Table 3.
Characteristics of the thirteen validation studies for nine dish-based dietary assessment tools
| Statistical tests | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Tool No. | Tool type | Study No. | First author (year) | Participant characteristics; number (female %); age (years), mean (sd) or range | Reference method | No. of times tools administered | Intakes used for analysis | Paired t test/Mann–Whitney U/Wilcoxon signed-rank test | The median (range) of correlation coefficients: Pearson’s (P), Spearman’s (S) | Cross-classification (CC)/κ | Bland–Altman analysis | Quality level* |
| 2 | FFQ | 2 | Park (2012)(29) | Apparently healthy Korean adults; 288 (60·1 %); age 44·7 (9·4), range 30–66 | Four seasonal 3-d DR | 2 (interval: 9 months) | Eleven food groups, energy, fifteen nutrients | (Mean) Food groups: all groups significantly differed Energy: not differed Nutrients: nine significantly differed |
Food group S (crude): 0·29 (0·15–0·72) S (energy-adjusted): 0·25 (0·15–0·70) Energy S (crude): 0·40 Nutrient S (crude): 0·31 (0·20–0·42) S (energy-adjusted): 0·29 (0·10–0·56) |
CC quintile: Food group: exact agreement: 22–47 %; gross misclassification: 1–5 % Energy: exact agreement: 31 %; gross misclassification: 2 % Nutrient: exact agreement: 18–32 %; gross misclassification: 2–5 % |
NR | Very good |
| 3 | FFQ | 3 | Kim (2015)(30) | Healthy Korean adults in the metropolitan area; 126 (50 %); age 42·7 (13·1), range 20–65 | Four seasonal 3-d DR | 2 (interval: 9 months) | Energy, thirteen nutrients | NR | Energy P (crude): 0·43 Nutrient P (crude): 0·37 (0·27–0·45) P (energy-adjusted and deattenuated): 0·38 (0·15–0·64) |
CC quartile: Energy: exact agreement: 35 %; gross misclassification: 5 % Nutrient: exact agreement: 22–43 %; gross misclassification: 1–10 % |
The narrowest LOA was found for carbohydrate, and the widest LOA was found for vitamin C. Carbohydrate and vitamin A showed proportional bias | Very good |
| 4 | FFQ | 4 | Yum (2016)(11) | Korean adolescents; 153 (chosen from 160 subjects with 50 % female); age range 12–18 | 8-d DR | 2 (interval: 3–4 weeks) | Energy, fifteen nutrients | NR | Energy P (crude): 0·83 P (deattenuated): 0·91 S (crude): 0·82 S (deattenuated): 0·90 Nutrient P (crude): 0·39 (0·10–0·71) P (deattenuated): 0·44 (0·13–0·79) S (crude): 0·36 (0·10–0·71) S (deattenuated): 0·41 (0·12–0·78) |
CC quartile: Energy: exact agreement: 57 %; gross misclassification: 1 % Nutrient: exact agreement: 27–48 %; gross misclassification: 1–13 % κ (range): Energy: 0·62 Nutrient: 0·07–0·48 |
(Protein, fat, vitamin A and β-carotene only) the narrowest limits of agreement were found for protein and fat and the widest for vitamin A and β-carotene | Very good |
| 6 | FFQ | 6 | Lin (2017)(7) | Rural Bangladeshi children and adults from forty-seven families;190 (54·2 %);age 31·3 (14·7) | Two 3-d DR† (summer and winter) | 1 | Seven food groups, energy, twenty-nine nutrients | (Mean) Food group: five significantly differed Energy: significantly differed Nutrient: twenty-four significantly differed |
Food group P (crude): 0·42 (0·16–0·75) P (energy-adjusted): 0·42 (0·21–0·85) P (deattenuated): 0·53 (0·25–0·90) Energy S(crude): 0·35 Nutrient P (crude): 0·31 (0·08–0·38) P (energy-adjusted): 0·39 (0·14–0·54) P (deattenuated): 0·54 (0·18–0·87) |
CC quintile: Food group: exact agreement: 24–37 %; gross misclassification: 3–11 % Energy: exact agreement: NR; gross misclassification: 9 % Nutrient: exact agreement: 24–43 %; gross misclassification: 1–10 % κ (range): Food group: 0·07–0·41 Nutrient: 0·08–0·43 |
Most nutrient intakes did not show significant proportional bias | Very good |
| 8 | FFQ | 8A | Whitton (2017)(31) | Chinese, Malay and Indian adults living in Singapore; 161 (50 %); age 44 (14) | Two 24-HR | 2 (interval: 6 months) | Energy, twelve nutrients | NR | Energy P (crude): 0·15 P (deattenuated): 0·04 Nutrient (first FFQ) P (crude): 0·34 (0·04–0·47) P (deattenuated): 0·44 (0·09–0·68) |
NR | NR | Good |
| 8 | FFQ | 8B | Whitton (2017)(31) | Chinese, Malay and Indian adults living in Singapore; 161 (50 %); age 44 (14) | Two fasting blood and overnight urine samples | 2 (interval: 6 months) | Six foods, one nutrient‡ | NR | Food (first FFQ) P (crude): 0·21 (0·11–0·47) P (energy-adjusted): 0·19 (0·14–0·48) P (energy-adjusted and deattenuated): 0·20 (0·15–0·51) Nutrient (first FFQ) P (crude): 0·12 P (energy-adjusted): 0·14 P (energy-adjusted and deattenuated): 0·15 |
