The elderly population continues to increase in most countries and inadequate nutrition is a common problem affecting their functional and physical status. A variety of physiological, psychological, economic and social changes that may compromise nutritional status accompany aging. Inadequate nutrition is a major problem for elderly people(Reference Kucukerdonmez, Koksal and Rakicioglu1). Most of the tools used to assess nutritional intake in large epidemiological studies were originally developed to be used in young and middle-aged subjects and, therefore, their validity and reliability when employed in older subjects remain uncertain(Reference Bartali, Turrini and Salvini2). Klein et al. (Reference Klein, Kinney and Jeejeebhoy3) concluded in their review of the literature that no gold standard exists for determining nutritional status.
Few studies have assessed the use and validity of dietary assessment methods in elderly people, particularly those classified in the oldest age group (75 years plus)(Reference Dumartheray, Krieg and Cornuz4). For the assessment of average long-term dietary intake in large numbers of individuals, FFQ have emerged as particularly useful tools since they give a better approximation of usual long-term dietary intake than short-term records, can be self-administered and are relatively inexpensive to use. Short-term recalls and diet records are expensive and unrepresentative of usual intake if only a few days are assessed and inappropriate for assessment of past diet(Reference Dumartheray, Krieg and Cornuz4).
In studying health and disease in elderly subjects, information may be needed concerning current or past dietary intake. Regardless of which time period was assessed, the validity of self-reported dietary information probably diminishes with increasing respondent age. In studying diet in elderly subjects, obtaining information from surrogate sources may improve information quality and provide data otherwise unavailable for deceased or incompetent subjects, such as those with memory loss or visual impairment(Reference Samet5).
Research conducted as part of the European Commission's EURopean micronutrient RECommendations Aligned (EURRECA) Network of Excellence has focused on extensive literature reviews addressing the validation of methods used to assess intake of micronutrients, n-3 fatty acids and of special population groups including pregnant women, infants, children, adolescents and elderly people(Reference Ashwell, Lambert and Alles6). In the present review, studies validating dietary methods for assessing micronutrient intake in elderly people are presented.
Materials and methods
A systematic literature search was performed in April 2008. The research question applied to the systematic review was, ‘Which dietary methods are reliable for the assessment of micronutrient intake in elderly people?’. The main stages of the review are illustrated in Fig. 1. The review included English, Spanish, French, Italian, Portuguese and German articles, without limits on time frame or country. Stage 1 of the review involved searching for publications using electronic databases (MEDLINE and EMBASE). The MeSH terms used in the general search were: nutritional assessment; diet; nutritional status; dietary intake; food intake; validity; validation study; reproducibility; replication study; correlation coefficient; correlational study in the title and abstract. A second specific search included the following words: elderly; elder; ‘aged 65 and over’; ‘older people’; ‘dietary assessment’; ‘dietary intake’; ‘nutrition assessment’; ‘diet quality’; reliability; reproducibility; validit*; correlate* as free text in the title and abstract. Additional publications were identified from references published in the original papers. At stage 2 of the review, the title and abstract were analysed by two independent reviewers and the exclusion criteria were applied (Table 1). At stage 3, studies that fulfilled the inclusion criteria were analysed for relevance to the research question.
The selected studies were then classified into three different types according to the reference method applied in the validation studies: (1) reference method assessing intake of < 7 d (including 24-h dietary recall (24HR), estimated dietary records (EDR) and weighed dietary records (WDR)), classified as reflecting short-term intake; (2) reference method assessing intake of ≥ 7 d, reflecting more long-term intake; (3) reference method that employed the use of a biomarker (BM).
Moreover, the different studies included in this review were scored according to a quality score system developed by EURRECA, which has been described in another article in this supplement(Reference Serra-Majem, Frost Andersen and Henriquez-Sánchez7). A total score was calculated according to the mean of the correlation coefficients weighted by the quality score of the validation study. It was considered a poor method for assessing specific nutrient intake when the mean weighted correlation was < 0·30. Methods whose mean weighted correlations were between 0·30 and 0·50 were regarded as acceptable for assessing nutrient intake. Good methods were those whose weighted correlation average were between 0·51 and 0·70, and finally, when the mean weighted correlation was >0·70, the method was considered very good.
