Validity of a Self-administered Food Frequency Questionnaire Used in the 5-year Follow-up Survey of the JPHC Study Cohort I to Assess Fatty Acid Intake: Comparison with Dietary Records and Serum Phospholipid Level

We compared fatty acid intake estimated from our 138-item food frequency questionnaire (FFQ) with 28-day weighed dietary records among a subgroup of JPHC Study Cohort I (102 men and 113 women), and with the corresponding two serum phospholipid levels (88 men). Spearman rank correlation coefficients between fatty acid intakes estimated from FFQ and intakes estimated from DR were as follows: saturated fatty acid, r=0.61 and r=0.60; monounsaturated fatty acid, r=0.50 and r=0.44; for energy adjusted value and Eicosapentaenoic acid (EPA), r=0.62 and r=0.55; docosahexaenoic acid (DHA), r=0.61 and r=0.50; for percentage of total fatty acid intake in men and women, respectively. Spearman rank correlation coefficients between fatty acid intakes estimated from FFQ and the corresponding serum phospholipid levels (% of total fatty acid) were as follows: EPA, r=0.43 and r=0.59; DHA, r=0.35 and r=0.49; for crude value (g/day) and percentage of total fatty acid intake, respectively. In conclusion, relatively high correlations were observed for SFA, MUFA and marine-origin n-3 polyunsaturated fatty acid, whereas we must take into account the indicator of each fatty acid intake when using the data of fatty acid intake assessed with FFQ for JPHC study.

and chronic disease has hardly been investigated in epidemiological studies.Several reports have shown a potential role for n-3 polyunsaturated acids (PUFA) in reducing the risk of chronic disease such as cancer and cardiovascular disease .1Japanese people consume much more n-3 PUFA derived from fish than Western populations, and there is a greater variation of n-3 PUFA intake in Japanese diet.2Also , Japanese people consume less saturated fatty acids (SFA) than Western populations .3I It is important to investigate the relationship between specific fatty acid intake and health effects in these populations .
Therefore, it is essential to assess fatty acid intakes in a large population.
In the present study, we compared fatty acid intake estimated with our 138-item food frequency questionnaire with 28-day weighed dietary records among a subgroup of JPHC Study Cohort I. We also evaluated whether there was any relation between fatty acid intake and the fatty acid composition of serum phospholipid .

MATERIALS AND METHODS
The study design and subject characteristics have been reported elsewhere in the present study.5 The survey method of dietary records (DR) and the computation method of nutrient intakes from the food frequency questionnaire (FFQ) have also been described elsewhere in this Supplement.6.7

Fatty Acid Intake Calculations
The daily intake of fatty acids was calculated using the fatty acid composition table of Japanese foods.8 The table has missing values for some foods, so we substituted fatty acid composition for the missing foods.9

