Dietary intake and biomarker status of folate in Swedish adults

Riksmaten Adults 2010–11 [18] is the most recent national dietary survey in Sweden and was conducted by the Swedish National Food Agency (NFA). The method has been validated [19, 20]. Statistics Sweden invited a representative national sample of 5008 adults aged 18–80 years to participate in the dietary survey and a subsample of 1008 of these to also participate in assessments of nutritional status (Fig. 1). The subsample was divided into seven regions according to affiliation to Swedish Occupational and Environmental Medicine Centers (OEMCs). Each region included the regional capital (Linköping, Lund, Stockholm, Umeå, Uppsala, Gothenburg and Örebro) together with two randomly selected counties in each region. An equal number of individuals were selected in each region independent of population size. Recruitment took place at four different occasions during the year to cover different seasons. Twelve individuals per part of the region and occasion were invited to participate (7 regions, 3 sites in each, the capital and two counties, four occasions and 12 individuals per round). The participation rate was 36% for the dietary survey and 30% for the subsample (Fig. 1). The study protocol was approved by the Regional Ethical Review Board in Uppsala.

The participants reported everything they ate and drank for four consecutive days, using a web-based food diary developed by NFA [21]. Food intake and status in all seasons and all days of the week were captured by carrying out the study from May 2010 to July 2011 and by randomly assigning starting days to participants.

To estimate food and folate intake, a food composition database consisting of 1909 food items was used. Intakes of foods and beverages (gram per day) were estimated using a portion guide, household measures, numbers of portions (cups, pieces, slices) and grams. The portion guide, available as a printed booklet and in the web tool, included 24 different food categories with four to eight different reference sizes in each category. For about 30% of the foods, folate content was analysed using an accredited trienzyme microbiological method (L. rhamnosus, Culture Collection of the University of Gothenburg, CCUG 21452, equivalent to L. casei American Type Culture Collection, ATCC 7469). For about 20% of the foods, folate values were imputed from similar foods or borrowed from other national food databases, companies or scientific literature. The values for the remaining foods were calculated from recipes using analytical data on folate in ingredients and factors for losses and gains of water and fat as well as losses of folate due to heat treatment [22].

Information on supplement use and frequency was collected using a questionnaire as recommended by Bates et al. [12]. As most available folic acid supplements in Sweden contain 500 µg of folic acid (the RI for pregnant and nursing women [3]), the doses of all registered folic acid supplements were set to 500 µg of folic acid for estimation of intakes from supplements. Foods included in the survey were controlled for voluntary fortification and if relevant also fortification level. Dietary folate equivalents (DFE) were calculated according to the formulae: 1 μg DFE = 1.0 μg food folate = 0.6 μg folic acid added to foods = 0.5 μg folic acid taken without food [23].

The Goldberg cut-off (Black), which defines the probability of adequately reported energy intake, based on body size and physical activity level, was used to identify low-energy reporters. Information on physical activity level (PAL) at work and leisure time was collected as part of the dietary record on a four-grade scale. Criteria for acceptable energy intake were a quota of energy expenditure to basal metabolic rate within the confidence interval (0.93–3.01).

Non-fasting blood samples for folate status were collected in coded sterile Vacutainer™ tubes from BD (Belliver Industrial Estate, Plymouth, UK) by nurses at the OEMCs. Samples were kept at −20 °C until analysis.

Whole blood samples for erythrocyte folate status were collected in 3.0-mL EDTA tubes, and analysed using a chemiluminescence immunoassay method at the Karolinska University Hospital, Sweden. Erythrocytes were hemolysed using ‘RBC folate Hemolyzing Reagent’ (Roche Diagnostics GmbH, Mannheim, Germany) and quantified as serum folate (range 45–1407 nmol/L, intra-assay CV 3%). Erythrocyte folate concentrations were not corrected for serum folate concentrations.

Plasma folate samples were collected in 3.5-mL PST tubes and analysed at the Clinical Chemistry and Pharmacology Department at Uppsala University Hospital, Sweden. Plasma folate (cat. no. 1P74-35, Abbott Laboratories, Abbott Park, IL, USA) was analysed by a chemiluminescence immunoassay method on an Abbott Architect ci8200 analyser. The laboratory was accredited according to SS-EN ISO/IEC 15189:2007. As part of the accreditation procedures, the laboratory participated in interlaboratory external proficiency testing schemes for P-folate from EQUALIS AB, Uppsala and Labquality OY, Helsinki. The total analytical imprecision of plasma folate measurements was 12 and 7 CV % at 4 and 35 nmol/L, respectively. Cut-off level was 8 nmol/L.