Results
A total of thirty-three publications(Reference Dumartheray, Krieg and Cornuz4, Reference Quandt, Vitolins and Smith8–Reference Mahalko, Johnson and Gallagher39) were selected for inclusion, with information on each validation study summarised in Table 2. Sixteen of the publications showed results from European countries (Norway, Denmark, United Kingdom, The Netherlands, Spain, Greece, Switzerland, Sweden, Italy and the Europe wide SENECA study), fourteen from the United States, two from Australia and one study was from New Zealand. The number of participants in the selected studies varied from 37 to 1286.
PSFFQ, picture-sort FFQ; 24HR, 24-h diet recalls; SFFQ, self-administered FFQ; BM, biomarkers; WDR, weighed dietary record; EDR, estimated dietary record; DH, diet history.
In seven of the studies presented(Reference van de Rest, Durga and Verhoef9, Reference Magkos, Manios and Babaroutsi11, Reference Montomoli, Gonnelli and Giacchi16, Reference Tucker, Chen and Vogel20, Reference Jackson, Little and Wilson36–Reference Cummings, Block and McHenry38), only one type of micronutrient was analysed, while in the rest of the publications included in this review, correlations for a wide variety of micronutrients were observed, and in total twenty-one micronutrients were analysed. In eleven of the studies presented(Reference Vioque, Weinbrenner and Asensio10, Reference Messerer, Johansson and Wolk12–Reference Montomoli, Gonnelli and Giacchi16, Reference Tucker, Chen and Vogel20, Reference Klipstein-Grobusch, den Breeijen and Goldbohm21, Reference Mares-Perlman, Klein and Klein28, Reference Nes, Frost Andersen and Solvoll30, Reference Munger, Folsom and Kushi31), dietary supplements were also analysed. Tables 3 and 4 show information on the correlation between methods and other statistics regarding validation studies in elderly people for twelve vitamins and nine minerals, respectively. Table 5 presents the classification of the dietary methods applied to the elderly according to the mean of the correlation coefficients weighted by the quality of different validation studies for each micronutrient included in this review. Nine studies(Reference Quandt, Vitolins and Smith8, Reference Magkos, Manios and Babaroutsi11, Reference Messerer, Johansson and Wolk12, Reference Morris, Tangney and Bienias14, Reference Tangney, Bienias and Evans15, Reference Wengreen, Munger and Wong17, Reference Kumanyika, Tell and Shemanski23, Reference Kumanyika, Tell and Fried24, Reference Munger, Folsom and Kushi31) validated FFQ against 24HR, and a total of twenty micronutrients were evaluated in these correlation studies. These assessment methods showed acceptable correlation for eight micronutrients, good correlation for another eight micronutrients, while four micronutrients (Mg, Se, thiamin and riboflavin) showed a very good correlation. Six different FFQ were validated against EDR(Reference Montomoli, Gonnelli and Giacchi16, Reference Klipstein-Grobusch, den Breeijen and Goldbohm21, Reference Rothenberg27, Reference Horwath29, Reference Potosky, Block and Hartman33, Reference Cummings, Block and McHenry38). In these studies, fifteen micronutrients were analysed, in which acceptable correlations were observed in five micronutrients, whereas another seven micronutrients presented a good rating and three micronutrients (Mg, P and Ca) showed a very good rating (r>0·7). Micronutrients with correlations < 0·3 (poor) were not observed in any of the studies in this category. Additionally, five FFQ(Reference Dumartheray, Krieg and Cornuz4, Reference Flood, Smith and Webb13, Reference Smith, Mitchell and Reay22, Reference Nes, Frost Andersen and Solvoll30, Reference Nelson, Hague and Cooper37) were validated against WDR, and a total of fifteen micronutrients were analysed. These assessment methods showed poor correlations for retinol and Zn, acceptable correlations for five micronutrients, while eight micronutrients showed a good correlation and therefore a good rating. It should be noted that micronutrients with correlations >0·7 (very good) were not observed in any study in this category. In addition, two studies presented results of the validation of FFQ against diet history (DH)(Reference Grootenhuis, Westenbrink and Sie26, Reference Jackson, Little and Wilson36) in which a total of eleven micronutrients were studied. These methods presented acceptable correlations only for vitamin A, good correlations for seven micronutrients, and very good correlations were observed in niacin, Fe and P. Four