Serum Phospholipid Fatty Acid
Peripheral venous blood was sampled just before 7-day DR in February and just after those in August 1994 in the Ninohe , Yokote and Saku areas.Blood was collected in the Ishikawa area in February and August 1995 under similar conditions .After leaving the blood for an hour at room temperature to facilitate clotting, the serum was separated by centrifugation .Serum was kept  and women, marine-origin n-3 fatty acid intake was lowest and the ratio of n-3 fatty acid intake to n-6 fatty acid intake was highest in Okinawa similar to the intake indicated by DR.Table 3 shows the mean and median fatty acid composition of serum phospholipid by area.SFA and MUFA compositions were lowest and that of PUFA was highest in Iwate.Marine-origin n-3 fatty acid intake was lowest and the ratio of n-3 fatty acid intake to n-6 fatty acid intake was highest in Okinawa similar to the intake according to both DR and FFQ.
Table 4 shows fatty acid intake levels estimated from DR and FFQ in 4 areas.Both in men and women, SFA and MUFA intakes estimated from FFQ were higher than intake from DR. Marineorigin n-3 fatty acid intake estimated from FFQ was lower than intake from DR in men, although no apparent difference was observed in women.Table 4 also presents the relations between fatty acid intake estimated from FFQ and fatty acid intake estimated from DR.For comparison, intake of fatty acid was expressed in three ways: crude intake (g/day), energy-adjusted value, and percentage of total fatty acid intake.Both in men and women, SFA and MUFA intakes estimated from FFQ were highly correlated with intake from DR when intake was expressed as an energy-adjusted value (r=0.61 and 0.60 for SFA, r=0.50 and r=0.44 for MUFA in men and women, respectively).Of the PUFA intake, marine-origin n-3 fatty acid intake estimated from FFQ was well correlated with intake from DR expressed in all three ways, and was especially higher when intake was expressed as a percentage of total fatty acid (r=0.62 and 0.55 for EPA, r=0.61 and r=0.50 for DHA in men and women, respectively).
Table 5 shows the mean fatty acid intake estimated from DR within quintiles of corresponding fatty acid intake estimated from FFQ.Both in men and women, SFA and marine-origin n-3 fatty acid intake from DR in the highest quintile was significantly higher than the corresponding fatty acid intake in the lowest quintiles.The result of cross-classification of the subjects by quintiles from the fatty acid intake from DR and fatty acid intake from FFQ are shown in Table 6.As for all fatty acid, more than 50% of the subjects were classified in the adjacent quintiles and less than 10% of them were classified in the extreme quintiles.
Table 7 shows the mean and median fatty acid composition of serum phospholipid in 4 areas and the relation between the serum phospholipid fatty acid level and fatty acid intake estimated from both FFQ and DR.The correlations of SFA composition in serum phospholipid with the respective intake estimated both from FFQ and DR were weak, although for MUFA, a significant correlation was observed for DR (r=0.39), when intake was expressed as a percentage of total FA intake.Marine-origin n-3 fatty acid compositions in serum phospholipid were well-correlated with the respective intake both estimated from FFQ and DR, and especially higher when intake was estimated from DR.When estimated with FFQ and DR, the correlations were higher when intake was expressed as a percentage of total FA intake (r=0.59 and 0.76 for EPA, r=0.49 and r=0.50 for DHA in FFQ and DR, respectively).
Table 8 shows the mean serum phospholipid fatty acid level (% of total fatty acid) within quintiles of corresponding fatty acid intake estimated from FFQ for men.Marine-origin n-3 fatty acid composition of serum phospholipid in the highest or next highest quintiles was significantly higher than the corresponding composition of serum phospholipid in the lowest quintiles.The results of cross-classification of the subjects by quintiles from the serum phospholipid fatty acid and fatty acid intake from FFQ are shown in Table 9.More than 70% of the subjects were classified in the adjacent quintiles in terms of the marine-origin n-3 fatty acids.Tables 10-14 show the cumulative percent contributions of the top 20 foods for SFA, MUFA, PUFA, n-3 fatty acid, and n-6 fatty acid, respectively.Dairy and meat products were important contributors of SFA.Vegetable oils were the largest contributor and meat was also an important contributor of MUFA.Although vegetable oils were the largest contributor, lean foods such as rice, miso, and tofu were also important contributors of PUFA and n-6 fatty acids.Even though vegetable oils were the largest contributors, many kinds of fish were also important contributors of n-3 fatty acid.