Data were expressed as median, lower quartile (Q ₁) and upper quartile (Q ₃) or as percentage. Folate intakes were presented including and excluding low-energy reporters. Erythrocyte and plasma folate concentrations were not normally distributed as tested using the Shapiro–Wilk test, so nonparametric methods were used. The Wilcoxon–Mann–Whitney rank sum test was used to test if sex, age, income, fruit and vegetable consumption or low-energy reporting affected intake and status. The Kruskal–Wallis test was used to compare if education level affected intake and status. Associations (folate intake and biochemical parameters) were assessed by Spearman’s ranked correlation coefficients.

Food patterns associated with folate status parameters were identified using factor analysis followed by robust regression analysis. Food groups (in gram per day) included in the factor analysis were: vegetables, pulses and roots; fruit and berries; potatoes; bread; rice and grains; pasta; porridge and gruels; breakfast cereals; meat including offal and blood products; poultry; sausages; fish and shellfish; egg; milk, fermented milk and yoghurt; cheese; spreads and butter; coffee, tea and water; fruit and vegetable juice; soft drinks, sports and energy drinks; beer, wine and spirits (alcoholic beverages); jam, marmalade and apple sauce; ice cream; candy and chocolate; buns, biscuits and cakes; sugar, syrup, honey and artificial sweeteners; pizza, pie and pirogue; pancakes, waffles and crepes; soup; sauces; and diet and nutritional supplements, i.e. bars, powder. Rotated factor loadings above 0.300 were included in the final analysis. To simplify the interpretation of the selected factors, the varimax (orthogonal) rotation was applied, this has as it is rational the provision of uncorrelated factors with a few large loadings and as many loadings as possible close to zero.

To maintain blood folate concentrations above cut-off values (plasma and erythrocyte folate concentrations below 6.8 and 317 nmol/L, respectively), NNR set average requirement (AR) to 200 µg/day and recommended intake (RI) to 300 µg/day [3]. Previous studies in Sweden report average or median folate intakes ranging from AR and up to RI (Table 4) [4‐9, 26‐31]. In this survey, nearly 75% reported a folate intake above AR and about 25% above RI. However, out of the 450 women aged 18–45 years in this survey, only 17 (3%) reported an intake according to RI for women of reproductive age (400 µg/day). Of these, 3 women aged 18–30 years had intakes between 400 and 410 µg/day, and the rest were aged 31–44 years. Among these one ate a fortified bar contributing with 235 µg folic acid per day, one reported an average intake (during the four days) of 23 g of reindeer liver contributing with more than 500 µg folate/day and six reported fruit and vegetables intakes between 600 and 1100 g per day. This indicates how difficult it is to meet RI for women of reproductive age from ordinary foods, although low-energy reporting of dietary intake may add uncertainty. However, estimations of folate intakes are always less certain than, e.g. estimations of more stable micronutrients due to a combination of more complex food analysis and uncertain factors for recipe calculations. Therefore, folate intakes should preferably be evaluated together with biomarker data. Moreover, the use of a web-based self-reporting method could introduce a random error as all participants may be considered as individual coders. In addition, interviewers might be able to retrieve more detailed information on the foods compared with the self-reporting method. On the other hand, using interviewers for coding might introduce systematic errors and furthermore external coding errors were minimised by using the self-assisted web-based dietary record.

Despite the voluntary fortification legalisation in EU, the number of fortified products on the Swedish market is scarce [32]. Some breakfast cereals (estimated to 40% of the market), some bars (about 30%) and some beverages (about 5%) are currently fortified with folic acid in Sweden. However, although only about 20% (n = 398) reported intake of folic acid fortified foods, those consuming such products had significantly higher folate intakes (296 µg/day, Q ₁ = 243, Q ₃ = 355) and erythrocyte folate concentrations (n = 48, 480 nmol/L, Q ₁ = 420, Q ₃ = 600) but not plasma folate concentrations (n = 51, 14 nmol/L, Q ₁ = 11, Q ₃ = 23). Due to the low consumption frequency of folic acid supplements and limited number of fortified foods, evaluation of DFE intakes resulted in a bimodal distribution making DFEs less relevant for evaluation of intakes in the Swedish population.

The EFSA Panel considered erythrocyte folate concentrations the most reliable biomarker of folate status [17]. This is the first study presenting national data on erythrocyte folate concentration in Sweden. Erythrocyte folate concentration (median 460 nmol/L) was low compared to other countries without mandatory folic acid fortification. For example, in Denmark median erythrocyte folate concentrations in pregnant women was 840 nmol/L (week 18, n = 404); however, one third of these women reported daily consumption of a supplement containing 100 µg folic acid [14]. In an Irish population not consuming folic acid fortified foods or supplements median erythrocyte folate concentration was 699 nmol/L (n = 200) [15]. On the other hand, one should be cautious comparing data on folate status since the analytical methods, sampling and storage procedures might affect the results. The microbiological assay, as used in the Irish study, has been reported to result in higher results than immunoassays. For example, in NHANES the previously used Bio-Rad radioassay yielded 29% lower results for serum folate concentrations [33] and 45% lower results for erythrocyte folate concentrations [34] than the microbiological assay.

