In nutrition research including clinical settings, a short FFQ could be a powerful tool for estimating dietary exposures of interest. By using the present short FFQ focusing on seafood and n-3 supplements, we found a strong correlation (r = 0.73) with the level of marine n-3 fatty acids in erythrocytes. This correlation was weaker when the intake of n-3 supplements was not included in the analysis (r = 0.61), however still strong enough in view of expected correlations when validating a FFQ against a biomarker in blood matrices . Other validation studies of different FFQs against total n-3 fatty acids, EPA or DHA in erythrocytes, plasma or serum reported correlation coefficients in the range of 0.35-0.67 [7, 15, 20, 21]. Similar correlations are reported when weighed records are validated against serum or plasma levels of marine n-3 fatty acids, as reviewed by Øverby et al . The lower correlation between the seafood FFQ and marine n-3 fatty acids in erythrocytes when the supplements were excluded illustrate the importance of including n-3 supplement registration in the FFQ. This was further emphasized by the finding that the users of n-3 supplements had significantly higher levels of marine n-3 in erythrocytes than non-users. The importance of including questions about supplements intakes are in agreement with other studies focusing on dietary assessment methods [22, 23]. At the same time, it is of importance to include seafood in the diet as it provides several important nutrients beyond the marine n-3 fatty acids. Further, increased seafood consumption is in accordance with recommendations of a healthy and balanced diet .
Very few validation studies are designed to compare dietary vitamin D intake with the biomarker 25(OH)D . Although no information on sunlight exposure was collected in the present study, we found a fairly good correlation between the FFQ and serum 25(OH)D. This might be explained by the fact that the present study was conducted in late spring (i.e. April), however adding questions regarding residence, ethnicity and sun exposure (e.g. sun beds, sun seeking holidays, time spent in daylight) are essential in the later revision of the present short FFQ. In absence of sufficient sunlight exposure, dietary vitamin D intakes become important . Further, the seasonal differences in the correlation between vitamin D intake and 25(OH)D are described in several studies. The MoBa study  showed a stronger correlation between vitamin D intake calculated by FFQ or by food record and plasma 25(OH)D concentration (0.45 and 0.51, respectively) during the winter months in comparison with the whole year (0.32 and 0.43). In the study by Youl and Hee-Kyung  frequent fish intake and regular exercise were significantly associated with higher level of 25(OH)D in healthy Korean men. Fish is regarded as one of the major sources of vitamin D in the Korean diet and their study was also conducted during a time period in which solar UV radiation was very low . A different aspect with the present study is that even if mean level of vitamin D was in the range of optimal status, still one fifth of the participants had suboptimal vitamin D-status (25(OH)D < 50 nmol/L), despite the fact that our study included healthy adults reporting to be apparently interested in diet and health. However, this underlines the growing global health concern about vitamin D insufficiency, particularly in Nordic countries during the winter season, and the present study mirrors the results of Holvik et al. .
Most studies use national food composition data when converting intakes of different foods into nutrient intakes. However, in the present short FFQ the frequency of seafood consumption and supplements intakes were converted into points rather than into quantities of intake, partly because the portion size of intake was not reported in all questions and partly because no Norwegian food composition table is currently publicly available for individual values of fatty acids. The Norwegian food composition table gives only the sum of saturated, monounsaturated or polyunsaturated fat in the foods . Nevertheless, it is important to consider that these food databases often includes analyses of nutrients from a limited number of foods, although these foods should be representative of the given food group. Further, the specific nutrient content in foods may vary within the same food group due to different brands and due to seasonal variations [7, 21].