studies validated FFQ with BM(Reference van de Rest, Durga and Verhoef9, Reference Vioque, Weinbrenner and Asensio10, Reference Tangney, Bienias and Evans15, Reference Tucker, Chen and Vogel20), in which poor correlations were observed for folate and β-carotene, whereas vitamins C and E presented good and acceptable ratings, respectively. On the other hand, four studies validated DH against EDR(Reference Pedersen, Fagt and Ovesen18, Reference van Staveren, Burema and Livingstone25, Reference Mares-Perlman, Klein and Klein28, Reference Mahalko, Johnson and Gallagher39) and twenty-one micronutrients were analysed, of which vitamins A, B12 and β-carotene showed acceptable correlations, eleven presented good correlations and seven micronutrients showed very good ratings. Micronutrients with correlations < 0·3 (poor) were not observed in any study in this category. Likewise, two other studies validated DH against WDR(Reference Nes, van Staveren and Zajkás32, Reference Osler and Schroll34) and a total of eight micronutrients were ranked, five showing good correlations, Ca and vitamin B6 with very good classification and poor correlations being observed for vitamin A. Comparisons of different dietary assessment methods in elderly people by vitamins and minerals are shown in Fig. 2. This figure demonstrates that DH using EDR as the reference method present better correlations for several micronutrients than other combinations of instruments and reference methods in these population groups.
CC, correlation coefficient; SFFQ, self-administered FFQ; PSFFQ, picture-sort FFQ; RE, retinol equivalent; 24HR, 24-h recall; WDR, weighed dietary record; EDR, estimated dietary record; VDAM, videotape dietary assessment method; DH, diet history; BM, biomarker; TE, tocopherol equivalents.
Mean values were significantly different: *P < 0·05; **P < 0·01; ***P < 0·001.
CC, correlation coefficient; 24HR, 24-h recall; PSFFQ, Picture-sort FFQ; EDR, estimated dietary record; SFFQ, Self-administered FFQ; DH, diet history; BM, biomarker; VDAM, videotape dietary assessment method; WDR, weighed dietary record.
Mean values were significantly different: *P < 0·05; **P < 0·01; ***P < 0·001.
EDR, estimated dietary record; WDR, weighed dietary record; 24HR, 24-h recall; BM, biomarker; DH, diet history; VDAM, videotape dietary assessment.
Moreover, these studies were classified according to which reference method was used reflecting short-term intake, long-term intake or BM. Seventeen studies were classified in group 1 with a reference method that reflected short-term intake ( < 7 d), in which nine studies applied 24HR(Reference Quandt, Vitolins and Smith8, Reference Magkos, Manios and Babaroutsi11, Reference Morris, Tangney and Bienias14, Reference Tangney, Bienias and Evans15, Reference Wengreen, Munger and Wong17, Reference Kumanyika, Tell and Shemanski23, Reference Kumanyika, Tell and Fried24, Reference Munger, Folsom and Kushi31, Reference Brown, Tharp and Dahlberg-Luby35), three publications used WDR(Reference Dumartheray, Krieg and Cornuz4, Reference Nes, van Staveren and Zajkás32, Reference Osler and Schroll34), another three papers applied EDR(Reference Pedersen, Fagt and Ovesen18, Reference van Staveren, Burema and Livingstone25, Reference Rothenberg27) and finally, two studies utilised a DH(Reference Grootenhuis, Westenbrink and Sie26, Reference Jackson, Little and Wilson36). Likewise, twelve other studies were classified in group 2 where the reference method reflected long-term intake ( ≥ 7 d), in which one study used 24HR(Reference Messerer, Johansson and Wolk12), four applied WDR(Reference Flood, Smith and Webb13, Reference Smith, Mitchell and Reay22, Reference Nes, Frost Andersen and Solvoll30, Reference Nelson, Hague and Cooper37) and another seven used EDR(Reference Montomoli, Gonnelli and Giacchi16, Reference Klipstein-Grobusch, den Breeijen and Goldbohm21, Reference Mares-Perlman, Klein and Klein28, Reference Horwath29, Reference Potosky, Block and Hartman33, Reference Cummings, Block and McHenry38, Reference Mahalko, Johnson and Gallagher39). Finally, in group 3 where dietary methods were validated against BM, five studies were identified(Reference van de Rest, Durga and Verhoef9, Reference Vioque, Weinbrenner and Asensio10, Reference Tangney, Bienias and Evans15, Reference Espeland, Kumanyika and Wilson19, Reference Tucker, Chen and Vogel20). Some articles present validations using more than one instrument(Reference Quandt, Vitolins and Smith8, Reference Tangney, Bienias and Evans15).