Median plasma folate concentration of 14 nmol/L in this study was rather high compared to previously reported plasma folate concentrations in Sweden ranging from 7.6 nmol/L up to 14.4 nmol/L (Table 4). Using the same analytical laboratory, Northern Europe (Sweden and Denmark) was found having significantly lower median plasma folate concentrations (10.7 nmol/L) than Central (13.9 nmol/L) and Southern Europe (13.7 nmol/L) [35]. In that study, plasma from the Malmö Diet and Cancer cohort was used [29], which is substantially higher than, e.g. from Northern Sweden [8] (Table 4) indicating that the difference might be even larger.

The prevalence of lower blood folate concentrations in the present study (4% using erythrocyte folate <340 nmol/L and 21% using plasma folate concentrations <10 nmol/L) was nearly identical with the pre-fortification prevalence in the USA (erythrocyte folate 3.5% and serum folate 24%) [36]. Folic acid supplement intake in US pre-fortification was 34% compared to 1% reporting folic acid supplement intake in this study (Table 1) and between 1% [5] and up to 32% [37] in Swedish cohort studies.

Prevalence rate of neural tube defects (NTD) in Sweden decreased significantly from 0.1% in 1999 to 0.08% in 2013 [38]. This might partly be explained by less than 1% reporting consumption of folic acid supplements during early pregnancy in 1999 compared to 15% in 2012 [38]. Our study was not designed to evaluate folate status in respect of risk of NTD. However, folate status during the first weeks of pregnancy is crucial for prevention of NTDs; hence, folate status among women in reproductive age was evaluated separately.

None of the women in reproductive age in this study had an erythrocyte folate concentration as recommended by WHO [24] to achieve optimal prevention against neural tube defects (>906 nmol/L). Median erythrocyte folate concentrations among women aged 18–45 years were 580 nmol/L (Q ₁ = 470 nmol/L; Q ₃ = 710 nmol/L), which according to Daly et al. [25] is associated with a nearly 3 times higher risk of NTDs compared with a status above 906 nmol/L. In countries with mandatory fortification, e.g. Canada median erythrocyte folate concentrations in the first trimester was 1280 nmol/L and only 10% have suboptimal concentrations (<906 nmol/L) [39].

Our results indicate difficulties in achieving an optimal folate status against prevention of NTDs via ordinary foods. Food groups associated with a higher folate status were a higher consumption of vegetables, pulses and roots as well as cheese and alcoholic beverages and lower consumption of meat. Apart from cheese and alcoholic beverages this is in line with the current Swedish dietary guidelines for adults as well as pregnant women. A high alcohol consumption results in intestinal malabsorption of folate, reduced liver uptake and increased urinary folate excretion [40] and is usually negatively associated with folate status, e.g. Pfeiffer et al. [36]. Hence, the positive association between folate status and alcohol intake in this survey probably indicates a food pattern beneficial for folate status. For example, both alcohol and cheese intake positively correlated with vegetable intake (R = 0.08, p < 0.001 for alcohol and R = 0.16, p < 0.0001 for cheese).

In order to achieve an optimal status during the first trimester, folic acid supplements and/or fortified foods appear to be the main choice, even for women consuming a diet rich in fruit and vegetables. However, targeted campaigns focusing on supplement use are proven to be less effective. In a survey in Denmark (n = 462), only 10% of the women reported taking folic acid supplements prior to pregnancy, despite the fact that more than 80% reported knowledge about the Danish recommendation to consume 400 µg folic acid supplements daily periconceptional [41]. In Sweden, only 21% among well-educated pregnant women (58% with college or university degree) reported folic acid supplements prior to pregnancy, despite that three in four had a planned pregnancy [42]. Thus, dietary advice, focusing on the whole diet and certain food groups, is important to facilitate improved folate intake and status. Unpublished results from simulation studies show that a balanced diet based on the NNR and Swedish dietary guidelines may provide amounts of folate in line with RI for women in childbearing ages (Elisabet Amcoff, personal communication).

This study aimed to be representative for the Swedish population. Nonetheless, the participation rate was low for both folate intake estimations (36%) and folate status assessments (30%). Participants had a higher education level than the average population, and this might have resulted in slightly higher estimated folate intakes (Table 2), which should be considered when interpreting the results. Neither education level nor income had an effect on folate status in this study (Table 3). However, the low participation rate in folate status assessment could have introduced bias. The participation rate was particularly low among men aged 18–44 years (participation rate between 23 and 29%) and immigrants (participation rate 27%) making the results less representable for these groups. For immigrants in Sweden, data on folate intake and status are limited. In a case–control study, the prevalence of high folate concentrations (defined as plasma folate concentrations >14 nmol/L) was significantly higher among pregnant women in Sweden born in a non-Nordic country compared to Nordic countries [28].