People with a body mass index ⩾30 under-report their dietary intake: A systematic review

Abstract

Under-reporting of total energy intake is a common and well-known source of measurement error in dietary assessment, and evidence suggests that this bias is particularly significant in obese individuals. After a multi-stage selection process of eligible papers, this literature review appraised 34 papers detailing the accuracy of self-reported dietary intake in people with an obese body mass index (BMI ⩾ 30). The available literature to date shows that having a body mass index ⩾30 is associated with significant under-reporting of food intake. Future research should look into identifying effective techniques to reduce this bias in clinical practice.

Keywords

24-hour dietary recall dietary intake nutrition assessment obesity self-reported energy intake under-reporting

Introduction

Over the past decade, obesity has become one of society’s major health challenges (Morris et al., 2014; Royal College of Physicians, 2013), illustrating an increasing burden on the health system (Selassie et al., 2011). Statistics demonstrate that currently approximately 67 per cent of men and 57 per cent of women in the United Kingdom are either overweight or obese, according to the categories defined by the WHO (World Health Organization) that classify an overweight body mass index (BMI) over 25 and an obese BMI over 30 (Sedghi, 2014). Further trends show that by 2030 there could be as many as 65 million more adults in the United States and 11 million more obese adults in the United Kingdom (Wang et al., 2011). According to recently released research by McKinsey & Company (2014), every year obesity costs the world $2 trillion in economic burden, through both direct medical cost and indirect cost of decreased productivity. Falling slightly behind smoking, obesity brings along a burden of $73 billion or around 3 per cent of the gross domestic product (GDP) in the United Kingdom (Consultancy UK, 2015). Taking these figures into account, obesity is a highly concerning issue in today’s society that calls for urgent and impacting action.

It is plausible that obesity must result from energy imbalance, and it has long been assumed that simple excess intake of kilocalories leads to overweight. However, this has not been demonstrated in the literature; a range of studies have shown that obese persons have repeatedly and consistently reported consuming the same or less energy than their normal weight counterparts (Lincoln et al., 1972; Myers et al., 1988). This is also reflected in clinical practice: dieticians frequently see morbidly obese patients who find it difficult to lose weight with a suspected underlying medical cause; however, the reason for their failure to lose weight often cannot be explained by metabolic imbalances (Buhl et al., 1995). The difficulty in fully understanding what people actually consume is that capturing food intake relies on self-reported food intake; therefore, accuracy can be compromised in a range of measurement tools used (Black et al., 1993; Scagliusi et al., 2008). The main limitation of food records is the under-reporting or misreporting of energy intake (EI) and representativeness of the overall diet (Livingstone et al., 1990; Poslusna et al., 2009). The prevalence of under-reporting in the general population ranges from 18 to 54 per cent but it can be as high as 70 per cent for particular subgroups (Macdiarmid and Blundell, 1998). Although nearly all people tend to underestimate their calorie intake, there is increasing evidence highlighting that the bias is much worse in people with a higher BMI (Lichtman et al., 1992; Livingstone and Black, 2003; Tooze et al., 2004). Assessing the accuracy of self-reports for EI in the obese is therefore important, as measurement error can conceal the effects of food and nutrient intake on health outcomes (Thompson et al., 2008).

In clinical treatments of obese persons, it is often assumed that they intentionally underestimate food intake to improve their self-esteem as a form of self-deception or self-presentation because they want to present themselves in a positive light to others (Muhlheim et al., 1998). Macdiarmid and Blundell (1998) identified different types of under-reporting taking into account the intentional background. These can be summarised as follows:

Food being eaten but deliberately NOT reported (intentional under-reporting);

Food consumption being reduced, or certain foods being avoided, during the period of study (intentional alteration of diet);

Food being eaten but genuinely forgotten (unintentional/ unknowing under-reporting).

Therefore, the intention of under-reporting needs to be distinguished when discussing underlying factors as individuals may have completely different motives that require different strategies to overcome their misreporting bias.

However, according to several studies, a high BMI alone is not a sufficient variable in explaining under-reporting. Additional factors that have been suggested in this context include poor body image and weight consciousness. Heitmann (1993), Crawley and Summerbell (1997) and Lafay et al. (1997) agree that inaccurate reports of food consumption appear to be associated with dieting and dietary restraint, which may clarify the strong link between under-reporting and an obese BMI to some extent.

