The most important consequences of zinc deficiency include the adverse fetal effects, as well as increased newborn [7
] and children death rate [3
]. However, the influence on the offspring is not the only effect proven for young women, as in the systematic review of Lomagno et al. [23
], conducted for young and premenopausal women, the better zinc status was indicated as associated with improved emotional and cognitive functioning. Taking it into account, not only pregnant women and those in preconception period would benefit from the control of the zinc intake, but also those who do not intend to be pregnant.
4.1. Food Frequency Questionnaires for Pregnant Women
A number of applied multiple-nutrients food frequency questionnaires include zinc intake assessment, while some of them were designed for a specific group of pregnant women. Such pregnancy questionnaires were validated e.g., in United States of America [24
], China [26
], Brasil [28
], or Japan [29
], but also in European countries—Great Britain [30
], Norway [31
], and Spain [32
It is indicated by authors that the intake during pregnancy changes significantly—it may change within each trimester, as well as it is in general less stable than in the case of non-pregnant women, so a significant day-to-day variation is observed [24
]. It may result from the appetite fluctuation, caused by nausea or vomiting, changing food preferences, as well as energy requirements [27
]. As the diet during pregnancy may change, the season of the year when women is pregnant is also important, as the summer to autumn period is indicated as characterized by the highest seasonal variation, resulting from the fruit and vegetables intake [26
Regarding the above-mentioned, it must be emphasized that a constant monitoring of intake during pregnancy is needed and food frequency questionnaires enabling assessment of multiple nutrients may be a good option. The dietary record may not be a better method than food frequency questionnaire, as Brantsæter et al. [31
] indicated that in their study pregnant women were omitting snacks which were considered as unhealthy, in spite of the fact, that they were asked not to change their typical dietary habits.
However, the indicated questionnaires were designed to assess the intake of many nutrients and, as a result, many questions are included. Consequently, it is indicated, that in the case of validation conducted by Brunst et al. [25
], it took about 20–30 min to complete the interviewer-administrated modified Block98 Food Frequency Questionnaire [33
]. As a result, while only the zinc intake is to be assessed, rather a zinc-specific brief questionnaire should be applied.
4.2. Food Frequency Questionnaires Assessing Zinc Intake
In the review of Serra-Majem et al. [34
] it was indicated, that for the zinc assessment, applied using food frequency questionnaires, the methods analyzing intake, such as records and recalls, had an acceptable correlation with validated food frequency questionnaires. Simultaneously, correlations were increased while weighted dietary records were applied and while the assessment of applied supplementation was included [34
At the same time, the validations conducted using biomarkers often are not so positive. For the electronic semi-quantitative food frequency questionnaire, including 235 items, adapted from the Blue Mountains Eye Study Food Frequency Questionnaire [35
], assessing inter alia zinc intake, it was observed, that between zinc intake and serum zinc concentration, there was no significant trend [36
]. The lack of association between zinc intake assessed using food frequency questionnaire and zinc biomarkers was explained by many authors. Fayet et al. [36
], interpreted it as resulting from the number of other dietary factors that affect zinc absorption [37
]. Simultaneously, Brunst et al. [25
] indicated the fact that serum zinc concentration reflects the recent intake from food products and supplementation, while the intake from questionnaire reflects the typical intake from diet and sometimes also from supplementation. Moreover, Brantsæter et al. [31
], validating food frequency questionnaire for pregnant women in the Norwegian Mother and Child Cohort Study (MoBA) indicated that just a few biological markers are directly related to the nutrients intake, so biomarkers use is limited, mainly due to the high cost [38
]. In the conducted study, the blood samples were not collected, as the aim of the study included validation in comparison with the other method of the dietary intake assessment, but it must be indicated, that in the following analysis, the validation in comparison with biomarkers of the zinc status will be needed.
