In infants and children, iron-deficiency anaemia (IDA) can have serious health consequences including impaired growth and cognitive development and weakened immune defence(Reference Beard1–Reference Yip3). Iron deficiency (ID) typically exists in three stages: low iron stores, reduced iron delivery to the tissues and IDA characterized by low Hb and reduced erythrocyte size(Reference Gibson4). The aboriginal people of Canada include three distinct groups: Inuit, First Nations and Métis(5), and there is evidence that the rates of anaemia and ID are higher among aboriginal children than among non-aboriginal children. Recent prevalence estimates for Inuit infants are 36–60 % for ID compared with 33 % for non-aboriginal Canadian infants(Reference Christofides, Schauer and Zlotkin6–Reference Zlotkin, Ste-Marie and Kopelman8). IDA is thought to affect 26 % of Inuit infants compared with 5 % for non-aboriginal infants(Reference Christofides, Schauer and Zlotkin6–Reference Zlotkin, Ste-Marie and Kopelman8). Information for the preschool age group of 3–5 years is lacking. Prevalence estimates for Canadian infants and children combined are 24 % for tissue ID and 5 % for anaemia from all causes(9). Current information on ID and IDA among Canadian Inuit preschoolers, however, is not available.
Studies among Inuit children show that dietary iron intake is most likely adequate(Reference Christofides, Schauer and Zlotkin6, 10–Reference Young, Moffat and O’Neil12). However, a nutrition transition in the Arctic is rapidly occurring, which warrants ongoing nutritional status assessment and biomarker monitoring. Further, infection with the human pathogen, Helicobacter pylori, has been postulated to contribute to ID, although the mechanisms remain unclear(Reference Annibale, Capurso and Lahner13–Reference Kostaki, Fessatou and Karpathios23). H. pylori infection is highly prevalent in Arctic populations(Reference Christofides, Schauer and Zlotkin6, Reference Bernstein, McKeown and Embil24, Reference Sinha, Martin and Sargent25) and may increase the risk for ID among Inuit children(Reference Baggett, Parkinson and Muth16). Therefore, a cross-sectional survey of Inuit preschool children in Nunavut was used to evaluate the prevalence and correlates of anaemia, ID and IDA. The study to date has identified a high prevalence of child food insecurity (56 %) as well as of other indicators of socio-economic disadvantage including household crowding (53·9 %), income support (42·7 %) and living in public housing (69·7 %)(Reference Egeland, Pacey and Cao26). The children are of normal stature(Reference Galloway, Young and Egeland27), but have a high prevalence of overweight based upon Centers for Disease Control and Prevention (CDC)(Reference Galloway, Young and Egeland27) and WHO international standards(Reference Egeland, Pacey and Cao26). Further, nearly half of the children consumed traditional Inuit food in the past 24 h and traditional food contributed to nutrient intakes, including iron(Reference Johnson-Down and Egeland28).
Experimental methods
Setting
This research is part of the Inuit Child Health Survey of preschool children in the Nunavut Territory of Canada, with details of the study design and demographic characteristics described elsewhere(Reference Egeland, Pacey and Cao26, Reference Johnson-Down and Egeland28). Using currently available population census information, we estimated that a sample of ninety to 100 children would provide 90 % power to detect a population prevalence of IDA of 10 (sd 5) %. A sample size of 300 was then desired as a minimum to allow for multivariable analyses and assessment of other correlates. Sixteen of the twenty-five communities in Nunavut were chosen to participate. The communities were selected to represent the three jurisdictional regions within Nunavut, to be geographically dispersed by latitude, and to represent small, middle and large-size communities. From these communities, Inuit children, aged 3–5 years, were randomly selected to participate in the survey using health centre lists of age-appropriate children. Recruiters were instructed to make three attempts to reach caregivers. Written informed consent was obtained from the children’s caregivers. The survey was developed by a steering committee consisting of partners from Inuit and community organizations, Nunavut health officials and McGill University and the University of Toronto. Certification of ethical acceptability for research involving human subjects was obtained from the McGill Faculty of Medicine Institutional Review Board. A scientific research licence was obtained from the Nunavut Research Institute.
Anthropometry
For the current report, age- and sex-appropriate BMI Z-scores were based upon the 2000 CDC growth reference(Reference Kuczmarski, Ogden and Guo29).
