Article Text
PDF
Abstract
Background People in Arctic Canada are undergoing a nutritional transition and increased prevalence of chronic disease. The Healthy Foods North diet and physical activity intervention was developed in 2007–2008 while working with populations in six communities in Nunavut and the Northwest Territories, Canada.
Methods Four communities received the 1-year intervention (eg, conducting workshops, cooking classes and walking clubs) and two communities served as controls. Among the 263 adult evaluation participants, food frequency questionnaires were used to assess dietary intake at baseline and postintervention. Changes in mean nutrient intakes, nutrient density and dietary adequacy from baseline to postintervention were determined. The intervention impact on nutrient intakes was assessed through multivariate linear regression analysis.
Results Post-intervention assessment showed a reductions in total fat, saturated, monounsaturated and polyunsaturated fatty acids, and increases in iron intake, only in the intervention group. More than a 4%-increase in the percentage of adherence to vitamins A and D recommendations was observed in the intervention group. After adjusting the regression models, respondents in the intervention communities significantly reduced their energy intake and increased their vitamins A and D intake.
Conclusions The Healthy Foods North is an effective programme to improve dietary quality among populations of the Canadian Arctic. Long-term interventions are expected to be important factors in the prevention of diet-related chronic diseases in these communities.
- NUTRITION
- Epidemiological methods
- DIET
Statistics from Altmetric.com
Introduction
Increasing rates of chronic diseases among Aboriginal populations in the Canadian Arctic may be attributable to the ongoing nutritional transition lowering their diet quality.1 ,2 ,3 Prior to contact with other societies, commonly consumed foods included locally harvested nutritionally rich products such as land and marine mammals, fish, birds and plants. Greater exposure to non-Aboriginal dietary practices resulted in dramatic dietary transitions, with a study conducted between 2007 and 2008 among Inuit and Inuvialuit populations revealing that people spent twice as much money on non-nutrient-dense foods compared with on traditional foods.4 The role of nutrient rich diets in the prevention of chronic disease has been well-documented, with the WHO identifying nutrition as a modifiable determinant for disease prevention.5
The growing prevalence of chronic diseases combined with the reduced life-expectancy experienced by Aboriginal populations, particularly in the Northern territories, has contributed to increased costs for the national healthcare system.6 ,7 The total health expenditure per capita in 2010 for Nunavut (NU) and the Northwest Territories (NT) were $11 811 and $9906, respectively, compared to $5452 on average for Canada.8 Hence, chronic disease prevention programmes emphasising positive changes in diet may be beneficial for Northern Canada's Aboriginal communities.9 ,10
The Healthy Foods North (HFN) programme was a culturally appropriate, community-based intervention developed for Inuit and Inuvialuit communities in NU and NT. HFN sought to improve diet through increased intake of essential micronutrients and decreased consumption of non-nutrient dense foods that contribute to high energy and fat intake. Baseline dietary assessments conducted for the programme found widespread inadequacies of micronutrient intakes and increased total energy intakes among adults.11 ,12 These observations were attributed to a high consumption of micronutrient poor, high-energy dense store-bought foods.13 ,14
HFN's objectives were to: (1) increase/maintain consumption of traditional and nutrient-dense foods, (2) promote food preparation methods without adding fat, (3) decrease processed food consumption and (4) increase physical activity.15 The objectives of the present study were to determine differences between baseline and postintervention intake of promoted nutrients (eg, vitamins A and D) and depromoted nutrients (eg, fat and sugar) and evaluate the impact of the HFN programme among Inuit and Inuvialuit communities.
Methods
The HFN programme was evaluated through a quasi-experimental study design. The programme was conducted in six communities in NU (n=3) and NT (n=3). Selected communities included diverse proportions of Inuit and Inuvialuit populations of varying socioeconomic backgrounds.16 The intervention was conducted in one semiremote and one remote community in NT between May 2008 and August 2009, and two remote NU communities between October 2008 and November 2009. One remote community each in NU and NT served as controls.
Based on themes identified during community participatory research,17 the HFN intervention included five phases: (I) 6 months of informative research to understand local concepts and cultural norms regarding healthy foods and food practices; (II) two 2-day workshops with stakeholders to identify foods and behaviours for intervention; (III) intervention refinement over 18 months to review evidence and finalise a list of foods and behaviours for intervention; (IV) 3-day training for trainees, local community health representatives, project coordinators and local store staff and (V) programme implementation and evaluation over 14 months. Some of the activities of the programme included healthy breakfasts, meal planning and cooking, sufficient intake of vitamins and minerals, pedometer challenges and walking clubs. Implementation sites included food stores, health clinics, offices and special community events. Promotional messages were communicated through local community media.15
Data collection has previously been described elsewhere.18 In brief, data collection was administered at baseline and 1 year following intervention.
