Diets of Arctic and Subarctic indigenous peoples have been the subject of growing interest because of concerns about the presence of organochlorine and heavy metal contaminants, as well as concerns over the health implications of a shift away from traditional food. Long-range atmospheric and oceanic transport and local sources of contamination have been described (Barrie et al. 1992, Lockhart et al. 1992, Muir et al. 1992, Shearer 1997, Thomas et al. 1992), and the role of traditional food as the main source of exposure has been emphasized (Kinloch et al. 1992, Kuhnlein et al. 1995b).

Traditional food is food, both plant and animal, harvested from the local environment, in comparison to market food, which is commercial food shipped from the south. Although traditional food is a source of contaminants, and apparently more so in the eastern than the western Canadian Arctic (Chan et al. 1995 and 1997, Kuhnlein et al. 1995b), it is also a source of important nutritional, social and cultural benefits (Kuhnlein 1995, Kuhnlein and Receveur 1996, Receveur and Kuhnlein 1997).

In the western Canadian Arctic, the territories of the Dene/Métis, food systems have been described in a few communities. Szathmary and co-workers (Ritenbaugh et al. 1996, Szathmary et al. 1987) investigated the relationships between dietary change and its potential effect on glucose metabolism among Dogrib Dene/Métis. Wein and colleagues described the food system of the Fort Smith Dene/Métis in terms of diet quality, with particular attention to the contribution of traditional food to the total diet (Wein et al. 1991a and 1991b). Our previous work focused on Sahtú Dene/Métis and characterized some of the benefits and risks of the current diet (Kuhnlein et al. 1991, 1994, 1995a and 1995b, Kuhnlein and Soueida 1992, Morrison et al. 1995).

Following these reports, The Dene Nation and Métis Nation of the Northwest Territories suggested that all Dene/Métis communities might benefit from studies of the relationships between diet and health and from assessments of type and quantities of traditional and market food being consumed. A participatory research procedure was therefore developed to 1) establish a baseline dietary intake that future dietary studies could use to assess change in food intake; 2) improve understanding of how food practices convey different benefits or risks with regard to nutrients and contaminants and also culturally and economically; and 3) identify food- and nutrition-related concerns and potential food and nutritional problems in the community.

Results of this study were reported to the communities through workshops, posters, radio interviews and a final report (Receveur et al. 1996). In this article we present methods used in this study and results of dietary assessment based on 24-h recalls. This analysis focused on describing nutrient intakes and identifying nutrients at risk of inadequate intake. We then evaluated the influence of traditional food on diet quality and examined the direction of dietary change by comparing intergenerational and between-community differences in dietary intake. Other reports will provide estimates of exposure to contaminants through traditional food consumption and consider the sociocultural and economic dimensions of the food system.

SUBJECTS AND METHODS

Participation in this study was extended to all 27 Dene/Métis communities except the communities of Dettah and Ndilo (both close to Yellowknife, the capital of the Northwest Territories), where similar projects were conducted by regional health services. Nine communities participated during March-April 1994 and again in October-November 1994 together with seven new communities (Fig. 1). The March-April interview period was selected to represent the time of the year with lowest traditional food consumption, and October-November was selected to reflect the peak of traditional food consumption.

Individual interviews were the primary instruments used in this study. The format of these was developed in consultation with members of the Dene Nation and the Métis Nation of the Northwest Territories, and the interviews incorporated comments and suggestions collected in 1993 during individual and community meetings in 15 Dene/Métis communities. The present project was approved by the McGill University Ethical Review Committee and a license was obtained from the Northwest Territories Science Institute. Each community developed a "research agreement" with the researchers (Scott and Receveur 1995).

