Higher n3-fatty acid status is associated with lower risk of iron depletion among food insecure Canadian Inuit women

Study population

The studied population was non-pregnant, self-identified Inuk women, 18 years of age or older from ISR, Nunavut, and Nunatsiavut in Arctic Canada. According to the 2006 Statistics Canada Census the Inuit population was 50, 485.

Sampling and data collection

A cross-sectional Inuit health survey of adults residing in randomly selected households was carried out between August to September, 2007, and August to October, 2008 in 36 Canadian Arctic communities [17]. Households were randomly selected from a list of house numbers in each participating community. Of the 2796 Inuit households approached, 1901 households (68%) agreed to participate, with a total of 2595 adult participants, of whom 61.5% were women. Forty-five women were excluded from current analyses due to possible pregnancy or lactation. After providing voluntary and informed consent, a 24-hour recall, semi-quantitative food frequency questionnaire, 5 general questionnaires, a fasting venous blood sample, and anthropometric measurements were obtained either on-land or utilizing the Canadian Coast Guard Ship Amundsen. Venous blood samples were obtained from 1403 (87.9%) female participants of whom 1309 women had both SF and high-sensitivity C-reactive protein (hs-CRP) measured and met the inclusion criteria. Appropriate regional research licenses were obtained and ethical approval for the survey was granted from the McGill University Faculty of Medicine Institutional Review Board. Survey design and implementation was directed by a participatory process [17].

Clinical assessment

Weight and body composition were measured using bioelectrical impedance analysis (Tanita TBF-300GS, Arlington Heights, IL, USA) and height was measured with a portable stadiometer (Road Rod 214, Seca, Maryland). Normal weight, overweight and obesity were defined by the WHO classification system [18]. Percent body fat was classified according to the manufacturer’s age-appropriate healthy body fat ranges. For women, an at-risk body fat percentage was defined as >33%, >34%, and >36% for ages 18-39, 40-59, and 60-99 y, respectively.

Laboratory analyses

Participants were requested to fast at least 8 hours overnight, prior to morning venous blood collection. Hemoglobin measures were obtained from venous blood drops or blood drops from a finger prick using the azidemethemoglobin method with HemoCue™ 201+ portable photometer (HemoCue, Inc., Lake Forest, California), and adjusted for cigarette smoking according to WHO guidelines [19]. Prevalence estimates of anemia were derived from venous samples only and categorized according to the WHO 120 g/L cut-off for non-pregnant women [19]. Serum samples were analyzed for ferritin (automated chemiluminescence assay; Liaison Ferritin;Diasorin, Italy), hs-CRP (auto-analyzer; Beckman Coulter, Brea, CA, USA), and soluble transferrin receptor (sTfR) concentration on a subsample (n = 652, ELISA assay; R&D Systems, Minneapolis, USA). Serum sTfR > 2.75 mg/L was defined as iron deficient erythropoiesis (IDE), as suggested by the manufacturer. Depleted iron stores was defined by SF < 15 μg/L or SF = 15-50 μg/L in the presence of acute inflammation (hs-CRP ≥10 mg/L). SF > 200 μg/L in the absence of acute inflammation was used to define elevated iron stores in order to compare across studies of Inuit populations [12] with sex and age-appropriate cut-offs for iron overload [20] also determined. Iron deficiency anemia (IDA) was defined as anemia + SF <15 μg/L or SF 15-50 μg/L + hs-CRP > 10 mg/L. Immunoenzymatic methods (ELISA) were used to detect IgG antibodies against H. pylori in serum (Calbiotech; Spring Valley,CA, USA) and RBC membranes were analyzed for fatty acid composition (Lipid Analytical Laboratories Inc., University of Guelph Research Park, Guelph, ON), as previously described [21]. Total EPA and DHA were expressed as % of total fatty acids and hereafter referred to as LC-PUFA (long chain-PUFA).