The FFQ were the main dietary method applied, which was validated in twenty-five studies(Reference Dumartheray, Krieg and Cornuz4, Reference Quandt, Vitolins and Smith8–Reference Wengreen, Munger and Wong17, Reference Tucker, Chen and Vogel20–Reference Kumanyika, Tell and Fried24, Reference Grootenhuis, Westenbrink and Sie26, Reference Rothenberg27, Reference Horwath29–Reference Munger, Folsom and Kushi31, Reference Potosky, Block and Hartman33, Reference Jackson, Little and Wilson36–Reference Cummings, Block and McHenry38). Accordingly, Fig. 3 shows only FFQ validation studies that assess micronutrient intake in elderly people using as a reference method a short-term or a long-term dietary assessment instrument or BM. Regarding the reference method that reflected short-term intake, very good correlations were observed for thiamin and riboflavin. However, when the reference method used reflected long-term intake, we observed that a greater number of micronutrients presented good correlations. Moreover, micronutrients with correlations < 0·3 (poor) were not observed when the reference method used reflected long-term intake. Finally, BM used as reference methods presented very good correlations for vitamin E. FFQ validation studies assessing micronutrient intake in elderly people, which included or did not include dietary supplements, are presented in Fig. 4. For most of the micronutrients, the correlation improved when the FFQ evaluated supplement use.
Validated dietary methods
In the thirty-three articles included in this review, twenty-five different FFQ had been validated(Reference Dumartheray, Krieg and Cornuz4, Reference Quandt, Vitolins and Smith8–Reference Wengreen, Munger and Wong17, Reference Tucker, Chen and Vogel20–Reference Kumanyika, Tell and Fried24, Reference Grootenhuis, Westenbrink and Sie26, Reference Rothenberg27, Reference Horwath29–Reference Munger, Folsom and Kushi31, Reference Potosky, Block and Hartman33, Reference Jackson, Little and Wilson36–Reference Cummings, Block and McHenry38). A DH was validated in six studies(Reference Pedersen, Fagt and Ovesen18, Reference van Staveren, Burema and Livingstone25, Reference Mares-Perlman, Klein and Klein28, Reference Nes, van Staveren and Zajkás32, Reference Osler and Schroll34, Reference Mahalko, Johnson and Gallagher39), in another study a 24HR was validated(Reference Espeland, Kumanyika and Wilson19) and one study validated a videotape dietary assessment method(Reference Brown, Tharp and Dahlberg-Luby35). All the FFQ were designed to capture usual diet, however, the time period covered ranged from habitual diet in the last 24 h (one study), the last month (one study), the last 3 months (one study), the last 6 months (one study), the last 12 months (twelve studies) or the last 3 years (one study). This information was not specified in fourteen studies. The studies covered a wide range of items (30–224 food items) that were included in the questionnaire. The frequency categories reported ranged from 5 to 11. Twelve studies developed self-administered FFQ to assess dietary intake in elderly people and in another ten studies, the FFQ were completed by an interviewer. Fig. 5 shows mean study quality-weighted correlation coefficients distributed by use of FFQ or DH as the study instrument. For most of the micronutrients, the correlation improved when the DH was the study instrument being validated.