Iron status and exposure to H. pylori
Venous or capillary sampling was used to obtain blood samples. When venepuncture was used, 3 ml of blood was collected into sodium heparin Vacutainer® blood tubes (Becton Dickinson, Franklin Lakes, NJ, USA). The Vacutainer tube was inverted gently ten times. One drop of whole blood was dispensed onto Parafilm (Pechiney, Chicago, IL, USA) using a Diff-Safe® blood dispenser (Alpha Scientific Corporation, Southeastern PA, USA). Hb was measured either from this drop or from capillary blood samples using the cyanmethaemoglobin method with a HemoCue™ 201+ portable photometer (HemoCue Inc., Lake Forest, CA, USA). Blood samples were centrifuged within 6 h of collection. Separated plasma was stored at −20°C during fieldwork and at −80°C after completion of data collection.
Ferritin was measured from plasma samples using an autoanalyser (Liason®; DiaSorin, Saluggia, Italy) and a ferritin integral (REF 313551, DiaSorin, Saluggia, Italy). Low, normal and high control samples were tested with each analysis. C-reactive protein (CRP) was measured using a SYNCHRON® autoanalyser (Beckman Coulter, Inc., Brea, CA, USA) and a high-sensitivity CRP (hsCRP) assay at the Montréal Children’s General Hospital, Montréal, Canada. Previous exposure to H. pylori was assessed using a qualitative ELISA (Pylori Detect IgG; Calbiotech, Spring Valley, CA, USA) for the presence of anti-H. pylori IgG antibodies in plasma. Logistical aspects of conducting carbon urea breath tests given the other research priorities precluded assessment of current H. pylori infection.
ID was defined as low ferritin (<12 μg/l)(2). IDA was defined as the presence of low ferritin coupled with low Hb. Other anaemia was defined as low Hb, but normal ferritin (≥12 μg/l). As the cut-off value for low ferritin in children is currently unclear, we also conducted analyses using ferritin <10 μg/l to define ID and IDA(Reference Zlotkin, Ste-Marie and Kopelman8, Reference Looker, Dallman and Carroll30). The presence of acute inflammation elevates circulating ferritin concentrations; therefore, prevalence estimates for ID and IDA were restricted to children in whom hsCRP was below 8 ng/ml(Reference Willows, Dewailly and Gray-Donald7). Analyses were also repeated using a lower hsCRP cut-off of 3 ng/ml(Reference Beard, Murray-Kolb and Rosales31, Reference Wander, Shell-Duncan and McDade32), but as no differences were identified, only results using the 8 ng/ml cut-off are presented.
Dietary intake
A 24 h dietary recall was conducted for each child participant with a non-consecutive day repeat recall conducted for a 20 % subsample. Interviewers were trained using a five-stage, multiple-pass interviewing technique. Food model kits were used to estimate portion sizes. Interviewers were asked about the child’s mineral and vitamin supplement use, including frequency and brand information to allow for determination of nutrient content. Each caregiver was asked to complete a past-month qualitative FFQ for their child on traditional Inuit foods and market foods high in iron. Food frequency information was entered using EpiInfo (CDC, Atlanta, GA, USA). The 24 h dietary recall information was entered using CANDAT (Godin London Incorporated, London, Ontario, Canada). Iron intake was obtained using the Canadian Nutrient File (Health Canada 2007) and a database of 2000 additional foods derived from standardized recipes and food labels. All dietary data entries were double verified for errors. Traditional food intake used in univariate analyses was dichotomized based on a frequency of consumption that was greater or less than the median.
The home environment
Interviewers conducted questionnaires for characteristics of the home including the United States Department of Agriculture’s eighteen-item Household Food Security Survey Module adapted for the Inuit populations(Reference Lawn and Harvey33, 34). The food security module was scored according to Health Canada guidelines and detailed methods and findings are described elsewhere(Reference Egeland, Pacey and Cao26). Household crowding was defined as living in a home with greater than the median number of people per household.
Statistical analyses
There was incomplete ascertainment of all variables for all children. Univariate statistical analyses were conducted using all available data (H. pylori, n 282; hsCRP, n 254; Hb, n 285; ferritin, n 253; iron intake, n 374). Weighted prevalence rates with 95 % CI for anaemia, IDA and ID were estimated. Sampling weights were based on the proportion of participating children in each community using the total number of age-appropriate children obtained from health centre lists as the denominator for calculating weights.