NR | NR | Good |
| 9 | FFQ | 9A | Date (1996)(4) | Japanese junior college students in a dietitian course; 67 (95·5 %);age range 19–26 | 56- or 63-d DR | 2 (interval: 1 week) | Energy, fourteen nutrients | NR | Energy P (crude): 0·65 Nutrient P (crude): 0·54 (0·35–0·70) P (energy-adjusted): 0·46 (0·21–0·74) |
NR | NR | Acceptable/reasonable |
| 9 | FFQ | 9B | Kobayashi (2011)(32) | Healthy Japanese children; 48 (female % NR); age range 3–11 | 4-d DR | 2 (interval: 1 month) | Energy, thirty-eight nutrients | (Mean) Energy: significantly differed Nutrient: twenty-seven significantly differed |
Energy P (crude): 0·57 Nutrient P (crude): 0·38 (0·09–0·71) P (energy-adjusted): 0·30 (0·01–0·68) |
NR | The intake of energy and eleven nutrients showed agreement between the two methods | Good |
| 9 | FFQ | 9C | Kobayashi (2011)(32) | Healthy Japanese children; 41 (female % NR); age range 12–16 | 4-d DR | 2 (interval: 1 month) | Energy, thirty-eight nutrients | (Mean) Energy: not differed Nutrient: fourteen significantly differed |
Energy P (crude): 0·31 Nutrient P (crude): 0·24 (–0·13 to 0·45) P (energy-adjusted): 0·29 (–0·01 to 0·63) |
NR | The intake of energy and eleven nutrients showed agreement between the two methods | Good |
| 10 | FFQ | 10A | Kobayashi (2011)(32) | Healthy Japanese children; 48 (female % NR); age range 3–11 | 4-d DR | 2 (interval: 1 month) | Energy, thirty-eight nutrients | (Mean) Energy: not differed Nutrient: ten significantly differed |
Energy P (crude): 0·66 Nutrient P (crude): 0·55 (0·33–0·73) P (energy-adjusted): 0·39 (0·03–0·69) |
NR | The intake of energy and eighteen nutrients showed agreement between the two methods | Good |
| 10 | FFQ | 10B | Kobayashi (2011)(32) | Healthy Japanese children; 41 (female % NR); age range 12–16 | 4-d DR | 2 (interval: 1 month) | Energy, thirty-eight nutrients | (Mean) Energy: significantly differed Nutrient: twenty-two significantly differed |
Energy P (crude): 0·33 Nutrient P (crude): 0·26 (–0·06 to 0·42) P (energy-adjusted): 0·34 (0·15–0·77) |
NR | The intake of energy and eighteen nutrients showed agreement between the two methods | Good |
| 11 | DR | 11 | Matsuzaki (2017)(14) | Japanese registered users of a dietary management website; 163 (100 %); age 39·3 (10·3) | 1-d online DR with photos | 1 | Energy, thirteen nutrients | (Median) Energy: significantly differed Nutrient: six significantly differed |
Energy S(crude): 0·87 Nutrient S (crude): 0·77 (0·59–0·82) S (energy-adjusted): 0·77 (0·49–0·84) |
CC quartile: Energy: exact agreement: 66 %; gross misclassification: 0 % Nutrient: exact agreement: 41–63 %; gross misclassification: 0–4 % κ (range): Energy: 0·70 Nutrient: 0·34–0·64 |
Energy and macronutrients: no obvious systematic errors Vitamins, minerals, dietary fibre: proportional bias |
Good |
| 12 | FFQ | 12 | Lee (1994)(15) | Middle-aged, middle-income Chinese; 74 (100 %); age 40·7 (11·9), range 30–60 | A typical day’s diet recall during the last month | 1 | Energy, twelve nutrients | (Mean) Energy: significantly differed Nutrients: five significantly differed |
Energy P (crude): 0·50 Nutrient P(crude): 0·46 (0·21–0·66) |
CC quartile: Energy: exact agreement: 57 %; gross misclassification: 0 % Nutrient: exact agreement: 33–69 %; gross misclassification: 0–20 % |
NR | Good |
κ, weighted kappa coefficient; DR, dietary record; NR, not reported; LOA, limits of agreement; 24-HR, 24-h dietary recall.
*
Evaluated by a scoring system developed by the EURopean micronutrient RECommendations Aligned Network of Excellence(25). See online supplementary material, Supplemental Table 2 for the score of each tool.
†
Recorded by the female head of the household in charge of food preparation and weighed by research members.
‡
The associations were investigated between urinary isoflavones and soya protein intake, serum carotenoids and fruit and vegetable intake, plasma eicosapentaenoic and DHA and fish and seafood intake, plasma PUFA and polyunsaturated fat intake and plasma odd-chain saturated fatty acid and dairy fat intake.