Reference methods used
Seventeen studies were classified in group 1, where the reference method reflected short-term intake, in which nine applied 24HR(Reference Quandt, Vitolins and Smith8, Reference Magkos, Manios and Babaroutsi11, Reference Morris, Tangney and Bienias14, Reference Tangney, Bienias and Evans15, Reference Wengreen, Munger and Wong17, Reference Kumanyika, Tell and Shemanski23, Reference Kumanyika, Tell and Fried24, Reference Munger, Folsom and Kushi31, Reference Brown, Tharp and Dahlberg-Luby35), three used WDR(Reference Dumartheray, Krieg and Cornuz4, Reference Nes, van Staveren and Zajkás32, Reference Osler and Schroll34), three applied EDR(Reference Pedersen, Fagt and Ovesen18, Reference van Staveren, Burema and Livingstone25, Reference Rothenberg27) and two studies administered a DH(Reference Grootenhuis, Westenbrink and Sie26, Reference Jackson, Little and Wilson36) as the reference method. Likewise, twelve other studies were classified in group 2, where the reference method reflected long-term intake (one 24HR(Reference Messerer, Johansson and Wolk12), four WDR(Reference Flood, Smith and Webb13, Reference Smith, Mitchell and Reay22, Reference Nes, Frost Andersen and Solvoll30, Reference Nelson, Hague and Cooper37) and seven EDR(Reference Montomoli, Gonnelli and Giacchi16, Reference Klipstein-Grobusch, den Breeijen and Goldbohm21, Reference Mares-Perlman, Klein and Klein28, Reference Horwath29, Reference Potosky, Block and Hartman33, Reference Cummings, Block and McHenry38, Reference Mahalko, Johnson and Gallagher39)). Dietary records ranging from 3 to 15 d for data collection were used as the reference method in a total of seventeen studies. The number of repeated 24HR ranged from 5 to 14, which were administered in person or by telephone.
Biomarkers
A total of five publications analysed BM, which were used to validate four FFQ(Reference van de Rest, Durga and Verhoef9, Reference Vioque, Weinbrenner and Asensio10, Reference Tangney, Bienias and Evans15, Reference Tucker, Chen and Vogel20), and one 24HR(Reference Espeland, Kumanyika and Wilson19). Some articles presented validation of more than one instrument, of which one study(Reference Tangney, Bienias and Evans15) also validated a 24HR using BM as the reference method. The BM analysed were: concentration of serum and erythrocyte folate; plasma concentration of β-carotene; vitamin C; vitamin E and one study analysed Na and K in urine samples.
Discussion
In this review, thirty-three validation studies(Reference Dumartheray, Krieg and Cornuz4, Reference Quandt, Vitolins and Smith8–Reference Mahalko, Johnson and Gallagher39) are described. The aim of this analysis was to determine the reliability of methods used to measure the usual intake of vitamins and minerals in elderly people and how these were validated. The different studies included in this review were classified according to which reference method was used, those reflecting short-term intake, long-term intake or BM. To rate the different studies, a quality score system was developed by the EURRECA Network. For each micronutrient and reference method category, a mean of the correlations adjusted by the quality of the different validation studies was obtained, and all methods were classified as poor, acceptable, good, or very good. Dietary assessment constitutes a difficult task for epidemiological studies in the elderly population. Limited recall due to fading memory or impairments in sight or attention may require more complicated approaches and as such, result in a higher respondent burden and low participation rates(Reference Klipstein-Grobusch, den Breeijen and Goldbohm21, Reference van Staveren, de Groot and Blauw40).
Short-term intake
Different FFQ were validated for which wide variations in the number of food items were observed (30–224 items). Grootenhuis et al. (Reference Grootenhuis, Westenbrink and Sie26) found that good agreement of mean nutrient intake and high correlation coefficients between the estimates of the self-administered semi-quantitative questionnaire and the DH method, the absence of non-constant bias for most nutrients and the ability of the questionnaire to classify individuals adequately into broad categories, demonstrated an acceptable relative validity. The results of Rothenberg et al. (Reference Rothenberg27) indicated that the FFQ considerably over-estimated vitamin C intake compared with 4-d EDR. However, using 4-d WDR as the reference method, Dumartheray et al. (Reference Dumartheray, Krieg and Cornuz4) demonstrated a good level of nutrient intake estimation by FFQ for the majority of the micronutrients assessed. The validation correlations were similar to another validation study conducted in the elderly(Reference Smith, Mitchell and Reay22). Most correlation coefficients for unadjusted nutrients were lower than the correlation coefficients after adjustment for energy. This demonstrates that the variability of the nutrient consumption is related to energy intake. Moreover, the correlation coefficient is higher for unadjusted nutrients compared with energy-adjusted nutrients when variability of the nutrient depends on systematic errors of over- or under-estimation(Reference Dumartheray, Krieg and Cornuz4).