Using community housing maps, households were recruited following random selection of household numbers. This method ensured sampling from areas with varied proximities to food stores and hunting areas. One adult (≥19 years), typically the main food shopper or preparer, was selected to participate from each household, provided they had lived in the community for at least 6 months. Pregnant and breastfeeding women were excluded owing to the differences in their nutritional requirements and possible changes in energy expenditure. Three attempts were made before selecting a new household to substitute for a non-responding household. Written informed consent was obtained from all participants. Baseline response rates ranged between 69–93% and 65–85% in the NU and NT communities, respectively. Individuals selected at baseline from both groups were also interviewed at the end of the intervention. Participants were remunerated with gift cards.
Baseline and postintervention data collection was carried out by community health workers, community members and university students, who were trained, evaluated and certified by the principal investigator (SS). Participants’ height and weight were measured in triplicate.18 Weight was adjusted between 1 and 2 kg for light to heavy clothing and recorded to the nearest 0.1 kg using a digital scale. Height was recorded to the nearest centimetre using a stadiometer. When participants declined measurements, self-reported measurements were noted.
Culturally appropriate quantitative food frequency questionnaires (QFFQ) previously developed19 ,20 and validated21 ,22 for the populations were used to assess dietary intake. The questionnaire collected information on the typical frequency of food and beverage consumption over the past 30-day period classified into eight categories ranging from ‘never’ to ‘2 or more times per day’. Three-dimensional food models (NASCO, Fort Atkinson, Wisconsin, USA), packages of commonly consumed store-bought foods, standard units and local household utensils were used to assist participants’ estimate portion sizes.18 Participants’ socioeconomic status was assessed using a questionnaire collecting information on sociodemographic variables and material style of life (MSL), an additive scale of ownership of 20 items in working condition (Cronbach's α=0.83).23 All interviewers were trained by the principal investigator (SS) in questionnaire administration and anthropometric measurements to ensure standardisation. For participants whose primary language was not English, either an interviewer fluent in the local language (Inuktitut or Inuinnaqtun) or an interpreter was used to conduct the survey. Interviews were carried out at participants’ homes and the majority were conducted in English.
Ethics statement
Institutional Review Board approval was obtained from the Committee on Human Studies at the University of Hawaii and the Office of Human Research Ethics at the University of North Carolina at Chapel Hill. Additionally, the Ethics Committee of the Beaufort Delta Health and Social Services Authority approved this project. Aurora Research Institute in the NT and the Nunavut Research Institute in NU provided research licences.
Statistical analysis
Data from all six communities were combined for analysis. Baseline differences in the communities’ demographic and socioeconomic variables by intervention assignment were analysed using a Student t test for continuous normally distributed variables and a χ2 test and a one-way analysis of variance for categorical variables.
To determine total daily nutrient intake, data from three datasets (food composition table, QFFQ and food items portion weights) were analysed by the Food Frequency Questionnaire Analysis Program in STATA (StataCorp LP, College Station, Texas, USA), programmed by the first author. The programme was also used to extract the main food source of each nutrient and to determine the percentage of contribution of different food items in total daily intake of the nutrient. The mean and SD of daily energy and nutrient intakes were calculated for all participants at baseline and postintervention. As the index of diet quality, nutrient densities per 1000 kcal were calculated by dividing each participant's daily nutrient intake by their energy intake (kcal), multiplied by 1000. Observations with energy outliers of >5000 kcal were excluded. No participant reported daily energy intake <500 kcal. The change in postintervention to baseline daily consumption of each nutrient was computed at an individual level by subtracting the baseline value from the postintervention value. A positive value of change for a given nutrient was interpreted as a greater intake of the nutrient postintervention compared to baseline. For normally distributed and skewed values of nutrients, paired t test and Wilcoxon signed-rank test were used, respectively, to determine if the mean intake values at baseline and post-intervention differed statistically.