DISCUSSION
In the present study, we examined the correlation between dietary fatty acid intakes estimated from FFQ and DR, and the serum phospholipid fatty acid level to evaluate the FFQ used in JPHC Study Cohort I. Absolute intakes of fatty acid estimated from FFQ were slightly higher than intakes from DR for SFA and MUFA, although no apparent difference was observed for PUFA; on the contrary, intakes from FFQ were lower than those from DR for marine-origin n-3 PUFA both in either men or women.
The highest correlations were observed between the intakes estimated from FFQ and DR for marine-origin n-3 PUFA both in men and women.It is reasonable to think that marine-origin n-3 PUFA intakes are relatively easy to estimate by FFQ because dietary fish greatly contributed to marine-origin n-3 PUFA such as EPA, docosapentaenoic acid (DPA), and DHA.When intake was expressed as a percentage of total fatty acid intake, the correlation was higher than expressed as the absolute intake (g/day) for the marine-origin n-3 PUFA.In contrast, as for SFA and MUFA, intake expressed as an energy-adjusted value was more closely correlated.This is consistent with observations from other studies, and the observed correlation coefficients in the present study were similar to those of other studies for SFA (0.54-0.72) and MUFA (0.49-0.56).1-15No significant correlation was observed for n-6 PUFA because its contribution was largely by cooking oil.(We did not inquire as to the kind of cooking oil in either the DR or FFQ.)In addition, n-6 PUFA contributed by many kinds of lean foods such as rice, miso, and tofu.Because these lean foods were consumed on a daily basis and might have a small between-person variability, the observed correlation coefficient was likely to attenuate.
Our study revealed a significant correlation between dietary intake of marine-origin n-3 PUFA estimated from both FFQ and DR and the corresponding FA composition of serum phospholipid.The correlation between EPA intake and the corresponding EPA composition of serum phospholipid was much higher estimated from DR than from FFQ.No apparent differences in correlations were observed for other marine-origin n-3 PUFA.The correlation observed for EPA not only estimated from DR but also from FFQ was higher than the correlation coefficient of 0.20-0.57observed in the previous studies for the correlation between dietary intake of EPA and EPA content in biological specimens such as adipose tissue, serum, and plasma."-10The correlations observed for DHA were similar to those reported in several studies (0.42-0.56).1-20Just as with the correlation between the intakes estimated from FFQ and from DR, when intake was expressed as a percentage of total fatty acid intake, the correlation was higher than expressed as the absolute intake (g/day) for the marine-origin n-3 PUFA.This is consistent with other reports.16,18. 20 The previous study reported that the marine-origin n-3 PUFA was less important for the smaller contributors to intake, since they were much less correlated with total fatty acid intake."We also observed a significant correlation between dietary intake of MUFA estimated from DR (expressed as percentage of total fatty acid intake) and the corresponding FA in serum phospholipid.In the previous study, the observed correlation between MUFA intake and those of biomarker was weak, 16-11 possibly because of the endogenous synthesis of MUFA from carbohydrate and the relatively large within-person variability and small between-person variability of oleic acid intake."18 It is expected that olive oil intake, the source of oleic acid, would be small, and the contribution of eggs or dairy product consumption to the MUFA would be relatively high in our study population.
In summary, we observed a relatively high correlation between fatty acid intake estimated from FFQ and intake estimated from DR for SFA and MUFA when expressed as an energy-adjusted value, and for marine-origin n-3 PUFA when expressed as a percentage of total fatty acid intake.We also observed a relatively high correlation between fatty acid intake estimated from FFQ and serum phospholipid fatty acid level for marine-origin n-3 PUFA.As for the marine-origin n-3 PUFA, the correlation was relatively high as expressed in all 3 ways, although when expressed as a percentage of total fatty acid intake the correlation was highest.These results suggested that when using the data of fatty acid intake assessed with FFQ for JPHC study, we should take into account the indicator of the intakes of each kind of fatty acid.         2 Data on subjects in Ishikawa PHC (14-day data) were counted twice for 28-day data. 2 Data on subjects in Ishikawa PHC (14-day data) were counted twice for 28 -day data .2 Data on subjects in Ishikawa PHC (14-day data) were counted twice for 28-day data. 2 Data on subjects in Ishikawa PHC (14 -day data) were counted twice for 28-day data . 1 Food codes and descriptions correspond to those of the Standard Tables of Food Composition, 4th revised edition in Japan by Science and Technology Agency. 2 Data on subjects in Ishikawa PHC (14-day data) were counted twice for 28-day data.

Table 5 .
Fatty acid intakes assessed with DR2 classified with quintiles of the intakes assessed with FFQ' for corresponding fatty acids (g/day) 1 DR, dietary records. 2 FFQ, food frequency questionnaire. 3 TFA, total fatty acid; SFA, saturated fatty acid; MUFA, monounsaturated fatty acid; PUFA, polyunsaturated fatty acid.4 Ratio compared to the lowest quintile.

Table 1 .
Fatty acid intakes (g/dav) assessed with DR' for 28-or 14-days by area

Table 3 .
Serum phospholipid fatty acid level (% of TFA1) in men by area

Table 4 .
Fatty acid intakes (g/day) assessed with DR' for 28-or 14-days and FFQ4 in 4 areas and correlations

Table 6 .
Comparison of FFQ1 with DR2 for fatty acid intake based on joint classification by quintiles (%)

Table 7 .
Serum phospholipid fatty acid level (% of TFA) in 4 areas and correlations with fatty acid intakes assessed with either DR' or FFQ4 in men (n=88)

Table 8 .
Serum phospholipid fatty acid level classified with quintiles of the intakes assessed with FFQ'

Table 9 .
Comparison of FFQ' for fatty acid intake with serum phospholipid fatty acid level based on

Table 10 .
Cumulative % contributionof the top 20 foods for saturated fatty acid assessed by DR 1 Food codes and descriptions correspond to those of the Standard Tables of Food Composition, 4th revised edition in Japan by Science and Technology Agency.

Table 11 .
Cumulative % contributionof the top 20 foods for monounsaturated fatty acid assessed by DR 1 Food codes and descriptions correspond to those of the Standard Tables of Food Composition , 4th revised edition in Japan by Science and Technology Agency .

Table 12 .
Cumulative % contributionof the top 20 foods for polyunsaturated fatty acid assessed by DR 1 Food codes and descriptions correspond to those of the Standard Tables of Food Composition, 4th revised edition in Japan by Science and Technology Agency.

Table 13 .
Cumulative % contributionof the top 20 foods for n-3 polyunsaturated fatty acid assessed by DR 1 Food codes and descriptions correspond to those of the Standard Tables of Food Composition, 4th revised edition in Japan by Science and Technology Agency.

Table 14 .
Cumulative % contributionof the top 20 foods for n-6 polyunsaturated fatty acid assessed by DR