The correlation coefficients between FFQ and 24HR showed by Quandt et al. (Reference Quandt, Vitolins and Smith8) were slightly below the range observed in other studies of older populations(Reference Morris, Tangney and Bienias14). This might be due to the method of FFQ administration that was conducted by telephone rather than by face-to-face interviews. Wengreen et al. (Reference Wengreen, Munger and Wong17) utilised three 24HR and two picture-sort FFQ to assess the ability of the FFQ to discriminate among individuals' usual dietary intake. Mean intakes estimated from 138-item picture-sort FFQ used by Wengreen et al. (Reference Wengreen, Munger and Wong17) were higher than those reported by Kumanyika et al. (Reference Kumanyika, Tell and Fried24), the latter being a similar study comparing multiple 24HR to picture- sort FFQ.
The correlation coefficients for the dietary history against the 3-d EDR obtained by Pedersen et al. (Reference Pedersen, Fagt and Ovesen18) were generally high except for vitamin A, probably because of large day-to-day variations for this nutrient. The correlations for dietary data are in the same range as those seen for various physiological measures, as intraindividual variability is also present in these types of measures as well(Reference Block and Hartman41). Pearson's correlation coefficients found by Osler & Schroll(Reference Osler and Schroll34) were of the order of 0·61, with the exception of vitamin A, for which the value was considerably lower (r 0·24). When the data were classified into tertiles (low, medium and high intakes), Pearson's correlations observed by Nes et al. (Reference Nes, van Staveren and Zajkás32), with most values falling between 0·5 and 0·8, were comparable with those observed in studies of elderly people by Mahalko et al. (Reference Mahalko, Johnson and Gallagher39), but were slightly higher than those reported by Potosky et al. (Reference Potosky, Block and Hartman33) However, comparisons with other studies are not straightforward since the reference methods vary in the number of days recorded, the independence of the days (consecutive or non-consecutive), between-person variation and how portions are measured (weighed or estimated). Potosky et al. (Reference Potosky, Block and Hartman33) demonstrated that the apparent validity of a quantitative FFQ increases when it is compared with a greater number of cycles of 4-d dietary records.
Utilisation of the videotape method of dietary assessment described by Brown et al. (Reference Brown, Tharp and Dahlberg-Luby35) resulted in the calculation of food amounts that were, on average, 6 % below amounts determined by recall methods. They found a difference of ≤ 10 % between means using the videotape method and measured amounts in recalls. These results provide strong evidence of the utility of the videotape method. The videotape method comes very close to representing actual intake, is reproducible, and may prove to be useful in identifying the effects of dietary composition on the health status of the elderly and other populations having a relatively high prevalence of cognitive or communication impairments. This study suggested that 24HR may be an inappropiate method of dietary assessment for the elderly due to the high percentage of error and misclassification that may occur due to impaired cognition resulting from disease or even medications affecting memory among others.