The impact of the intervention was assessed using multivariate linear regression models.24 For this study, only results relating to promoted nutrients (dietary fibre, total folate, and vitamins A and D) and depromoted nutrients (fat and sugar), energy intake and nutrients with significant change (α<0.001) from baseline to postintervention were reported. Postintervention nutrient intakes were regressed on a series of independent variables including intervention assignments, the difference between mean baseline and individual baseline nutrient intake (to adjust for regression to mean),25 baseline energy intakes, age, sex, body mass index (BMI), physical activity, smoking, education, MSL scale, households on income support and employment status. To avoid overadjustment, baseline energy intake was excluded when added sugar, and fat and its relevant variables (eg, saturated and unsaturated fatty acids), were included in the model.
All the nutrients with clear dietary reference intakes (DRIs) values26 ,27 selected for the regression models were also assessed for the percentage of adherence to DRI. Participants’ intakes were compared to gender and age-specific estimated average requirements (EARs), to examine dietary adequacy. For dietary fibre and sodium, the EAR was not available; thus adequate intake (AI) was used.
Statistical analyses were run through Stata V.11 (StataCorp LP). Subject to the possibility of obtaining false-positive results (type I errors) by conducting multiple pair-wise tests on the study data, all p values were considered statistically significant at α<0.001 for two-sided tests.
Results
Seventy-six per cent of participants (n=378) completed the postintervention assessment. Sex proportion was different between those who completed the follow-up and those who did not (18% male and 82% female vs 30% male and 70% female, respectively, p=0.01). The two groups were not different for other demographic and health-related indices including age, education, marital status, BMI and smoking. Baseline demographic characteristics of the complete evaluation sample are presented in table 1. After excluding observations with missing data and/or extreme energy intakes (<500 or ≥5000 kcal), 39 men and 224 women were included. No analysis was performed on drop-outs following intervention. Mean age was 47.14 (SD=14.08) years for intervention groups and 42.77 (SD=10.87) years for control groups (p=0.01). Distribution of participants over gender, BMI, education, MSL score and employment categories was not statistically different between control and intervention communities. A larger proportion of participants in the control, compared to the intervention group, were from households with member(s) on income support (53.85% vs 38.37%, respectively; p=0.02).
Demographic comparison of intervention and control groups at baseline
The mean (SD) energy and nutrient intakes at baseline and postintervention are presented in table 2. Contribution of fat in total energy intakes for the intervention group reduced by 1.92% after intervention (95% CI −2.84 to −1.00; p for paired t test=0.0001). Additionally, significant reduction in ω-6 fatty acid (−2.03 g/day, 95% CI −3.44 to −0.62; p for Wilcoxon signed-rank test=0.0007) was observed in this group. No statistically significant changes in mean energy and nutrient intakes had been observed for postintervention compared with the baseline in the control group.
Mean (SD) of nutrient intake/day by treatment group, and difference in postintervention to baseline
When postintervention nutrient density values (per 1000 kcal) were compared to baseline nutrient density values, reduction in fat (−2.13 g/day, 95% CI −3.16 to −1.11; p for paired t test=0.0001), saturated fatty acids (−0.88 g/day, 95% CI −1.32 to −0.43; p for paired t test=0.0001), monounsaturated fatty acids (−0.92 g/day, 95% CI −1.33 to −0.50; p for paired t test<0.0001), polyunsaturated fatty acids (−0.51 g/day, 95% CI −0.78 to −0.24; p for paired t test=0.0003) and increases in iron (1.06 mg/day, 95% CI 0.46 to 1.65; p for paired t test=0.0006) were observed in the intervention group (data not shown). No statistically significant changes for these dietary factors were detected for the control group.
The percentage of individuals by intervention assignment who showed intake equal to or greater than the EAR or AI at baseline and postintervention stages for selected nutrients is presented in table 3. For vitamins A and D, the proportion of participants adhering to the recommended level of intake increased in the intervention group (4.65% (95% CI −4.43 to 13.73) and 4.06% (95% CI −2.83 to 10.97), respectively) but decreased in the control group (−15.39% (95% CI −29.01 to −1.76) and −8.80% (95% CI −17.74 to 0.16), respectively). The percentage of adherence to recommended fat intake was reduced more in the intervention group compared to the control group (−15.70% (95% CI −23.55 to −7.84) vs −3.30% (95% CI −12.71 to 6.12)).
Frequency, percentage and percentage change of participants with nutrient intake equal or higher than the DRI by treatment group
After adjustment, participants receiving intervention significantly reduced their energy intake (β=−317, 95% CI −570 to −64) but increased their vitamins A and D intake (β=232.38, 95% CI 104.04 to 360.72 and β=53.46, 95% CI 1.88 to 108.80, respectively) from baseline to postintervention (table 4).