Usual iron intake from the 24 h recall was estimated from observed intake using Software for Intake Distribution Estimation (Iowa State University, 1996). Adjustments were made for sequence and day of week of the recall. Within-person variability was estimated using information from the 20 % subsample of repeat recalls. The percentage of children below the age-appropriate Estimated Average Requirement (EAR) for iron was determined(35). The frequency of consumption of iron-containing traditional and market food was calculated, both for consumers only and for all children. The three outcomes of interest were ID, IDA and other anaemia. Univariate analyses of outcome and exposure variables were performed using a χ 2 test or Fisher’s exact test when cell sizes were <10. Relative risks (RR) and 95 % CI were calculated for relevant exposure variables. As sex differences were noted in ID, a post hoc t test examined sex differences in dietary iron intake. Multivariable logistic regression was performed to examine independent effects of exposure variables when variables were of borderline significance (P < 0·10) in univariate analyses; adjusted OR were calculated from regression coefficients. For all analyses, a P value <0·05 was considered significant. Weighted prevalence rates and dietary adequacy analyses were determined using the SAS statistical software package version 9·1 (SAS Institute, Cary, NC, USA). All other analyses were performed using the STATA statistical software package version 10·0 (StataCorp., College Station, TX, USA).
Differences in Hb levels obtained from capillary and venous blood samples were evaluated to determine the feasibility of combining the data in an overall assessment of anaemia prevalence. Mean Hb concentration in capillary blood samples (114 (sd 12) g/l) was significantly lower than that in venous blood samples (118 (sd 7) g/l, Wilcoxon rank P < 0·01), perhaps attributed to a dilution effect associated with finger prick sampling. There were, however, no socio-economic differences noted between those who provided a capillary or a venous blood sample, which suggested that anaemia prevalence rates would not be biased by excluding capillary blood samples from the analyses. Thus, analyses involving the Hb reported herein were based solely upon venous blood samples.
Results
Of the 644 homes approached, a total of 537 homes were successfully contacted, of whom seventy-five (11·6 %) refused upon initial contact and seventy-four (13·8 %) cancelled or did not attend their clinic appointment. The overall participation rate was 72·3 % (n 388). Venous blood was obtained from 289 of 388 children (74·7 %) and capillary blood samples from seventy-nine of 388 children (20·4 %). Five per cent of the children did not undergo any blood sampling.
Population characteristics
Fifty-three per cent of the participating children were female and the mean age was 4·4 (sd 0·9) years. Daycare was attended by 38·3 % of children. Sixteen per cent of caregivers reported that they gave their children a nutritional supplement containing iron, most commonly multi-vitamin and mineral supplements. The weighted prevalence of exposure to H. pylori was high at 46·1 % (95 % CI 40·1, 52·1; Table 1). The median hsCRP concentration was 0·65 ng/ml (25th percentile: 0·2 ng/ml; 75th percentile: 2·2 ng/ml). Overall, fourteen children (5·1 %; 95 % CI 2·2, 8·0) had high hsCRP concentrations of ≥8 ng/ml. These children, along with twenty-one children whose hsCRP status could not be determined, were excluded in determining the prevalence of ID and IDA as well as from univariate and multivariate analyses.
Table 1 Prevalence of ID, anaemia, IDA and Helicobacter pylori infection among participating children: Nunavut Inuit Child Health Survey, 2007–2008
ID, iron deficiency; IDA, iron-deficiency anaemia.
†Presence of anaemia coupled with ID.
‡Hb <110 g/l (in 3–4-year-olds) or <115 g/l (in 5-year-olds).
Prevalence of iron deficiency and anaemia
The mean ferritin concentration was 19·1 (sd 10·1) μg/l and the median was 16·6 μg/l (Fig. 1a). Overall, 18·0 % (95 % CI 12·7, 23·3) of children were iron deficient using a ferritin cut-off of 12 μg/l (Table 1). Using a ferritin cut-off of 10 μg/l, the prevalence of ID decreased to 10·9 % (95 % CI 6·5, 15·2; Table 1). IDA was found in 5·4 % (95 % CI 2·3, 8·6) of children and when the lower ferritin cut-off was used, the prevalence of IDA decreased to 3·3 % (95 % CI 0·9, 5·8).