In the review analyzing assessment methods for zinc intake [34
], the correlation coefficients for the food frequency questionnaires, validated using methods assessing the dietary intake, were analyzed and it was indicated, that the correlation coefficients for zinc depend on the food items number in food frequency questionnaires. It was stated, that the number of food items lower than 100 is associated with a slightly higher values of correlation coefficient than for higher number of items [34
The length of the form and a number of questions may be a crucial issue and, as was revealed by Serra-Majem et al. [34
], the lower number of questions in the form may be not only easier for respondents, but also associated with a higher validity. Regarding the above-mentioned, the constant analysis of questionnaires is necessary, while redundant food items should be excluded to simplify the questionnaire [39
For the zinc, it was observed by Samman et al. [40
], in the study conducted in Australia, that assessed the validity of the short food frequency questionnaire while the various number of food products from the questionnaire were included into calculation. In the indicated study, it was observed, that the number of food product items in the questionnaire may be successfully reduced from 74 to 37 (food products contributing to 80% of the total zinc intake), as the similar results in the validation were still observed [40
4.3. Validations Conducted Using Bland–Altman Index
The Bland–Altman plot is recommended to be used instead of the other methods of assessment, in spite of the fact, that in practice it is applied by researchers less often than the other methods [13
]. In the present study, the assessment of validity and reproducibility of the ZINC-FFQ, conducted using the Bland–Altman plot indicates a positive validation, or borderline positive validation, on the basis of the commonly applied criteria [41
]. In general, a Bland–Altman index of 5% is indicated as a borderline [41
]. Such level of a maximum of 5% was observed for the ZINC-FFQ in the assessment of reproducibility (3.3%), while for the validity, the level of 5.5% was observed for FFQ1, but also 6.7%—for FFQ2. The lower validity level for FFQ2 may be associated with the fact, that FFQ2 was filled in 6 weeks after the FFQ1, while FFQ1 was filled in at the similar time as the 3-day dietary record.
However, other authors validating their brief food frequency questionnaires to assess the zinc intake, often observe even higher levels of Bland–Altman index, but they still interpret the questionnaires as a good methods in the assessment. Samman et al. [40
], analyzing a food frequency questionnaire of a various number of items, administrated during a personal interview, in a group of 22 young women, being validated against a 7-day weighted dietary record, for the number of 37-item, indicated the successful validation. The number of 20 individuals, out of 22, within the limit of agreement, indicated a Bland–Altman index of 9.1%, being both higher than 5% indicated as a borderline in the assessment [41
] and higher than Bland–Altman indexes observed in the validation of ZINC-FFQ in the present study.
Similarly, the Bland–Altman plot was analyzed in the study of Alsufiani et al. [42
], in the assessment of the 64-item food frequency questionnaire adapted from the questionnaire previously validated by Samman et al. [40
], by adjusting food products list to the nutritional habits in the population of Saudi Arabia. The number of 95 individuals, out of the diverse population of 100, within the limit of agreement, both in the assessment of reproducibility and validity, indicated a Bland–Altman index of 5.0%. The indicated Bland–Altman index may be interpreted as a borderline significant, that was confirmed by authors, who stated, that the reasonable validity and a high repeatability was observed [42
In spite of the fact that the validity observed in the present study (especially in the case of FFQ2) was lower than recommended, the reproducibility of 3.3% was higher than for the indicated studies of Samman et al. [40
] and Alsufiani et al. [42
]. As a result, it may be supposed, that ZINC-FFQ may be particularly useful in the repeated measurements, while the researcher may benefit from the high reproducibility.
The observed higher reproducibility than in the mentioned study [42
] may result from the different procedure chosen during designing the questionnaire. In the study of Samman et al. [40
] food items of a zinc content no lower than 0.5 mg/100 g were chosen and afterwards, during analysis, they removed food product items, until they obtained 37 items contributing 80% of total zinc intake in the assessed group. Alsufiani et al. [42
] adapted the questionnaire previously validated by Samman et al. [40
], so the procedure was similar. At the same time, in the present study of the ZINC-FFQ, food items characterized by zinc content of 0.01 mg per 100 g or higher were chosen and afterwards they were clustered into 46 food product items. Such procedure may result in a higher accuracy of estimation, as no products were removed from the questionnaire during the designing procedure.