Impact of the Healthy Foods North intervention programme on postintervention nutrient consumption (results from fully adjusted regression models)*
Discussion
This study evaluated the impact of the HFN programme on nutrient intake among Inuit and Inuvialuit communities in the Canadian Arctic. The intervention was associated with a decline in energy intake and increased intake of vitamins A and D.
After identifying the dietary inadequacy of essential nutrients such as vitamins A and D,19 ,20 intake of traditional foods rich in these nutrients were promoted as part of the programme's intervention. Although there was only a 4% increase in the proportion of participants adhering to the DRI postintervention, the increase of dietary vitamin D intake following intervention may play an important role in adherence to recommended intakes. This is particularly relevant for Aboriginal populations in the Canadian Arctic who experience limited solar ultraviolet B absorption owing to darker skin pigmentation, high geographical latitudes, limited sun exposure during winters28 and, consequently, limited dermal vitamin D production. Similarly, the increased daily intake of vitamin A, observed as almost a 4.7% increase in adherence to DRI postintervention, may be considered as a substantial improvement in dietary quality. Further analysis demonstrated traditional foods as primary dietary sources of vitamins A and D and the consumption of this food group increased by 21% from baseline to postintervention in intervention communities versus only 3.3% in control communities (data not shown). A food group analysis showed a 95 g/day (95% CI −36 to 225) difference between the intervention and control groups for the preintervention to postintervention change in traditional food intake, suggesting an overall improvement in traditional food intake in the intervention compared with the control group, although not significant statistically. Dairy products were other sources of increase in vitamin A intake (by 12%) in the intervention group. While the dietary source of vitamin A and potassium are similar in many settings, reduction in coffee, tea and chips intake (as the main sources of potassium intake in both groups) can justify the potassium reduction among the study population (data not shown). These results indicate the programme's success in promoting vitamins A and D intake, given the substantial reduction in adherence to these (−15.4% and −8.8%, respectively) in control communities. From a public health point of view, these results indicate the necessity and possibility for the improvement of general population knowledge and skill regarding healthy eating among the Aboriginal remote communities through a culturally appropriate bottom-up community-based intervention programme.
In other studies, community-based diet intervention has also been shown to be successful in improving vitamin A intake. Home gardening has been reported as a popular and effective strategy for enhancing vitamin A intake.28 A significant improvement in vitamin A intake was observed in a food-based approach in Mozambique through encouraging the communities to produce and consume orange-fleshed sweet potato.29 The HFN project focused more on animal sources of vitamin A, considering the limitations for farming in the Canadian Arctic.
Exposure to the intervention also led to decreased total energy intake. Since the introduction of store-bought foods to Arctic communities, overconsumption of energy-dense foods has increased.29 ,30 Excessive intake of energy can lead to weight gain, and obesity coupled with nutritionally poor diets is linked with increased risk of developing chronic diseases.5 ,31 ,32 The magnitude of reduction in energy intake after this intervention (5%) is less compared to 18% energy reduction in the America On the Move study33 or 25% reduction in Buzzard et al's study.34 However, participants in those studies were overweight and breast cancer survivors with possibly greater motivation for reduced energy intake compared to our study population who reside in an area with reduced access to healthy food options.
Intake of total fat, and saturated and unsaturated fatty acids, except ω-3 fatty acid, significantly decreased only in the intervention group. Results of the regression models indicated that changes could not be linked to the intervention alone. However, the intervention communities substantially reduced fat intake from depromoted foods (unpublished data). Decreased fat intake was the main source of total energy reduction.
An increase in iron intake postintervention was another significant improvement in the diet quality of the intervention communities. However, an intervention effect was not observed in the regression analysis. Traditional foods, particularly caribou, were the main sources of iron throughout the study.