Fig. 1 (a) Distribution of Hb concentration values among Inuit preschoolers, where anaemia is defined as Hb below 110 g/l (3–4-year-olds) or 115 g/l (5-year-olds). (b) Distribution of ferritin concentration values among Inuit preschoolers, where ferritin below 10 or 12 μg/l is defined as iron deficiency in children (Nunavut Inuit Child Health Survey, 2007–2008)
The mean Hb concentration was 118 g/l (sd 8·0; Fig. 1b). The weighted prevalence of anaemia from all causes was 16·8 % (95 % CI 12·0, 21·6). Among children with anaemia, 4·3 % had moderate-to-severe anaemia (Hb < 100 g/l), whereas 95·7 % had mild anaemia (Hb ≥ 100 g/l, but below age-specific cut-off). ID (ferritin <12 μg/l) explained 31 % of observed anaemia. When ID was defined as ferritin below 10 μg/l, it explained 20 % of the observed anaemia. Thus, 69–80 % of the anaemia observed in Inuit preschoolers was most likely due to causes other than low iron stores.
Dietary iron intake
Iron intake was normally distributed with a mean of 15·6 (sd 11·0) mg/d and median of 13·5 mg/d. Only 0·3 % of children had iron intake below their age-specific EAR (1–3 years: 3·0 mg/d; 4–8 years: 4·1 mg/d). Based on the FFQ, children commonly consumed various iron-rich foods such as caribou meat (84·2 %), ringed seal meat (49·5 %), beef (82·6 %) and breakfast cereals (96·3 %; Table 2).
Table 2 Frequency of consumption of traditional and market food sources of iron among Inuit children aged 3–5 years: Nunavut Inuit Child Health Survey, 2007–2008
†Owing to the limited geographical availability of this food, N is reduced because certain communities were not asked about this traditional food item.
Correlates of iron deficiency and iron-deficiency anaemia
In univariate and multivariate logistic regressions, no significant correlates emerged for ID or for IDA when using the ferritin cut-off value of 12 μg/l. However, when using the lower ferritin cut-off value (10 μg/l), age and sex were significantly associated with ID (Table 3) but not with IDA (data not presented given the small number of IDA events). In univariate analyses, boys were more likely to be iron deficient than girls (RR = 3·22, 95 % CI 1·33, 7·78), and children aged 3–4 years were more likely to be iron deficient than 5-year-olds (RR = 3·31, 95 % 1·02, 10·71). In a multiple logistic regression model for ID containing age, sex and BMI Z-scores, only boys were significantly at risk (adjusted OR = 2·28, 95 % CI 1·17, 8·25). However, there were no significant differences in mean dietary iron intakes between boys (15·5 (sd 5·8) mg/d) and girls (15·6 (sd 6·1) mg/d).
Table 3 Univariate analyses for correlates of iron deficiency (ferritin <10 μg/l) and anaemia from other causes: Nunavut Inuit Child Health Survey, 2007–2008
*P < 0·05; ***P < 0·001.
Correlates of other anaemia
Correlates for anaemia from causes other than ID were examined (Table 3). Three- to 4-year-olds were less likely to be anaemic from other causes (5·3 %) than 5-year-olds (19·2 %; RR = 0·28, 95 % CI 0·13, 0·60). Children residing in a home with six or more individuals were more likely to be anaemic (12·3 %) than those living in less crowded homes (4·8 %; RR = 2·56, 95 % CI 1·04, 6·34). Children who were treated for an ear infection in the past 12 months were more likely to be anaemic (14·5 %) than children not requiring treatment for an ear infection (6·6 %; RR = 2·20, 95 % CI 1·01, 4·76). There were no differences in mean BMI Z-scores or in hsCRP concentrations between children with and without anaemia from other causes. In a multiple logistic regression model adjusted for BMI Z-scores and sex, 3–4-year-olds were less likely to be anaemic relative to 5-year-olds (OR = 0·11, 95 % CI 0·08, 0·58), while household crowding (OR = 2·30, 95 % CI 1·37, 12·31) and past-year treatment for an ear infection (OR = 1·35, 95 % CI 1·05, 7·21) were associated with a significantly elevated risk for having anaemia from other causes.
Discussion
The present study is the first to report population-level prevalence estimates of ID and IDA for Inuit preschoolers in the Nunavut Territory, Canada. National prevalence estimates for Canadian preschoolers are currently not available. However, in comparison with American preschoolers, Inuit preschoolers have a higher prevalence of ID and IDA. In the USA, ID affects 4·5 % of children aged 3–5 years and IDA is found among 0·5 %(Reference Cardenas, Mulla and Ortiz19), whereas in the present study 10·8–18·0 % were iron deficient and 3·3–5·4 % had IDA, depending upon the cut-offs used. Natives of Alaska(Reference Baggett, Parkinson and Muth16) as well as Canadian Inuit infants(Reference Christofides, Schauer and Zlotkin6, Reference Willows, Dewailly and Gray-Donald7, 9) have been observed to have higher rates of ID than the general population(Reference Christofides, Schauer and Zlotkin6, Reference Cardenas, Mulla and Ortiz19). The prevalence rates determined from the present study would most likely be defined as mild according to WHO thresholds for population-level ID(36). However, improvements in iron status are possible given that rates of deficiency continue to be higher among Inuit preschoolers compared with preschoolers of the general US population.
