Table 3.
Results and outcomes of studies included in a systematic review of the validity of dietary assessment methods used in adults (≥18 years) when compared with the method of doubly labeled water (DLW)A.
| References | Results | Significance of results | LOA | Correlations EI:TEE | Individual level | Group | Overall quality score∧ |
|---|---|---|---|---|---|---|---|
| Andersen et al. (24) Norway | NS difference between mean EI and mean TEE: −229 kcal/d (± 485). Accuracy was not affected by weight or BMI. | Substantial variability in the accuracy of FFQ at the individual level. FFQ can provide a more accurate measure of the mean EI for groups rather than for individuals. |
−1,195 to 717 kcal/d | r = 0.36, p = 0.15 | Under-report: 47%; over-report: 12%. | Under-report: 10% | Positive |
| Arab et al. (25) USA | Difference in mean EI and mean TEE: 223 kcal/d (diet day) and 662 kcal/d (DHQ). Significant difference (p < 0.05) between MPR and FFQ for participants who under-report. | Validity of MPR was superior to that of the FFQ. Ethnicity affects EI accuracy: more under-and-over reporting among whites than blacks, regardless of the method. | Not reported | Diet day: r = 0.45; DHQ: r = 0.33. Correlations improved with each increased day of recall. |
Under-report: 34% (White) vs. 25% (Black) by MPR; 19% (White) vs. 9% (Black) by FFQ. | Under-report: 9% by MPR; 27% by FFQ | Neutral |
| Barnard et al. (26) Australia | Increased misreporting of EI was associated with increased EE but not with age, sex, BMI or body fat. EI significantly different (p = 0.005) between sexes for both DH and FR. NS weight change over study period. |
Highly active participants or those with variable dietary and exercise habits are more likely to misreport EI. | Not reported | DH: r = 0.90; FR: r = 0.79 |
Adequate-report: n = 7 (5 males, 2 females). | Under-report: 47% (female) vs. 1% (male) by DH; 41% (female) vs. 18% (male) by FR | Positive |
| Bathalon et al. (27) USA | EI accuracy affected by dietary assessment method (p < 0.05). Reported EI significantly (p < 0.05) lower in restrained eaters. Significant weight change in both groups: −33 g/d (Unrestrained eaters) and −28 g/d (restrained eaters). | Under-reporting higher in restrained eaters. Reporting accuracy tended to be higher for WFR than for 24 h recall or FFQ. Assessing dietary hunger and restraint may help to identify subjects likely to under-report dietary intake. | Not reported | 24 h recall: r = 0.06, p = 0.66; FFQ: r = 0.06, p = 0.66; WFR: r = 0.13, p = 0.33 |
Not reported | Under-report: 11% (unrestrained eaters) vs. 19% (restrained eaters) by WFR; 18% (unrestrained eaters) vs. 24% (restrained eaters) by 24 h recall; 23% (unrestrained eaters) vs. 26% (restrained eaters) by FFQ | Positive |
| Beasley et al. (28) USA | EI was more highly correlated with TEE among true reporters (within 25% of EI) compared to non-concordant reporters. | Usual intake was correlated with estimated intake and more highly correlated in true reporters compared to non- concordant reporters. | Not reported | r = 0.79 (true reporters) vs. r = 0.54 (non-concordant reporters) | Reported as true and concordant reporters but values not provided | Not reported | Positive |
| Black et al. (29) UK | Difference in mean EI and EE affected by BMI: 0.73 (post-obese participants) vs. 0.89 (non-obese participants). EI accuracy not affected by sex: 0.89 (women) vs. 0.88 (men). | EI under-reported in both sexes. Greater under-reporting for post-obese participants. |
Not reported | r = 0.47, p < 0.001 | Under-report: n = 6. | Under-report: 11% | Positive |
| Black et al. (30) UK | Mean differences were −1.15 (±1.75) MJ/d for weighed records and −0.43 (±2.40) MJ for diet history. EI accuracy not affected by dietary assessment method: 0.89 (WFR) vs. 0.98 (DH). Mean weight change 0.4 kg (±2.2). |
EI under-reported using both methods. Better ranking of individuals by WFR. |
WR: −0.2 to −2.6 MJ/d; DH: 1.1 to −2.0 MJ/d | WFR: r = 0.48 DH: r = 0.11 FFQ: r = 0.45 24 h recall: r = 0.44 7 days record: r = 0.24 |