A key component of HFN that contributes to its success is the community development and engagement approach, which integrates the work of key local, territorial and national stakeholders, public as well as private. The importance of active multisectoral participation in the success of population-based interventions has been addressed by the WHO and the FAO.5 Formative research on Inuit diet, and on traditional and contemporary attitudes and practices, provided a strong basis for the development of the HFN programme.17 All these components support a sustainable and effective intervention at the community level.35
This is the first nutritional intervention to be conducted among the Inuit and Inuvialuit populations in NU and NT. After only 1 year, several positive impacts were observed, including improvements in some nutrient intakes and decreased energy intakes. The communities in this study are remote and each one must be flown into; they are otherwise only accessible for a few months in the winter by ice road. Only two communities were relatively close; both received the interventions. Therefore, the likelihood of intervention contamination or change in availability of different food groups was low over a year of intervention. Therefore, changes in nutrient intake measured in this and the other study36 as well as any change in food intake pattern37 could be interpreted as the HFN effect at the individual level. Other interventions effectively decreased chronic disease risk factors usually over longer intervention periods, such as the 3-year Pathways project on Indigenous people in North America,38 ,39 or the 5-year Hartslag Limburg community intervention in the Netherlands.40 Therefore, it is reasonable to expect ongoing HFN intervention will benefit the communities.
The baseline evaluation showed that participants from intervention and control communities were similar in terms of gender, level of education, income, smoking and BMI status. However, participants in the intervention group were on average 4 years older than the control group (p=0.001). Nevertheless, selection bias is not a concern in this study because the participants were recruited through a random sampling method. In addition, the 4-year age difference between participants from the intervention and control communities is comparable with the 4-year general population age difference in these two community groups (average general population age in intervention and control are 27 and 23 years, respectively). Finally, we adjusted the regression models for any confounding effect of age. Men were under-represented in the current sample, and therefore the results may not be generalised to male Inuit and Inuvialuit populations; this was because household members were recruited based on their authority regarding food purchases and preparations. Diversity in response rate could be related to less health consciousness and less motivation to engage in healthier behaviours in communities with lower response rate and, therefore, representing non-response bias. This study was limited by possible recall bias, which is a distinctive restriction of FFQs and response bias as a consequence of the interviewer-administration method. However, the prospective design of the study eliminates the possibility of differential recall bias between participants from the two groups. The 24 h recall is often favoured for assessing dietary adequacy since it is not limited to a food list, which may lead to the omission of some consumed foods. However, with input from local residents, the community-specific QFFQs were developed to capture most foods among a limited list of accessible foods in the communities.19 ,20 The relative short period of the diet intake assessment (past 30 days) is a limitation for this study to overcome the possible confounding effect of seasonality and to extend the intervention effect on long-term intakes of nutrients of interest. The effect of intervention on postintervention nutrient intakes was measured after removing the confounding effects of age, sex, baseline energy and relevant nutrient intakes, BMI, smoking and socioeconomic factors. The possibility of a threat for internal validity of this study (ie, history, maturation) is low due to the comparison design intended for intervention and control (two group designs). In addition, the communities are remote and relatively isolated and the participants did not receive any concurrent health-related interventions during the study period. Ultimately, using the difference of mean and individual nutrient value in the regression models,25 the risk of regression to the mean is not a case for estimating intervention effects.
As a conclusion, integrated multi-institutional, multisectoral, community-based programmes, such as Healthy Foods North, can be implemented over a longer duration to improve dietary quality.
What is already known on this subject?
Healthy Foods North is a community-based intervention addressing known micronutrient deficiencies among Canada's Inuit and Inuvialuit populations.
What this study adds?
Healthy Foods North was associated with reduction in energy intake and increase in vitamin A and D intake.
Acknowledgments
The authors are grateful to our project participants, community staff, Aurora Research Institute, the Government of Nunavut and the Government of the Northwest Territories. The project was supported by American Diabetes Association Clinical Research award 1-08-CR-57, Government of Nunavut Department of Health and Social Services, Government of Northwest Territories Department of Health and Social Services, Health Canada, Public Health Agency of Canada, and the Nunavut and Northwest Territories Public Health Association.
References
Footnotes
Contributors MP analysed and interpreted the data and drafted the manuscript. CR oversaw data collection, all field activities and programme implementation, and critically reviewed the paper. FK, JG and AC contributed to interpretation of results and finalising the manuscript. JG also assisted in developing an in-depth interviewer guide and interviewer training. SS developed the conception and design of the study and oversaw all activities. All authors critically reviewed the content and approved the final version submitted for publication.
Funding Government of Canada-Health Canada; Government of Nunavut Department of Health and Social Services; and American Diabetes Association.
Competing interests None.
Ethics approval Committee on Human Studies at the University of Hawaii and the Office of Human Research Ethics at the University of North Carolina at Chapel Hill. Additionally, the Ethics Committee of the Beaufort Delta Health and Social Services Authority approved this project. Aurora Research Institute in the NWT and the Nunavut Research Institute in NU provided research licenses.
Provenance and peer review Not commissioned; externally peer reviewed.