After negotiation of the "research agreement" with community leaders, each community was informed through posters and radio messages of the intent of the project. A random sample of 10% of the Dene/Métis households or 25 households, whichever was larger, was drawn from existing band, housing and utility lists. One man and one woman from each household were invited to participate. When one man or one woman was not present in the household for interviewing, another person from an additional randomly selected household was contacted. In small communities (Tsiigehtchic, K'áhbamit'úé, Jean Marie River, Pehdzeh Ki) all households were invited to participate. Although a random sample of sufficient size is the best guarantee that results will be generalizable to the community at large, bias may still occur. For example, traditional food intake may be underestimated if the interviews are conducted at a time when a large number of high consumers of traditional food are out of the community (i.e., "on the land"). Care was taken to avoid interviewing at those particular times. Dietary interviews started in October, after the peak of the fall hunt, and again in March, the time of lowest expected traditional food use. Participation was confidential and voluntary; 91% of selected persons agreed to participate.

In each community, a project coordinator, generally a research-trained dietitian, worked with the local interviewer for the duration of data collection to provide guidance and ensure quality control. Interviewers were trained to administer interviews in English or the local language (there were four local languages in the study area), at the respondent's choice. Each interview lasted approximately 1 h and included the following:

To facilitate recalls, each interviewer was provided with an illustrated index of all traditional food species listed in the traditional food frequency questionnaire, three-dimensional portion models (bowls, cups and spoons locally available) and two-dimensional serving-size representations of bannock (a frequently consumed quick-bread usually of white flour) for use in the 24-h recall. Each record was checked in the field for completeness. Respondents reporting daily intakes of more than 17,672 kJ or less than 4418 kJ were contacted again to ensure the exactitude of the record: when in doubt, a second recall was completed and the first was discarded (this occurred for eight of the 1012 dietary records). Alcohol intake is not included in the data set. Alcohol consumption is prohibited in some communities, and it can therefore be expected that reports of alcohol consumption may not be comparable across communities.

Nutrient analyses were performed using two food composition databases: 1) a database of the composition of traditional Dene/Métis food derived from published reports (Appavoo et al. 1991, Kuhnlein et al. 1994, Morrison and Kuhnlein 1993) supplemented with traditional food previously analyzed from samples collected in the eastern Arctic (Kuhnlein and Soueida 1992, Kuhnlein et al. 1991); and 2) a market food composition database (Murphy and Gross 1987) derived from Agricultural Handbook No. 8 series adjusted to include Canadian-particular food items and nutrient fortification levels (Thompson and Brulé 1992). Dietary fiber and sucrose contents of traditional food were assumed to be zero for animal food and equivalent to those of commercial berries in the case of local berries, the only traditional plant food reported consumed during the seasons of interview. There were no missing values for market food. For traditional food, 110 out of the 1214 occurences reported in the 24-h recalls had missing food composition data. Forty of the 110 were for moose and caribou ribs, for which moose or caribou flesh was substituted. Other traditional food items with missing values included rarely eaten food parts such as moose nose, tongue or stomach, for which a substitute was used (moose flesh in those examples).

A total of 385 and 677 interviews were completed in late winter and fall, respectively. Records showing daily intake of coffee and/or tea exceeding 4 L (n = 32) and records from pregnant or lactating women (n = 18) were excluded from this analysis. Such levels of tea-coffee consumption were found to markedly affect energy and nutrient intakes, and these records represent a special subgroup to be analyzed separately rather than left in the data set where they disproportionally influence population estimates. Pregnant or lactating women were excluded because the number was insufficient for dietary evaluation. Table 1 presents the final data set by region and community. For each community, population size, latitude and road access are presented because these community-level variables can readily influence food availability. The proportional age distribution of our sample is presented in Table 2 and agrees well with the most recent census estimates (Northwest Territories Data Book 1990).