29% were not classified in the same third of the distribution for energy | Under-report: 2% by DH; 11% by WFR | Positive |
| Blanton et al. (31) USA | NS difference between mean EI and mean TEE for MPR and FR. Under-reporting by 28% for DHQ and FFQ. | MPR is valid for measuring EI at group level. FR is a valid dietary assessment method. FFQ and DHQ underestimates EI compared to DLW. |
24 h recall: −775 to 930 kJ/d; 14 days FR: −1,325 to 346 kJ/d; FFQ: −3,713 to 1,367 kJ/d; DHQ: −3,868 to −1,513 kJ/d |
MPR: r = 0.53, p = 0.02 FR: r = 0.41, p = 0.07 FFQ: r = 0.25, p = 0.29 DHQ: r = 0.15, p = 0.53. |
Not reported | Under-report: 28% by FFQ and DHQ NS under-reporting for MPR and FR |
Positive |
| Boushey et al. (32) USA | NS difference between mean EI and TEE. Under-reporting of 12% for men and 10% for women. NS weight change over study period. |
Image-based mobile FR as accurate as traditional dietary records. Males more likely to under-report than females. |
−1,700 to 700 kcal/d | r = 0.58 (p < 0.01) | 852 kcal/d (men) vs. 444 kcal/d (women). | Over-report: 2% of participants | Positive |
| Champagne et al. (33) USA | NS difference between mean EI and TEE for Dietitians. Non-dietitians significantly (p < 0.05) under-reported EI by 429 kcal/d. | Dietitians reported EI more accurately than non-dietitians. Professional experience and interest in WFR may explain increased accuracy in estimating EI. | Reported in graphical form only | Not reported | Not reported | Under-report: 10% | Positive |
| Christensen et al. (34) Sweden | Significant (P < 0.001) difference between mean EI and mean TEE estimated by both FFQs. | EI under-reported when using the FFQs. validity of WFR superior to that of FFQs. | WFR: −5,800 to 2,246 kJ/d; Mini FFQ: −9,200 to 1,092 kJ/d; FFQ: −8,500 to 1,569 kJ/d | Normal FFQ: r = 0.42, p < 0.01, Mini FFQ: r = 0.38, p < 0.01 |
Not reported | Under-report: 30% by normal FFQ; 36% by mini FFQ; 17% by WFR | Positive |
| Farooqi et al. (35) Sweden | Significant (P < 0.001) under-reporting between mean EI and mean TEE using DH (28%) and FR (20%). EI accuracy was affected by BMI for DH (r = −0.47) and FR (r = −0.50). NS weight change over study period. |
Both DH and FR result in under-reporting of EI in COPD participants. Greater under-reporting by DH than FR. | DH: 5,000 to −500 kJ/d; FR: 5,000 to −1,900kJ/d |
DH: r = −0.05, p = 0.85, FD: r = 0.19, p = 0.45 |
More women were valid reporters based on the FR than on DH | Under-report: 28% by DH; 20% by FR | Neutral |
| Fassini et al. (36) Brazil | NS weight change over study period. | EI under-reporting more prevalent in control group, over-reporting more prevalent in clinical SBS group. | SBS group: −10.3 to 3.9 MJ/d Control: −1.3 to 6.9 MJ/d |
SBS group: r2 = 0.64 Control group: r2 = 0.01 |
Not reported | Under-report: 2.9 MJ/d (control); Over-report: 3.2 MJ/d (SBS group) |
Neutral |
| Ferriolli et al. (37) Brazil | Under-reporting of EI highly prevalent. Difference in mean EI and TEE: −17.7%. | EI under-reporting highly prevalent in urban-living Brazilians age 60–75. | Not reported | Not reported | Not reported | Under-report: 13% (female) vs. 20% (male) | Positive |
| Freedman et al. (38) USA | EI under-reported when using FFQ and MPR. | Less under-reporting by MPR than FFQ. | Not reported | Not reported | Not reported | Under-report: 12% (females) vs. 8% (males) | Neutral |
| Gemming et al. (39) New Zealand | MPR+SC reduced under-reporting by 6% (women) to 8% (men) compared with the MPR alone (P < 0.001). The increase in EI was largely from snack foods. NS weight change over study period. | Use of wearable camera significantly reduced under-reporting for both females and males as compared to MPR only. | Not reported | MPR: r = 0.68 (men) vs. 0.82 (women); MPR + SC: r = 0.61 (men) vs. r = 0.81 (women). | Not reported | Under-report: 13% (females) vs. 17% (male) by MPR; 7% (female) vs. 9% (male) by MPRc | Positive |