Dietary assessment

Dietary assessment was conducted by trained interviewers using a single 24 hour recall with a four stage, multi-pass approach [22] and a 42-item semi-quantitative food frequency questionnaire. Portion sizes were estimated with a graduated, three-dimensional food model kit (Santé Québec). Recall data were entered into CANDAT software (Godin London Inc,. London, Ontario, Canada) and nutrient analyses obtained from the 2007b Canadian Nutrient File (CNF) and additional databases as described previously [21]. There were no missing nutrient values in the analysis. Recall data were available for 1248 female participants after 25 recalls were excluded due to incompleteness. Dietary iron adequacy was assessed by the Dietary Reference Intake (DRI) probability method (pre-menopausal women) or the Estimated Average Requirement (EAR) cut-point method (post-menopausal women) using the SIDE method [23] and SIDE software [24] in which within-subject variation estimates for iron intake were obtained from dietary surveys with Canadian Inuit populations [9]. Intake of iron and dietary enhancers and inhibitors of iron were analyzed for women using 24-hour recall data. Twenty-nine 18 year olds (with different DRI requirements for their age group) were excluded from this intake analysis. Reported food items were classified as TF (derived from hunting, fishing or gathering) or market foods (imported foods available for purchase). Nutritional supplement and medication use was recorded by a nurse. Supplement content was not included in nutrient analysis of the 24 hour recalls as the majority of supplement users could not recall the brand or amount of supplement taken. TF frequency data were available for 1229 female participants. Frequency of TF use over the past 12 months was documented for in-season and off-season consumption of each item. Seasons were established according to locally developed regional wildlife harvest calendars and intakes adjusted to frequency per month (assuming 30.4 days per month).

Questionnaires

Questionnaires for sociodemographic, health and household characteristics were adapted from Greenlandic and Nunavik (Canada) Inuit Health Surveys [25] and the Aboriginal Peoples Survey [26], The household questionnaire included a version of the 18-item USDA Household Food Security Survey Module [27], with details of the questionnaire and classification of household food security described elsewhere [28]. A household food security score was dichotomized into secure or insecure and applied to each adult household member for analyses. Current smoking status was assessed as yes or no and quantified by cigarettes per day. Menopausal status was self-reported and defined by absence of a regular menstrual period or age ≥50 y if menstrual status not reported. Women were asked about contraceptive use to account for known effects of hormonal contraceptives (estrogen and/or progesterone) and intrauterine device use on iron status [29].

Statistical analysis

SF and sTfR concentrations were log₁₀ transformed to improve normality of the respective distributions. Weighted prevalence estimates of iron status are provided with 95% confidence intervals. Sampling weights reflected the proportion of participating women using Statistics Canada’s Census data of age-appropriate Inuit women by community. Age categories for prevalence estimates were based on the DRI recommendations for iron intake [30], although age-groups 51-70 and ≥70 were combined due to small sample size among the elderly. Independent determinants of depleted iron stores vs adequate iron stores were assessed with a multivariable logistic model, with community included as a random effect to improve model fit. A priori selection of variables known or suspected to be related to iron status were evaluated. The model was based on 1062 women with SF available and hs-CRP ≤ 10 mg/L to reduce invalid SF results due to inflammation. Low-grade inflammation (hs-CRP 3-10 mg/L) was included as a control variable. Sample size limited interaction testing to only 2x2 interactions between main effects. Model specification was verified with STATA linktest and standard model diagnostics were performed. Generalized linear models were used to investigate the relationship between LC-PUFA status and obesity. All analyses were performed in STATA (version 11; StataCorp LP, College Station, TX). P values were all two-sided and significance was set at P ≤ 0.05, with adjustments for multiple comparisons as appropriate.

Study population

Mean age ± SD of the study population was 42 ± 15 y (range: 18-90 y), with 25% of women overweight and 43% obese using WHO criteria. Underweight was infrequent (1%). Seventy-one percent of women currently smoked, with a median 10 cigarettes/day (IQR: 6-12). Prescription or non-prescription medication use was reported by 43% of women, oral or hormonal contraceptives were used by 11% of premenopausal women and intrauterine device by 4%. In all, 70% of women were classified as premenopausal. High blood pressure, diabetes mellitus, and high cholesterol were self-reported in 28%, 7%, and 12% of women respectively and a low hs-CRP concentration (<3 mg/L) was observed in 71% of female participants. Sixty-two women reported taking an iron-containing supplement and were included in analyses because of lack of difference in SF between supplement users and non-users. TF consumption was reported by 57.3% of women on the day prior to the survey, representing 6.8% of energy intake. The majority of the sample (61%) was classified as married and 66% reported living with an active hunter in the home.