Iron intake in this population is most likely adequate as only 0·3 % of children had intake below the EAR. In addition, breakfast cereals, many of which are iron-fortified, and beef and caribou were frequently consumed in this population. Among the same study population, only 0·1 % had intake below the EAR for vitamin C(Reference Johnson-Down and Egeland28), suggesting that low iron bioavailability due to low vitamin C intake is unlikely. While over-reporting of portion sizes on 24 h dietary recalls is possible, the energy intake reported in the present study is similar to that observed in the Canadian Community Health Survey(37). Dietary iron intake levels are also similar in the present study in comparison with others in Inuit, Métis and American children(Reference Kuhnlein and Receveur11, Reference Young, Moffat and O’Neil12, Reference Kuhnlein, Soueida and Receveur38, Reference Moshfegh, Goldman and Cleveland39).
The finding that more boys than girls were iron deficient is difficult to explain since there are no sex differences in iron requirements in the preschool age group. It has recently been shown that obesity and being overweight are associated with greater risk of ID, perhaps due to low diet quality, increased iron requirement due to higher blood volume as well as decreased iron absorption induced by chronic low-grade inflammation(Reference Brotanek, Gosz and Weitzman40–Reference Zimmermann, Zeder and Muthayya44). In another analysis of the Inuit Child Health Survey, boys were noted to have higher BMI-for-age Z-scores than girls(Reference Galloway, Young and Egeland27). However, we found no significant associations between ID and BMI Z-scores, and BMI Z-scores did not alter the relationship between sex and ID in logistic regression. It is possible that the association between BMI Z-scores and ID is relevant to Inuit preschoolers, but our study was not powered to detect this perhaps weak or moderate association.
In addition, the present study used a 24 h dietary recall and a 20 % random subsample repeat recall, which does not provide enough information to estimate nutrient intake for individuals. As such, nutritional ID or anaemia cannot be ruled out. However, since meat and dietary iron intake was high in the population overall, nutritional causes are unlikely to explain the observed ID and anaemia found among preschoolers.
The 45·4 % prevalence of H. pylori exposure observed in the present study is high and consistent with other studies of Canadian First Nations and Inuit and Alaskan native children(Reference Christofides, Schauer and Zlotkin6, Reference Baggett, Parkinson and Muth16, Reference Bernstein, McKeown and Embil24, Reference Sinha, Martin and Sargent25, Reference Parkinson, Gold and Bulkow45). In contrast, lower rates of H. pylori exposure have been reported (5·5–7·1 %) for American and Canadian children(Reference Cardenas, Mulla and Ortiz19, Reference Segal, Otley and Issenman46). We found no association between ID and H. pylori exposure among Inuit preschoolers. The present study is limited in that H. pylori infection assessment was related to previous exposure(Reference Malaty, Haveman and Graham47–Reference Okuda, Miyashiro and Koike49), thereby precluding comparisons with studies evaluating current infection. However, in various epidemiological studies and case reports in which H. pylori was shown to be independently associated with iron status, the association emerged primarily among older children(Reference Baggett, Parkinson and Muth16, Reference Cardenas, Mulla and Ortiz19, Reference Carnicer, Badia and Argemi20–Reference Kostaki, Fessatou and Karpathios23, Reference Barabino, Dufour and Marino50). For example, among the natives of Alaska, H. pylori was independently associated with ID for children aged 9 years and above, but not in younger age groups(Reference Baggett, Parkinson and Muth16). A causal relationship between H. pylori and iron status has yet to be established, but possible mechanisms involve bacterial damage to gastric glandular tissue and competition for iron in the stomach(Reference Annibale, Capurso and Lahner13, Reference Ashorn15, Reference Baysoy, Ertem and Ademolu18, Reference Dhaenens, Szczebara and Husson51–Reference Sarker, Davidsson and Mahmud53). In addition to the limitations explained above, perhaps the young age of the study group is relevant in explaining the lack of association between iron status and previous H. pylori exposure.