| Hagfors et al. (40) Sweden | NS difference between mean EI and TEE in both Mediterranean-type diet group and control. NS weight change over study period. | DH useful for estimating EI and DH not biased by dietary interventions. | Not reported | Not reported | Under-report: n = 3 | Under-report: 1% | Positive |
| Hise et al. (41) USA | Mean EI represented 99% (±18%) of TEE. NS difference between EI and TEE for both sexes, however, females slightly under-reported (3%) and males over-reported (3%). NS weight change over study period. | WFR + 24 h recall is a valid method for measuring EI in a group of overweight and obese individuals but caution should be taken when using it on an individual level. | −1,109 to 1,063 kcal/d | r = 0.71 | Women: 38 to 398%; men: 30 to 44% | Under-report: 3% female. Over-report: 3% for males | Positive |
| Howat et al. (42) USA | NS difference between EI estimated by FR and MPR, both under-reported compared with TEE. Training made no difference in validity or reliability but help improve portion size estimates. NS weight change over study period. | FR and MPR may reliable methods, however, are likely to under-report EI. Training may help improve portion size estimates. | Not reported | Not reported | Not reported | Under-report: 21.4% | Positive |
| Hutchesson et al. (43) Australia | Difference between mean EI and TEE: −2,301 kJ/d. NS weight change over study period. | EI under-reported by overweight and obese females when using web-based FR. | −1,267 to 169 kcal/d | Not reported | Under-report: 44%; over-report: n = 0. | Under-report: 20% | Positive |
| Johnson et al. (44) USA | EI misreporting negatively associated with BMI: r = −0.36, p < 0.05. NS weight change over study period. | EI under-reported when using MPR at group level. Overweight and obese females are more likely to under-report. | Not reported | 24 h recall: r = 0.22, p < 0.20 | Under-report: n = 12; over-report: n = 1; adequate-report: n = 22. | Under-report: 17% | Positive |
| Kaczkowski et al. (45) Canada | TEE was significantly (p < 0.01) higher than reported EI in each age group. NS difference in reporting accuracy among age groups. NS weight change over study period. | EI under-reported when using multimedia FR. | Not reported | Not reported | Not reported | Under-report: 24% | Positive |
| Koebnick et al. (46) Germany | Mean EI was both over-and-under reported compared with TEE (−49 to 34%). Negative association between EI accuracy and BMI: r = −0.39, p = 0.04. Accuracy was not affected by sex. | EI tends to be UR when using FR. FR is more useful for estimating EI on a group level than an individual level. | 3.5 to −6.4 MJ/d | r = 0.69, p < 0.01 | Under-report: 4%; Over-report: 21% |
Under-report: −1.7 ± 2.6 MJ/d | Positive |
| Koehler et al. (47) Germany | NS difference between mean EI and mean TEE. Significant (p < 0.01) proportional bias toward under-reporting in those with high EI | FR useful for estimating EI on a group level but not an individual level. | −1,371 to 1,174 kcal/d | r = 0.69, p < 0.05 (after removal of implausible reporters) | Not reported | Under-report: 98 kcal/d | Positive |
| Kroke et al. (48) Germany | Difference between mean EI and mean TEE for both methods were strongly and highly significantly correlated r = 0.74, p < 0.001. EI accuracy affected by BMI: r = 0.50, p = 0.007. NS weight change over study period. | EI under-reported by both FFQ and 24 h recall. Possible relation between under-reporting and obesity. | −1,673 to 478 kcal/d | FFQ: r = 0.48. No p-value reported | Not reported | Under-report: 22% | Neutral |
| Lins et al. (49) Brazil | NS difference between mean EI and mean TEE when using FFQ (p = 0.89). Significantly (p = 0.03) higher number of under-reporters in FFQ (n = 24) than 24 h recall (n = 13). Higher % body fat associated with over-reporting in FFQ but not for 24 h recall. | FFQ useful for estimating EI for groups but lack of precision for individuals. FFQ more useful than 24 h recall estimating EI in low-income populations. | 24 h recall: 870 to −1,545 kcal/d; FFQ: 1,500 to −1,888 kcal/d. | Not reported | Under-report: 20% by 24 h recall; 36% by FFQ. Over-report: 5% by 24 h recall; 33% by FFQ. |