Table 1. Community characteristics and sample size in late winter (March-April) and fall (October-November) 1994 [View Table]

Table 2. Sample distribution by region, sex and age for participating communities [View Table]

Data were entered using Epi-Info, version 6 (USD Inc., Stone Mountain, GA). After extensive data checking, including double entry of a 10% random subset, data were analyzed with SAS, version 6.10 (SAS Institute, Cary, NC). Descriptive statistics included means or least square means with associated SEM. Least square means were used to adjust for unbalanced sample sizes across communities, age groups and seasons (Searles et al. 1980). In most cases, nutrient intakes did not meet the assumption of normality, and differences between groups were therefore tested by Kruskall-Wallis nonparametric ANOVA (Zar 1984). Macronutrient intakes, expressed as a percentage of total energy intake, were analyzed with parametric ANOVA. Post-hoc comparisons of group means were made with Tukey's multiple pairwise comparisons (Zar 1984). P < 0.05 was considered significant in all statistical tests.

RESULTS

Table 3. Average daily energy, nutrient, fiber and sucrose intakes in Dene and Métis communities, by season, sex and age1 [View Table]

Intakes were tested among age groups to assess generational differences and between men and women, adjusted for energy intake, to assess whether men and women tended to consume similar types of food. Differences between the sexes and among age groups were few in March-April and more pronounced in October-November, the period of peak traditional food consumption. A consistent trend across seasons was the higher carbohydrate intake in the youngest generation, which can be accounted for by greater sucrose intake. Another less significant trend was greater saturated fat intake in the youngest generation. Other differences between men and women and among age groups were less readily interpretable and may be in part obscured as a result of the different number of communities surveyed in each season (nine communities in March-April and 16 in October-November).

Table 4 further characterizes age, sex and season as sources of variation in dietary intake by comparing age and sex differences in food intake (classified in food groups) for the subset of nine communities surveyed in both seasons. Market food use showed no significant seasonal variation, whereas traditional food did for fish; the intake of land animals remained stable. Berries and bird consumption did not vary by season of interview because intakes during these seasons were minimal. Both bird hunting and berry picking (as well as other plant gathering) are activities limited to a short time window in spring and summer, respectively, two seasons not included in this data set. Besides season, sex and age effects were present as well as age × sex interactions. Adjusting for age, women compared with men tended to derive less of their dietary energy from market meat and more from grains with these differences becoming more marked in older generations. Although intakes of other food groups were similar for younger men and women, differences appeared with increasing age. Older generations consumed less of dairy products, fruits and vegetables, and mixed dishes and more traditional fish and land animals. Other age-related effects were peculiar to the oldest group and included higher grains and lower fat and sweet intakes among older women, compared with men of the same age. Table 4 also shows how market meats were substituted for traditional land animals among younger generations.

Table 4. Percent of total energy intake from market and traditional food groups, by sex and age for nine Dene and Métis communities interviewed in both seasons (n = 723)1 [View Table]

In addition to age, sex and seasonal variation, geographical variation is important (Fig. 2). In some communities, traditional food constitutes a greater percentage of total energy intake than in others. It appears that more northern communities derived a larger proportion of dietary energy from traditional food. Differences in traditional food intake among communities may be related to community characteristics such as population size, road access and availability of affordable market food, proximity to animal migration routes, and prevalent fishing and hunting practices. Of particular interest is how differences in traditional food use are likely to be reflected in differences in diet quality and nutrient intakes.

Table 5. Average daily food weight, energy, nutrient, fiber and sucrose intakes in Dene and Métis communities with high and low traditional food use (TFU), by sex and age for fall 19941,2 [View Table]

Whereas Table 5 provided a description of the nutrient implications of the observed variation in traditional and market food use across communities, Tables 6-8 illustrate the current extremes on the continuum of dietary change. Comparison of food records with market food only or with a mix of traditional and market food shows how, when market food only was consumed, the macronutrient profile shifted towards higher carbohydrate (and sucrose in particular for women), fat and saturated fat. Records without traditional food were lower in protein, iron, zinc, copper, magnesium and phosphorus, nutrients found in great quantity in wild meats. Food records without traditional food were also higher in sodium. Sodium values represent content in food and exclude the sodium added at the table. Whether traditional food was consumed or not, vitamin A and calcium intakes remained low, suggesting that traditional food sources rich in these nutrientssuch as caribou, moose or fish liver (vitamin A), soups or stews cooked with bones, and fish with small bones and skin (calcium)may not be consumed in sufficient amounts to make a difference. Similarly, dietary fiber intake remained low when market food only was consumed.