Long-term intake
Messerer et al. (Reference Messerer, Johansson and Wolk12) assessed the validity of a self-administered FFQ and showed that overall, adding information about dietary supplement use increased the validity of micronutrient estimates by 13 % based on a self-administered FFQ. Furthermore, correct classification of micronutrient estimates into quintiles increased by 14 %. In line with these findings, Mares-Perlman et al. (Reference Mares-Perlman, Klein and Klein28) showed that the inclusion of dietary supplements increased the overall correlation coefficient from 0·52 to 0·71 among older men. Klipstein-Grobusch et al. (Reference Klipstein-Grobusch, den Breeijen and Goldbohm21) evaluated the relative validity of micronutrient intake estimated by a FFQ adapted for dietary assessment in the elderly as compared to 15-d EDR. The correlation coefficients observed in the present study ranged from 0·5 to 0·9 for crude and from 0·4 to 0·8 for adjusted data, indicating relatively good validity and being similar to results of validation studies in which either a FFQ or DH were administered to an elderly population(Reference van Staveren, Burema and Livingstone25–Reference Munger, Folsom and Kushi31, Reference Mahalko, Johnson and Gallagher39). Cummings et al. (Reference Cummings, Block and McHenry38) found that the daily Ca intake of elderly Caucasian women measured by interview-administered FFQ developed by Block and colleagues correlated well (r 0·76) with daily Ca intake calculated from 7-d EDR. This is consistent with the observation of Byers et al. (Reference Byers, Marshall and Fiedler42), who found that measurement of the intake of a specific nutrient by brief food frequency interviews was reasonably highly correlated with measurements made by longer and detailed food frequency interviews. Similar findings were found by Nelson et al. (Reference Nelson, Hague and Cooper37) who showed that the Ca intake FFQ correlated well (r 0·69) with Ca intake measured by 7-d WDR.
Biomarkers
Espeland et al. (Reference Espeland, Kumanyika and Wilson19) found that estimates of Na intake from 24-h dietary recalls were an average of 22 % lower than those from 24-h urine collection. Diet-based estimates of K intake exceeded those from urine assays by 16 % overall. In contrast, Na intake estimated from the urine collections were 7 % lower than intakes estimated from food records. K intakes estimated from urine were 8 % lower than those obtained from food records(Reference Mickelsen, Makdani and Gill43, Reference Schachter, Harper and Radin44).
On the other hand, four FFQ were validated against BM. These assessment methods presented poor correlations for folate and β-carotene, good correlation for vitamin C and very good correlation for vitamin E. The FFQ developed by van de Rest et al. (Reference van de Rest, Durga and Verhoef9) to assess folate intake over the past 3 months in Dutch elderly people showed a weak positive correlation between folate intakes estimated with the FFQ and serum folate concentrations (r 0·14), but not erythrocyte folate (r 0·05). This could be explained by the fact that the serum folate reflects recent intake and the erythrocyte folate reflects long-term intake(Reference Crews, Alink and Andersen45) and in this study, FFQ assessed food intake in the previous 3 months. Dietary intakes of β-carotene estimated by three different FFQ(Reference Vioque, Weinbrenner and Asensio10, Reference Tangney, Bienias and Evans15, Reference Tucker, Chen and Vogel20) were validated against plasma concentrations of this micronutrient. Vioque et al. (Reference Vioque, Weinbrenner and Asensio10) demonstrated that the correlations between usual intake of this micronutrient assessed by FFQ and their plasma concentration changed when the participants were grouped by BMI category. This finding was supported by Tucker et al. (Reference Tucker, Chen and Vogel20), who has shown that the correlations between carotenoid intakes and plasma concentrations improved after adjustment for BMI, plasma cholesterol concentrations and smoking status, particulary for α- and β-carotene, in men but not in women. The study by Tangney et al. (Reference Tangney, Bienias and Evans15) was designed to assess the performance of the modified Harvard FFQ with elderly people from two ethnic groups, differing cognitive abilities and various educational backgrounds from the Chicago Health and Aging Project. Because short-term memory declines with age, the reliability of multiple 24HR among elderly adults is often questioned(Reference Dubois and Boivin46). Multiple 24HR interviews may burden respondents, contributing to low participation rates and compromising general applicability. The FFQ require a more generic representation of intake(Reference Smith47). BM are subjected to error as well(Reference Tangney, Shekelle and Raynor48), but such errors are independent of those inherent in 24HR and FFQ. The Chicago Health and Aging Project correlations for vitamin E were considerably higher than those reported by the Dutch group (r 0·29 and 0·14 for men and women, respectively), in which considerably fewer older adults were represented(Reference Ocké, Bueno-de-Mesquita and Pols49). The Chicago Health and Aging Project correlations for vitamin C were lower than those reported by Jacques et al. (Reference Jacques, Sulsky and Sadowski50) for men and women from 40 to 83 years of age (r 0·44). BM should reflect the amount of nutrient present in the diet, which may vary according to the metabolism of the given nutrient. In the case of vitamin C, serologic response is seen several hours after postprandial spikes occur. For folate, most elderly have low intakes of this nutrient from food sources as well as being on medications, both of which can interfere with the correlation of intake with BM(Reference Potischman51).