Under-report: 13% | Positive |
| Lissner et al. (50) USA | In both obese and non-obese men and women, MPR was more accurate in determining EI. However, both methods under-reported. | Validity of MPR tends to be lower in the group with obesity. No significant difference in validity between obese and non-obese groups for FFQ. | Not reported | 24 h recall: r = 0.39 (non-obese); vs. r = 0.16 (obese) (p < 0.01) FFQ: r = 0.17 (non obese) vs. r = 0.08 obese (p = 0.23) | Not reported | Under-report: 7% (non-obese males) vs. 16% (males with obesity) by MPR; 8% (non-obese females) vs. 20% (females with obesity) by MPR. 24% (non-obese males) vs. 31% (males with obesity) by FFQ; 25% (non-obese females) vs. 29% (females with obesity) by FFQ. | Neutral |
| Livingstone et al. (51) Ireland | When split into thirds of EI, the EI ratio of EI: TEE in the upper third was close to 1.0 with [mean (SE) 0.96 ± 0.08 for females and 1.01 ± 0.11 for males (NS)]. Participants in middle and lower thirds of EI significantly under-reported. | Overall, EI was under-reported when using WFR but could be useful for estimating EI in participants with higher EI's. | Not reported | Not reported | 19 individuals considered accurate reporters (±2SD) | Under-report: 18% (females) vs. 19% (males) | Neutral |
| Lof et al. (52) Sweden | Significant correlation between EI:TEE and BMI (r = −0.352, p < 0.05). NS weight change over study period. | EI under-reported when using 24 h recall in females aged 21–41 y. Females with higher BMI have a higher tendency to under-report EI. | Not reported | Not reported | Under-report: n = 18 | Under-report: 22% | Positive |
| Lopes et al. (53) Brazil | Sex affected reporting accuracy for MPR: more females (29%) under-reporting compared with males (6%) (p < 0.05). NS difference between both sexes when using FR. NS difference between EI and TEE for males using both methods or by BMI and age. |
EI under-reported by both methods. Both methods more useful in estimating EI in males than females. | MPR: −2,204 to 439 kcal/d; FR: −2,043 to 516 kcal/d | Not reported | Under-report: 32% by food record; 20% by 24 h recall. | Under-report: 31% (females) vs. 24% (males) | Positive |
| Mahabir et al. (54) USA | Females who were overweight tended to under-report EI more than normal weight females. | EI under-reported by both methods. Greater tendency for females who were overweight to under-report than healthy weight females. | FFQ: 700 kcal to −2,800 kcal/d | Not reported | Not reported | Under-report: 37% by FR; 42% by DHQ | Neutral |
| Martin et al. (55) USA | Customized prompts did not improve accuracy of mean EI compared with mean TEE, under-reporting by 270 kcal ±748 or 8.8%. No relationship to BMI status. | RFPM is a valid method of estimating EI and is not affected by individual's BMI status. | Reported in graphical form only | Not reported | Not reported | Under-report: 34.3% | Positive |
| Martin et al. (56) Canada | EI accuracy was not affect by BMI status. NS weight change over study period. | EI under-reported when using WFR in healthy middle-aged females. | Not reported | r = 0.46, p = 0.01 | Not reported | Under-report: 20.2% | Positive |
| Medin et al. (57) Norway | EI underestimated by both Web FFQ and 24 h recall. NS weight change over the DLW period. | Web FFQ should be used cautiously, however, they seem reasonable for estimating macronutrients and most food groups. | Web FFQ: ± 1.96 | Web FFQ: r = −0.18 24 h recall: r = 0.34 | n = 14 of 29 women were deemed adequate reporters. | Under-report: 6% by Web FFQ; 17% by 24 h recall | Positive |
| McClung et al. (58) USA | There is a higher tendency to over and under-report using FR. NS weight change over the DLW period. | PDA is a valid method of estimating EI in a group. Both PDA and FR are less useful in estimating EI at an individual level. | −1,472 to 1,394 kcal/d | PDA: r = 0.60, p < 0.05 FR: r = 0.45, p > 0.05 | Not reported | Under-report: 8% by FR Over-report: 5% by PDA | Positive |