A weakness of the calculation of points is that the same total number of points might reflect a very different amount of seafood intake, e.g. eating seafood for dinner three times or more per week gives four points (Table 1). In comparison, you will also get a total of four points by eating seafood for dinner once a month plus eating seafood as spread one to two times per week. If you have seafood for dinner three times per week, the seafood intake is about 360-450 grams assuming a portion size of 120-150 grams. This intake is considerably higher compared to seafood for dinner once per month (portion size of 120-150 g) together with seafood as spread one to two times per week (portion size of 25-50 g), where the total seafood intake will be in the range of 53-135 g (assuming that the dinner contributes with approximately 28-35 grams plus the intake from spread of 25-100 grams). Even so, the short FFQ was able to adequately identify participants according to their erythrocyte level of marine n-3 fatty acids. The degree of strongly misclassification data meaning being classified into opposite quartile was very small, and 96% of the participants were classified into the same quartile or adjacent quartile by the short FFQ and by the marine n-3 fatty acids. This was similar to that reported in other studies [27–30]. Further, an important finding of the present study was that the correlation between seafood intake using the FFQ covering the whole diet and marine n-3 fatty acids in erythrocytes (0.67) was similar to what we found when we used the short FFQ (0.73). This also applies to the correlation with both FFQs and vitamin D.
Today there is no method available to measure dietary exposure among individuals or groups of individuals without error. Therefore, it is important that a new FFQ is tested for reproducibility and validity . In the present study we have only used an independent validity check of intakes as we have used biomarkers and we have not tested the relative validity by assessing the short FFQ to a reference method (i.e. weighted food records). The advantage of using a biomarker as a reference method is that the errors of the biomarker method are different from the errors of a dietary method. Further, we did not test the reproducibility of the short FFQ. However, the questions in the short FFQ are based on two different Norwegian FFQs covering the whole diet which both have been validated [16, 17]. Since we found a significant correlation between the short FFQ and the whole FFQ (the whole FFQ was filled in six months after the short FFQ), we believe that this strengthen our validation, although this design cannot fully replace a measure of reproducibility.
The marine n-3 fatty acids in adipose tissue or in blood are the most frequently used biomarkers for intake of seafood . Interestingly, we found a larger correlation between the short FFQ and the DPA level than reported in several other studies [28, 29]. The content of DPA in seafood varies in accordance to the fattiness of the species, and the content of DPA in seafood is normally lower than the level of EPA and DHA . Although, the intake of DPA is lower compared with EPA and DHA, the DPA level in erythrocytes is higher than the EPA level. We can't give a definite explanation, however we suggest that this is due to some methodological issues. The method for determination of fatty acids in some studies [21, 26, 28] find low correlation for DPA compared to our study and some other studies using a different method for the determination of fatty acids [16, 17].
In a study by Brantsæter et al , the authors explore several biomarkers for total fish intake and suggest that blood arsenic concentration appears to be a potential biomarker for fish and seafood intake. Arsenic intake, mainly in the form of arsenobetaine, is largely determined by the amount of seafood consumed in areas without arsenic-contaminated drinking water exposure . Examination of results in the present study revealed no correlation between blood arsenic concentration and the short FFQ (data not shown). However, in our study the arsenic concentration was determined in erythrocytes and not in whole blood . On the contrary, our study confirms the finding that urinary iodine excretion (μg/L) is not suitable as biomarker for seafood intake . This is shown in our study as a higher urinary iodine excretion of participants in quartile 2 than in quartile 3 (Table 4) and can be explained by the fact that milk and dairy products are consumed more regularly (i.e. every day) than seafood (i.e. weekly). Important iodine sources in the Norwegian diet are seafood, milk and dairy products and eggs . Worth mentioning is the tendency that the participants with urinary iodine concentration less than 100 μg/L reported lower consumption of dairy products and seafood than participants with urinary iodine concentration above 100 μg/L. Another important aspect is that we have used median iodine concentration from spot samples. To avoid misinterpretation of the data on the iodine status due to variation in the urinary volume, use of estimated 24-hour urinary iodine excretion adjusting for age and gender specific creatinine excretion has been suggested .
The present validation study was conducted among middle-aged adults with a higher consumption of seafood compared to national dietary studies performed previously in Norway [36, 37]. While convenient for feasibility (e.g. recruitment), whether the present study population was fully representative of the Norwegian adult population in terms of socio-demographic variables remains unanswered. When recruiting volunteers for the study we announced for participants to participate in a seafood study. This might have resulted in that non-seafood consumers avoided signing up for the study, as only one participant reported no seafood intake.