Conclusion
The aim of the present review was to determine the reliability of methods used to measure the usual intake of vitamins and minerals in elderly people and to evaluate how these were validated. The estimation of micronutrient intake is often a difficult task and can present extra challenges in elderly people. This population often suffers from diminished functionality and cognitive decline, which may hamper diet assessment and requires tailored approaches to assess dietary intake(Reference de Vries, de Groot and van Staveren52). Declining short-term memory can imply that the 24HR method may be particularly inappropiate, and several investigators have demonstrated that dietary recall ability decreases with age. The very elderly can become easily fatigued and frustrated with long-DH interviews and may take far longer than younger people to complete them, as they are particularly prone to digressions in interview situations(Reference Horwath29, Reference van Staveren, de Groot and Blauw40). When frequency methods are used for assessing micronutrient intake, the inclusion of dietary supplements improves their reliability for most nutrients, with notable differences observed for folate, retinol, vitamins A, D, E and Zn. The FFQ administered to elderly people had a wide range of included food items, varying from 30 to 224. The frequency categories reported were from 5 to 11. Future research to clarify the number of food items and frequency categories that are to be included in the questionnaires needs to be developed for this population group. Comparing FFQ methods used for assessing micronutrient intake with short-term reference methods, very good correlations were observed for thiamin and riboflavin. Nevertheless, a poor correlation was observed for β-carotene. When FFQ using long-term intakes as reference methods are compared, we have observed that a greater number of micronutrients present good correlations. They are also very good for measuring long-term intake of P and Mg. Micronutrients with correlations < 0·3 (poor) were not observed when the reference method used reflected long-term intake. Micronutrient intake correlates better with long-term rather than short-term daily intake. Additionally, BM used as reference methods present very good correlations for vitamin E and poor correlations for folate.
According to this systematic review, when comparing different validation methods, the DH presents better correlations when EDR are used as the reference method. When we analyse the mean of correlation coefficients weighted by study quality and their distribution by FFQ or DH as validated dietary methods, we observed that most of the micronutrients improved the correlation when the DH was used as the study instrument. Mares-Perlman et al. (Reference Mares-Perlman, Klein and Klein28) used a modification of the DH questionnaire developed as part of the Health Habits and History Questionnaire by Block et al. (Reference Block, Hartman and Dresser53) The questionnaire assessing dietary intake had three parts: a FFQ; additional food intake questions; questions about vitamin and mineral supplement intake. DH questionnaires of this type require less time and cost to administer and analyse than multiple food records or multiple 24HR. As such, this dietary assessment method is more economical in large epidemiological studies and also reduces the potential for response bias that can be introduced due to lower response rates associated with more rigorous methods(Reference Mares-Perlman, Klein and Klein28).
Acknowledgements
The studies reported herein have been carried out within the EURRECA Network of Excellence (www.eurreca.org), financially supported by the Commission of the European Communities, specific Research, Technology and Development (RTD) Programme Quality of Life and Management of Living Resources, within the Sixth Framework Programme, contract no. 036196. This report does not necessarily reflect the Commission's views or its future policy in this area. A. O.-A. wrote the first manuscript; A. O.-A., A. S.-V. and J. D.-A. contributed to the planning of the search and analysed the articles included in this review; J. V. and L. G. revised and discussed previous drafts; L. S.-M. contributed to the planning of the search, decided the analysis and presentation of results and created the quality assessment tool of the articles and revised and discussed previous drafts. There has been no conflict of interest. The authors of the present paper would like to thank Dr Margaret Ashwell, Dr Janet Lambert, Dr Adriënne Cavelaars, Dr Olga Souverein and Mrs Sandra Crispim for their technical contribution to this publication. The authors also thank the Health Sciences Library of the University of Las Palmas de Gran Canaria, Lourdes Ribas and Nuria Melián for their contribution in the collection of articles analysed in this review. Special thanks to Joy Ngo, RD for her help in editing the English version of the manuscript.