| Moshfegh et al. (59) USA | Greater under-reporting of EI with higher BMI. | MPR may be useful for estimating EI in normal weight adults but there is a tendency to under-report as BMI increases. | Not reported | r = 0.32 (males) vs. 0.25 (females) P-values not reported | Under-report: 20%; over-report: 5% | Under-report: 12% (female) vs. 10% (male) | Positive |
| Most et al. (60) USA | BMI had a significant effect on EI accuracy (p = 0.02). African American women reported significantly lower EI compared with white females (p = 0.04). Accuracy better when app used on own phone vs. study provided phone. |
SmartIntake—a smartphone application significantly underestimates food intake. | Not reported | Not reported | Not reported | Under-report: 36.6% | Neutral |
| Nybacka et al. (61) Sweden | Accuracy not affected by ethnicity. | EI under-reported by both methods. FR may be more accurate in estimating EI in a group than FFQ. | FR: 3,000 to −7,000 kJ/d FFQ: 5,000 to −9,000 kJ/d |
FR: r = 0.12 (men) vs. 0.33 (women) FFQ: r = 0.17 (men) vs. −0.05 (women). |
Under-report: 40% by FR; 57.5% by FFQ. Over-report: 15% by FR; 5% by FFQ. |
Under-report: 20% by FR; 18% by FFQ | Positive |
| Okubo et al. (62) Japan | EI:TEE ratio was significantly (P < 0.05) lower for males than females. Significant (P < 0.01) mean weight change in males by −23 ± 55 g/d. | EI under-reported when using DHQ (FFQ) for both males and females. | Not reported | Overall: r = 0.35 (P < 0.001) r = 0.34 (men) vs. 0.22 (women). |
Under-report: 58% (male) vs. 32% (female). Over-report: 10% (male) vs. 18% (female) | Under-report: 5–6% (females) vs. 9–16% (males) | Positive |
| Park et al. (63) USA | Average weight change was −0.3 ± 3.7% for men and 0.1 ± 4.4% for women. Under-reporting highest in participants with obesity and highest for FFQ's. | All EI were under-reported when compared to the DLW method. EI from ASA24 were comparable with 4DFR and both provided the best estimates for dietary intakes. | Not reported | Not reported | Under-report: 13- 32% (male) vs. 21–35% (female) by ASA24; 7–24% (male) vs. 15–20% (female) by FR; 32–46% (male) vs. 20–52 (female) by FFQ | Under-report: 15–17% by ASA24s; 18–21% by 4DFRs; 29–34% by FFQs | Positive |
| Persson et al. (64) Sweden | Mean weight change throughout study period was −0.5 ± 1.9. | FR may be useful for estimating EI in geriatric patients. | Not reported | Total: r = 0.81. r = 0.78 (women) vs. r = 0.80 (men). All p < 0.01. | Agreement by tertile only | Over-report: 8% | Neutral |
| Pettitt et al. (65) UK | Significant (p = 0.04) difference between mean EI estimated by FR and FRMC. NS weight change over study period. | FR with camera provides a more accurate estimation of EI than FR, however, both EI was under-reported by both methods. | 14 days FR: 750 to −4,900 kJ/d 2 d FR + camera: 7,800 to 0 kJ/d 2d FR + 2d camera: 0 to −1,100 kJ/d | Not reported | Not reported | Under-report: 34% by FR, 30% by FRMC | Neutral |
| Pfrimer et al. (66) Brazil | Significant difference between mean EI and mean TEE for different body fatness in females but not in males. | EI under-reported for FFQ and MPR. Females had greater tendency to under-report in both methods. Higher body fatness associated with higher rates of under-reporting, especially for females. | MPR: 479.8 to −971.5 kcal/d; FFQ: 1,303.4 to −1,891.3 kcal/d | FFQ r = 0.19, p = 0.22, 24 h recall: r = 0.25, p = 0.11 | Under-report: 31% by 24 h recall; 4.5% by FFQ | Under-report: 15.2% (female) vs. 7% (male) by FFQ; 27% (female) vs. 14.2% (male) by MPR | Positive |
| Ptomey et al. (67) USA | NS difference between mean EI and mean TEE. NS weight change over study period. | DP + R may be useful for estimating EI in overweight and obese young adults. | −1,606 to 1,667 kcal/d for females; −1,266 to 1,460 kcal/d for males | Not reported | Within 10% of the TEE in 35% of participants (31% of men vs. 29% of women). | Over-report: 6.8% | Positive |