Although 10·8–18·0 % prevalence of mild ID warrants continued public health attention, it is reassuring that the prevalence of IDA, the most severe form of ID, is low. However, apart from sex, we found no significant correlates of iron status in Inuit preschoolers in Nunavut. Iron absorption is one relevant issue that was not explored in the present study. It has been postulated that the Inuit may have adapted to excessive and deleterious iron intake through lowered iron absorption, which may now have implications for low iron stores given the nutrition transition away from iron-rich traditional foods(Reference Gessner54). However, little evidence is currently available to support or refute this hypothesis(Reference Baggett, Parkinson and Muth16, Reference Gessner54).
Finally, in the present study, anaemia from all causes was found in 16·8 % of Inuit children and only 20–31 % of this anaemia was explained by low iron status. Other studies in children and infants have shown similar results in which only a portion of the observed anaemia is explained by ID(Reference Gamble, Palafox and Dancheck55–Reference White57). Dietary causes of anaemia other than low iron intake include deficiencies in vitamin A, folate, vitamin B12 and riboflavin(Reference Fishman, Christian and West58). However, again, given the high meat and cereal intake noted, most of these micronutrient deficiencies are unlikely. Vitamin A deficiency could be evaluated in future research given that an earlier study found that young Inuit adults had a greater likelihood of having a retinol activity equivalent falling below the EAR than older Inuit adults, attributed to the changing pattern of traditional food consumption by age(Reference Egeland, Berti and Soueida59). However, in analyses of the present study population, the majority of children had a vitamin A intake above the EAR(Reference Johnson-Down and Egeland28).
Anaemia not attributable to ID could be related to acute inflammation. In acute infection, inflammation reduces erythrocyte half-life and the acute phase response blocks iron export proteins trapping iron inside cells(Reference Abshire and Reeves60–Reference Weiss62). Interestingly, we found that children who required treatment for an ear infection in the past year, when compared to those who did not, had a significantly greater risk for being anaemic. Ear and respiratory infections are common among Inuit children(Reference Banerji, Bell and Mills63, Reference Alaghehbandan, Gates and MacDonald64). One hypothesis is that recurring infections during childhood may result in mild anaemia, although the reverse may also occur, where anaemia may increase the risk of infection(Reference Walter, Olivares and Pizarro65, Reference Levy, Fraser and Rosen66). We also found that household crowding significantly predicted anaemia. Household crowding is a relevant health indicator for young children who spend the majority of their time at home with increased risk for person-to-person spread of infections(Reference Banerji, Bell and Mills63, Reference Baker, McNicholas and Garrett67–Reference Kovesi, Stocco and Fugler69).
Limitations
Owing to the rarity of IDA in this population, our study was underpowered to detect significant correlates. The cross-sectional design prevents exploration of causal relationships between correlates and ID, IDA and anaemia.
Conclusion
Inuit children aged 3–5 years have higher rates of ID, IDA and anaemia than non-aboriginal children. The rates observed, however, were not excessively high and do not warrant immediate intervention. The present study also revealed that ID explains only 20–31 % of low Hb and that anaemia was generally mild. The role of household crowding, history of infections and age need to be evaluated for their role in exacerbating anaemia. The study found that Inuit preschoolers have iron intake levels that are most likely adequate for their age, which is an important positive finding for children likely to be exposed to rapid nutrition transition in the Arctic. Health promotion efforts are best placed at encouraging beneficial dietary behaviours that already exist in Nunavut’s communities and in extending these beneficial behaviours in efforts targeting high-risk women of reproductive age and infants. Caregivers should be commended for feeding preschoolers many different iron-rich foods, especially iron-rich traditional meats, and encouraged and supported to continue these practices while reducing intake of foods that are less nutrient-rich but contribute to high energy intake.
Acknowledgements
Funding for the present study was provided through the Government of Canada International Polar Year, the Canadian Institutes for Health Research, Indian and Northern Affairs Canada and Health Canada. There are no conflicts of interest to declare. A.P. facilitated the collection of field data, conducted background research, performed data analyses and drafted the manuscript. H.W. helped with lab analyses and data interpretation and with final editing of the manuscript. G.M.E. was the principal investigator of the Nunavut Inuit Child Health Survey; she developed the content and design of the survey as well as guided the statistical analyses, interpreted the results and reviewed and edited the manuscript. All authors have approved the final version submitted for review. The authors would like to acknowledge the Nunavut Inuit Health Survey Steering Committee and extend a special thanks to Laureen Pameolik, Kathy Morgan, Christine Ekidliak and Nancy Faraj for field survey work and to Louise Johnson-Down for assistance with dietary intake analyses and quality control.
