake and lifestyle behavior. At some level, this validates respondents' self-identification as vegetarian.
Evidence for a higher level of 'health consciousness' among vegetarians in our sample was provided by findings of increased use of nutrient supplements, higher intakes of several nutrients (fiber, magnesium, potassium), higher intakes of fruits and vegetables, a considerably lower prevalence of smoking, and among women, higher physical activity and a lower BMI. Many of these findings have been reported in other studies, although most reports from convenience samples have not found differences in smoking or exercise behavior by vegetarian status [7, 16–21]. It is likely that convenience sampling resulted in recruitment of more 'health conscious' participants and therefore did not detect differences. Thus our findings provide population-level support for the concept that vegetarians have healthier lifestyle practices than the general population of non-vegetarians.
Vegetarians were also more likely to consider 'maintaining/improving health' when choosing/avoiding foods, to choose foods for the nutrients they contain and to avoid foods for their fat content. These findings provide additional evidence of health consciousness, and are consistent with research reporting that health concerns and benefits are a primary reason for adopting a vegetarian lifestyle [22, 23], although we did not assess motivation for adopting a vegetarian diet. They are also consistent with a population-based study in the Netherlands that found vegetarians were more likely to report health considerations when purchasing food . That study, however, did not report nutrient intakes.
A novel aspect of our analysis was that, in addition to assessing differences in nutrient intakes, we also compared the prevalence of inadequate nutrient intakes using the EAR cut-point method . As assessed by the proportions with total usual nutrient intakes below the EAR, vegetarians were less likely to have an inadequate intake of magnesium, and female vegetarians were also less likely to have inadequate intakes of folate, vitamin C, thiamin and vitamin B6. Although there were no differences by vegetarian status in the proportions with zinc intakes below the EAR, this may not be an accurate reflection of zinc adequacy, as the requirement for dietary zinc may be as much as 50% greater for vegetarians . Similarly, iron requirements of vegetarians are estimated to be 80% greater than those of non-vegetarians . However, the adequacy of iron intakes was not assessed in our study because the iron requirement distribution is skewed, and therefore the EAR cut-point method cannot be used to estimate the prevalence of inadequacy . Finally, although adequacy of vitamin B12 intakes is often identified as a concern for vegetarians, in our sample the prevalence of inadequate intakes was similar by vegetarian status. This is likely due to the fact that almost all vegetarians used dairy products and eggs, as well as to the high prevalence of B vitamin supplementation among vegetarians.
Although our vegetarian sample was small, our results provide suggestive evidence of gender differences. For example, vegetarian women had a lower age-adjusted BMI and waist circumference, and a lower prevalence of overweight/obesity, while no differences were seen between vegetarian and non-vegetarian men. This may have been due to the higher frequency of physical activity reported by vegetarian women (but not men), as energy intake did not differ by vegetarian status for either sex. Reports from convenience samples often suggest that vegetarians have lower BMI and/or a lower rate of obesity [2, 7, 22, 25–27]. Conversely, other convenience samples, in which energy intakes and physical activity were similar between vegetarians and non-vegetarians, did not detect differences in BMI between groups [16–19, 28, 29]. In the population-based CSFII, self-identified vegetarians had lower energy intakes and age-adjusted BMI . However, a major limitation of that report was that analyses were not conducted by gender. Accordingly, if vegetarians were more likely to be female, as observed in our sample and another population-based sample , vegetarians' mean energy intake and BMI would appear to be lower because of women's lower mean energy intakes and BMI.
The distribution of macronutrient intakes also provided suggestive evidence of gender differences. Carbohydrate as a percentage of energy was higher among vegetarians, as was also found in the CSFII vegetarian analysis  and the majority of convenience sample studies [18, 22, 26, 27, 30, 31]. Other studies have also reported lower percentage energy from fat [8, 9, 22, 27, 32] and protein [8, 18, 22, 27, 28, 30–32]. In our sample, only male vegetarians had a lower proportion of energy from protein and only female vegetarians consumed less energy from fat.
We also observed gender differences in motivations for choosing/avoiding foods. Only male vegetarians were more likely to report considering heart disease and high blood pressure when choosing/avoiding foods and to report avoiding foods because of their cholesterol or saturated fat content. This is consistent with the higher prevalence of heart disease among the male vegetarians in our sample, who we speculate may have chosen to follow a vegetarian diet as a result of heart disease. Because we did not assess motivation for adopting a vegetarian diet, this cannot be ascertained, and in any case, the study's cross-sectional design precludes causal inferences. Female vegetarians, on the other hand, were not more concerned about heart disease, but were more likely to consider cancer, osteoporosis and food allergies/intolerances when choosing/avoiding foods and to avoid foods because of their salt content. They were also less likely to consider weight gain when choosing/avoiding foods. It has been suggested that some young women may adopt a vegetarian lifestyle in an effort to lose weight [33, 34]; however, this does not appear to be true for our population-based sample.
While our findings suggest that variation by gender may exist in vegetarians' dietary habits and lifestyle behaviors, the study limitations should be acknowledged. First, although the sample was considered representative of the province of British Columbia, it was not nationally representative, which means that inferences cannot be made about the Canadian population. Also, the response rate, although typical of other studies of this kind, was not optimal. Second, the absolute number of self-identified vegetarians was small and therefore caution must be used when interpreting the apparent gender differences. We had limited power to detect gender-by-vegetarian status interactions. Finally, data on dietary intake and lifestyle behaviors were based on self-reports, and it is known that dietary intakes are underreported . This would be problematic if differences existed in the extent of underreporting by vegetarian status. However, based on similar reported energy intakes of the two groups, it appears unlikely that differential underreporting occurred.
We do not believe that our observations of higher 'health consciousness' among vegetarians were confounded by other differences between vegetarian and non-vegetarian groups. First, although the prevalence of vegetarianism was higher among women than men, we conducted analyses separately by gender. Second, vegetarians tended to be younger than non-vegetarians, so age was included as a covariate in nutrient intake and anthropometric analyses. Third, although vegetarians were more likely to be single and to report low-income status, consideration of these differences did not affect our observations.