| Rafamantanantsoa et al. (68) Japan | NS correlation between the difference between mean EI and mean TEE and BMI only physical activity. NS weight change over study period. | High intensity physical activity and body composition are important predictors of TEE. | −1,069 to 725 kcal/d | Not reported | Not reported | Under-report: 6% | Positive |
| Rollo et al. (69) Australia | Mean EI:TEE ratio was 0.76 ± 0.2 and 0.76 ± 0.17 for NuDAM and WFR, respectively. NS relationship between both NuDAM and WFR. NS weight change over study period. | EI under-reported by both NuDAM and WFR. Validity of both methods are similar. | Not reported | Not reported | Under-report: NuDAM (n = 3) vs. WFR (n = 4). Over-report: n = 0 |
Under-report: 24% by NuDAM and WFR. | Positive |
| Rothenberg et al. (70) Sweden | Mean EI:TEE ratio: 0.88 ± 0.22. | DH appears to underestimate EI by 12%. | Not reported | r = 0.27 (p > 0.05) | Under-report: n = 4 Over-report: n = 1 |
Under-report: 12% | Positive |
| Sagayama et al. (71) Japan | Significant difference between initial and final body weight at 73.0 ± 7.9 kg vs. 73.2 ± 8.2, respectively. | EI underestimated in light and middle weight wrestlers. | Not reported | Not reported | Not reported | Under-report: 17% (light weight wrestlers) vs. 23% (middle weight wrestlers) | Neutral |
| Sawaya et al. (72) USA | EI accuracy not affected by BMI, sex and age. NS weight change during study period. | Most accurate method for younger females was 24 h recall and FFQ (Willett) for older females. Although these methods may be suitable for estimating EI at the group level, none are reliable for at individual level. | Not reported | Willet FFQ: r2 = 0.40, p = 0.05; Block FFQ: r2 = 0.44, p = 0.04. |
Not reported | Under-report: 19% (younger female) vs. 22% (older female) by WFR. No under/over reporting (younger female) vs. 25% (older female) by 24 h recalls. 384 kcal/d by FFQ (Willet); 679 kcal/d FFQ (FHCRC/BLOCK) | Positive |
| Scagliusi et al. (73) Brazil | All three methods showed a lack of concordance with TEE: MPR r2 = 0.02; Food record r2 = 0.03; FFQ r2 = 0.16. Obese participant's under-reported more than normal weight participants for MPR and FR, but not FFQ. Ethnicity was associated with reporting accuracy (p = 0.01). BMI was negatively correlated with reporting accuracy for MPR. | FFQ had higher rate of misreporting compared to food FR and 24 h recall, which show similar rates of under-reporting. Weight status affects reporting accuracy and should be considered in studies that rely on self-reports of food intake in females. | MPR: −1,919 to 830 kcal/d; FR: −1,844 to 688 kcal/d; FFQ: −2,235 to 958 kcal/d |
MPR: r = 0.47 (p < 0.01) FR: r = −0.39; (p < 0.01) FFQ: r = −0.10; (p = 0.42). |
Under-report: n = 16 by MPR; n = 19 by FR; n = 35 by FFQ | Under-report: 21% by MPR; 22% by FR and 24% by FFQ | Positive |
| Schulz et al. (74) USA | There were NS correlations between EI estimates with both methods and measures of body size. | Both FFQs and 24 h recall under-reported but have comparable accuracy in assessing EI in Native American populations. | Not reported | FFQ: r = 0.48, p = 0.03 24 h recall: r = 0.64, p = 0.03 |
Not reported | Under-report: 20% by 24 h recall; 20% by FFQ | Neutral |
| Shook et al. (75) USA | Participants were divided into tertiles based on BMI by sex. The difference between estimated EI and DLW was 520, 527, and 788 kcal/d for each tertile. | EI underestimated by 24 h recall and estimates less accurate with increasing weight status. | Not reported | R2: 0.23 | Not reported | Under-report: 611 kcal/d | Neutral |
| Subar et al. (5) USA | Under-reporting tended to increase with BMI and with increased energy intake. EI accuracy was not affected by age. Over total 3 month study period participants gained weight (1.1 kg for men, 0.5 kg for females). |