4.4. Validations Conducted Using Other Methods
In spite of the fact, that the Bland–Altman plot is the major method in the validation of questionnaires, it is allowed to use the κ
statistic in the assessment of reproducibility and validity, for a small numbers of ordered categories [13
]. It is very rarely applied by researchers, but in the present study of the ZINC-FFQ, it confirmed the substantial agreement in the case of reproducibility and a slight agreement (values lower than 0.20) in the assessment of validity.
Authors more often use the analyses of the tertiles or quartiles distribution and adequacy assessment. According to the criteria of Masson et al. [43
], when more than 50% of individuals are equally classified, using the compared methods, and less than 10% of individuals are classified into opposite categories, the validity is confirmed. In the present study, the criteria were fulfilled in the assessment of reproducibility of ZINC-FFQ (72.22% correctly classified and 3.33% grossly misclassified), but not in the assessment of validity (43–47% and 16–20% correctly classified and grossly misclassified, respectively, for FFQ1 and FFQ2). It confirms as previously indicated a very good reproducibility, but a minor validity.
The criteria of Masson et al. [43
] were fulfilled in the previously mentioned assessment conducted by Alsufiani et al. [42
], as in the comparison with the 3-day dietary record for zinc intake, 62% of individuals were correctly classified into tertiles and 2% were grossly misclassified, that confirmed a positive validation. The criteria were also fulfilled in the assessment conducted by Heath et al. [44
], while the Meal-Based Intake Assessment Tool (MBIAT) was analyzed (consisting of 630 food products sorted into 16 groups) in healthy male individuals and 60% of individuals were correctly classified into quartiles, while there was no grossly misclassified individuals.
However, similarly as in the case of the Bland–Altman index, in other validation studies are also observed values, that do not fulfill the criteria of Masson et al. [42
]. In the study of Fayet et al. [36
], in the assessment of electronic semi-quantitative food frequency questionnaire, including 235 items, in young female adults, in comparison with the results of zinc intake obtained on the basis of repeated 24-h dietary recall, 26.4% of individuals were correctly classified into quartiles and 11.3% were grossly misclassified, while questionnaire was interpreted as a relatively valid.
The criteria of Masson et al. [42
] are also formulated for the analysis of correlation and it is indicated, that while the correlation coefficient is higher than 0.5, the validity is confirmed. In the present study of ZINC-FFQ, it was analyzed and confirmed in the assessment of reproducibility (p
= 0.0000; R
= 0.7440). Similarly, it was confirmed in the reproducibility assessment of the questionnaire conducted by Heath et al. [44
= 0.64 and R
= 0.73, for various periods between repeated measurements), as well as in the assessment of validity of the questionnaire conducted by Samman et al. [40
< 0.001; R
= 0.81 and p
< 0.001; R
= 0.76 for 74-item and 37-item questionnaire, respectively).
However, similarly as in the previously indicated cases of the Bland–Altman index and analysis of tertiles/quartiles, in other validation studies are also observed values that do not fulfill the criteria of Masson et al. [43
]. It was observed in the study of Alsufiani et al. [42
], that the criteria of Masson et al. [43
] were fulfilled in the assessment of reproducibility (R
= 0.758), but not in the assessment of validity (R
= 0.410), for the questionnaire interpreted as characterized by a high repeatability and a reasonable validity. Similarly, in the study of Lacey [45
], for a Zinc Assessment Tool (ZAT), validated in United States of America, in male and female students, the validity was observed, while compared with the 3-day dietary record, only for female (p
< 0.001; R
= 0.30), but not for male respondents (R
= 0.21), while for both cases correlation coefficient was lower than 0.5.