Iron status and anemia

Prevalence of anemia was moderate and similar across all age-groups (Table 1). IDA accounted for approximately half of the anemia cases, with the majority of IDA cases observed in premenopausal women. There was little evidence of iron deficient erythropoiesis (Table 1) but depleted iron stores were common and 74.5% of the population had SF < 50 μg/L. Elevated iron stores were rare and severe iron overload (SF > 700 μg/L) was absent. When using age-appropriate cut-offs for iron overload [20] rates were 0.3% (18-30 y), 0.9% (31-50 y) and 4.6% (≥51 y). SF concentration was low and stable in women until after age 48 (Figure 1) when it was more than 2-fold higher, whereas sTfR was low in all age-groups (Table 2). Adjusting for CRP < 10 mg/L and CRP < 3 did not appreciably alter SF concentrations or prevalence of iron depletion.

Age, y	Anemia2		Depleted iron stores		Elevated iron stores		Iron deficiency anemia		Iron deficient erthrypoiesis3		Iron intake < EAR4
	n	%(95% CI)	n	%(95% CI)	n	%(95% CI)	n	%(95% CI)	n	%(95% CI)	n	%
18-30	169	17.9	333	40.3	333	-	166	11.7	153	1.0	312	15.7
		(12.7-24.6)		(34.3-46.6)				(7.4-18.2)		(0.1-6.9)
31-50	326	21.3	583	37.0	583	0.9	316	15.4	294	6.2	583	17.3
		(16.7-26.8)		(32.7-41.6)		(0.3-3.0)		(11.5-20.3)		(3.3-11.4)
≥51	202	24.9	344	9.2	344	5.9	182	3.9	186	0.7	326	0.6
		(18.1-33.3)		(6.3-13.2)		(3.4-10.1)		(1.5-9.6)		(0.2-3.0)
Total	697	21.7	1260	29.4	1260	2.2	664	11.1	633	3.3	-	-
		(18.3-25.5)		(26.7-32.3)		(1.3-3.6)		(8.7-14.0)		(1.9-5.8)

¹ Where anemia = hemoglobin < 120 g/L; depleted iron stores = serum ferritin <15 μg/L or ferritin 15-50 μg/L + CRP > 10 mg/L; elevated iron stores = ferritin > 200 μg/L + CRP ≤ 10 mg/L; iron deficiency anemia = depleted iron stores + anemic; iron deficient erythropoiesis = serum soluble transferrin receptor (sTfR) > 2.75 mg/L. EAR = estimated average requirement.

²Analyses on a subset with venous blood sampling for hemoglobin determination.

³Analyses on a subset with soluble serum transferrin receptor measurements.

⁴Analyses on adjusted intakes for women aged 19-50 y using the probability method and women aged > 50 y using the EAR cut-point method, with women aged 18 y excluded.

Median intakes from carbohydrate, protein and fat (as% energy) were 45.2%, 17.3% and 29.4%, respectively. Prevalence of dietary iron inadequacy (adjusted intakes below the age-specific EAR) was much higher among pre-menopausal women (15.7-17.3%) than postmenopausal women (<1%). A small portion of the sample (4.1%) had adjusted iron intakes above the upper limit DRI (45 mg). Heme iron comprised 25.0% of total dietary iron intake for pre-menopausal women and 49.3% of total iron for postmenopausal women. Median intakes of vitamin C, vitamin A, and calcium were less than the EAR (Table 2). Tea and coffee were consumed by 43.8% and 3.7%, respectively, of women on the previous day. In order of contribution, the top five sources of iron for women 19-50 y were: TF meats (23%), baked products (13%), market food (MF) meats (13%), breads (10%) and mixed foods with meat (9%). For women over 50 y the greatest iron sources were: TF meats (44%), baked products (12%), bannock (8%), breads (7%), and MF meats (6%).

Correlates of iron status