Under-reporting of EI is higher with FFQ compared to MPR. Females under-reported EI to a greater extent than males for both methods. | Not reported | 24 h recall: r = 0.39 (women), r = 0.24 (men) FFQ: r = 0.19 (women); r = 0.10 (men). | Under-report: 22% (male) vs. 22% (female) by 24 h recall; 50% (male) vs. 49% (female) by FFQ | Under-report: 12 to 14% (male) vs. 16 to 20% (female) by MPR; 31 o 36% (male) vs. 34 to 38% (female) by FFQ | Positive |
| Svendsen et al. (76) Norway | Accuracy not affected by sex. Mean weight change in all participants 0·1 kg ± 1·0 (range −3.6 to 1.8 kg). | WFR and FFQs UR EI in obese males and females. | Not reported | Not reported | Under-report: 56% by FFQ; 53% by WFR. Over-report: 8% by FFQ; 2% by WFR | Under-report: 14% (male) vs. 21% (female) by FFQ; 28% (male) vs. 31% (female) by WFR | Neutral |
| Svensson et al. (77) Sweden | SDQ under-reporting was significantly (p = 0.02) higher in females with overweight and obesity (43%) vs. normal weight (22%). Significant correlation between SDQ and FFQ EI to TEE difference (r = 0.62; P < 0.001). Greater under-reporting in those with higher TEE values. | Both SDQ and FFQ under-reported EI in pregnant and non-pregnant females to a similar extent. A short SDQ is as accurate as a more extensive FFQ in estimating EI in females on a group level. | SDQ: Non-pregnant females = −2,003 to 362 kcal/d; Pregnant females = −957 to 2,121 kcal/d |
SDQ r = 0.14 FFQ r = −0.05 both NS | Not reported | Under-report: 30% (non-pregnant females) vs. 21% (pregnant females) | Positive |
| Tanskanen et al. (78) Finland | Reported EI of 11.5 ± 3.2 MJ/d was significantly lower than the mean TEE (15.5 ± 1.6 MJ/d): under-reporting of 26 5% (p < 0.001). | Pre-filled food diaries under-reported EI in male military personnel undergoing basic training. | Not reported | r = 0.44 (no p-value) | Not reported | Under-report: 26% | Positive |
| Tran et al. (79) USA | NS difference between EI as estimated by telephone MPR compared to in-person (p = 0.36). NS weight change over the 14 days period 0.2 (−1.6 to 2.8 kg). |
Telephone administered MPR have similar under-reporting as in-person recalls in estimating EI | −811 to 969 kcal | Not reported | Not reported | Under-report: 15% (MPR administered via telephone) vs. 18% (MPR administered in person) | Neutral |
| Weber et al. (80) USA | In both assessments under-reporting was significantly (p = 0.03) higher in obese compared to lean females. Difference between mean EI and mean TEE was 4.6 MJ (obese females) and 3.1 MJ (lean females). | Normal weight and obese females under-reported EI, although the magnitude of under-reporting may be influenced by the database used to assess dietary intake for normal weight females. | NDS: −74 to 1,824 kcal/d; N3: −120 to 1,859 kcal/d |
Not reported | Not reported | Under-report: 23% (lean females) vs. 39% (females with obesity) by N3; 30% (lean females) vs. 38% (females with obesity) by NDS | Positive |
| Yuan et al. (81) USA | ASA24s had lower validity than SFFQ2. SFFQ2 had lower validity than one 7DDR. Averaged 7DDRs had the highest validity. | SFFQ2 provided reasonably valid measurements. The ASA24 needs further evaluation for use in large population studies. |
Not reported | SFFQ2: r = 0.70, 7-day DR: r = 0.63 ASA24: r = 0.28 |
Not reported | Under-report: 15% by SFFQ; 21% by 7-day DR, 17% by ASA24 | Positive |
∧
Study Quality assessed by American Dietetic Association tool.
A
Abbreviations included in above table as defined as follows, LOA, limits of agreement; Y, years; SD, standard deviation; DLW, Doubly labeled water; SFFQ, semiquantitative food frequency questionnaire; FFQ, food frequency questionnaire; BMI, body mass index; MPR, multiple-pass 24 h dietary record; SC, Sensecam; DH, diet history; FR, Food record; WFR, weighed food record; SDQ, short dietary questionniare; PDA, personal digital assistant; NUDAM, Nutricam diet assessment method; RFPM, remote food photography method; DR, dietary record; NS, not significant; EI, energy intake; NDS, Nutrient Data System; N3, Nutritionist III; SBS, Short bowel syndrome; DP + R, digital photographs with dietary recalls; FHCRC, Fred Hutchinson Cancer Research Center.