Table 6. Percentage of energy contributed by carbohydrate, protein and fat on days with and without traditional food intake in Dene and Métis communities1 [View Table]

Analyzing the intakes of additional nutrients for which food composition data exist for market food but not for traditional food suggests that vitamin D, vitamin E and folic acid intakes may also be of concern when market food only is consumed (Table 8). Although additional vitamin D and vitamin E may be provided by traditional food, folic acid intake is unlikely to improve when traditional food is consumed, unless liver is often used.

Table 8. Average micronutrient intake in Dene and Métis communities on days without traditional food for nutrients with only market food composition data1 [View Table]

DISCUSSION

Macronutrient profiles of the diets were similar to those reported for the Dogrib Dene/Métis (Ritenbaugh et al. 1995, Szathmary et al. 1987) and Sahtú Dene/Métis (Kuhnlein et al. 1995a). Wein and co-workers reported a lower contribution of protein and a higher dietary fiber intake in their study of Fort Smith residents (Wein et al. 1991a). These different dietary components suggest that the latter population did not consume as much traditional food as other Dene/Métis communities. In our study as well, the most southern community (Fort Smith) was identified as a low traditional food user compared with other Dene/Métis communities (Fig. 2).

Table 7. Average vitamin A, minerals and fiber intakes on days with and without traditional food intake in Dene and Métis communities1 [View Table]

Although previous work documented the importance of considering age, sex and food type (traditional vs. market) as sources of variation in dietary intake among Dene/Métis (Kuhnlein et al. 1995a, Morrison et al. 1995, Szathmary et al. 1987, Wein et al. 1991a), limited information has been reported on between-community variation in dietary intake (Morrison et al. 1995, Ritenbaugh et al. 1995). The study of variation in diet among age groups and among communities offers an opportunity to use cross-sectional data to characterize the process of dietary change. This methodology may be particularly appropriate to the study of indigenous peoples in rapid transition, because collection of time-trend data of the required magnitude to detect dietary change is often not feasible. For indigenous peoples who formerly relied entirely on locally harvested products, the percentage of market food now included in the diet can act as an indicator of change. Comparing nutrient intakes on days with or without traditional food further illustrates how nutrient intake can be maximally affected in the transition.

The need for a better understanding of the process of dietary change has, however, been emphasized in relation to ecological, economic and cultural factors responsible for loss of traditional systems around the world (Johns et al. 1994) and potential implications such as loss of culture-specific knowledge, increase in sedentary lifestyle and diet-related chronic health conditions (Kuhnlein and Receveur 1996).

For the Dene/Métis, this shift away from traditional food use is characterized by an increase in absolute energy intake and increases in the relative contributions of carbohydrate (particularly sucrose), fat and saturated fat. These changes alone are likely to have significant negative effects on the health of the population. Other implications for health include overall low intakes of calcium, vitamin A, folate and dietary fiber.

When patterns and consequences of dietary change for indigenous peoples were reviewed, Kuhnlein and Receveur (1996) concluded that negative nutritional status consequent to loss of a traditional food system can be countered under certain conditions such as adequate income, availability of good-quality market food, and education in the use of the new market food. It is our opinion that none of these conditions are met for the great majority of Dene/Métis.

Our work showed that the Dene/Métis traditional food system is still used extensively but that there is a trend for decreasing use in the younger generation and that some communities rely less on their traditional food system than others. This pattern of change calls for initiatives to document the current health status of this population and to prevent potential negative health consequences.

ACKNOWLEDGMENTS