In order to evaluate the quality of dietary reports, it is important to understand and to differentiate the various methods that are commonly used in clinical practice and research in regard to their methodological quality and limitations. The available tools identified can be divided into three general categories, including (1) recall of foods eaten, (2) diet histories or retrospective questionnaires and (3) diet records. Diet recalls aim to assess recent EI quantitatively and usually involve remembering all foods that have been consumed during the previous 24 hours (Morgan et al., 1978). A frequently reported bias with this method is the lack of representativeness, as it may not adequately report eating behaviour on a habitual level, which is often the intended focus of diet-related studies. On the other hand, diet histories and Food Frequency Questionnaires (FFQs) have been associated with more accurate estimates of usual or habitual dietary patterns; however, this method involves other problems regarding the recall and seasonality of particular foods that are included (Hill and Davies, 2001). Diet records require participants to weigh or quantify the total foods and drinks consumed over a period usually ranging from 3 to 7 days, typically using household measures (Barrett-Connor, 1991; Gibson, 2005). A potential bias with FFQs is the potential effect of social desirability, which may lead to over-reporting of foods perceived as ‘good’, and under-reporting of foods perceived as ‘bad’. There are several methods available to assess the accuracy of self-reported dietary intake, but the ones that are most commonly used across the literature are Goldberg cut-offs and the doubly labelled water (DLW) method, which are both based on the fundamental principle of energy metabolism (Schoeller, 1990). Goldberg cut-offs were developed by Goldberg and his colleagues in 1990. The underlying principle is to test the agreement between PAL (physical activity level) and reported EI in relation to BMR (basal metabolic rate) using a specific cut-off that identifies under-reporters.

The problem with Goldberg cut-off values is that they do not take the wide range of PALs into account (Livingstone et al., 1990, 1992; Westerterp et al., 1986). Therefore, the degree of under-report assessed through the Goldberg method may be underestimated (Hill and Davies, 2001).

Since its development in the 1980s, studies comparing self-reported EI with energy expenditure (EE) have increasingly used the DLW technique. The DLW technique for measuring total energy expenditure (TEE) involves enriching the water within the body with the use of water labelled with a stable isotope of hydrogen and a stable isotope of oxygen. Although the DLW is generally considered as the gold standard of EE (Schoeller, 1990), it is involved with much higher costs than the alternatives which may limit its use. Consequently, although dietary assessments all have their own advantages and limitations, they are all prone to misreporting.

Considering the evident tendency of people with excessive weight to under-report to a higher extent than others, there is scope to explore the impact of an obese bodyweight on reporting accuracy, which has not been done systematically to date.

The purpose of the current review is, therefore, to examine the accuracy of self-reporting EI in people with a BMI ⩾30. It can furthermore provide details about factors which are associated with the extent of under-reporting in this population, such as demographic factors (e.g. age, social class or sex), psychosocial influences (e.g. social desirability, extraversion or disinhibition) or clinical abnormalities (e.g. depression or binge eating disorders). This will afford the opportunity to identify specific risk groups within this BMI category, who are more likely to under-report and therefore require particular attention when evaluating their data of self-reported food intake.

The results of this review can potentially be used to highlight the following:

Overall tendency/extent of under-reporting food intake across obese samples in comparison to non-obese samples;

Identification of valid or non-valid methods for self-reporting EI for people with an obese BMI;

Factors associated with under-reporting and accurate reporting in obese samples;

Recommendations about improving accuracy of dietary self-reports for this population;

Areas where further research may be required.

Materials and methods

Literature search and selection

A search of the literature was conducted via various established databases, including PsychINFO, MEDLINE, CINAHL and Web of Science. A broad search strategy was adopted to capture each self-report method for food intake within the target population (Supplementary Figure 1). In addition, the reference lists of relevant articles identified through the database search and existing systematic reviews were searched manually to identify further suitable studies.

The search was limited to articles published from 1 January 1982 to 18 March 2016. The reason for limiting the search to articles published after January 1980 was that a previous literature review identified that until 1982 it was not possible to determine the validity of dietary assessment procedures before the first objective method, the DLW technique, was developed (Barrie & Coward, 1985; Schoeller and Van Santen, 1982).

Furthermore, any studies that include a substantial number of participants younger than 18 years were excluded. A previous review has highlighted that dietary assessment among children is difficult because they are more susceptible to environmental influences than adults. Research indicates that self-reports by children are prone to considerable errors, which are caused by age-related deficits in attention, perception, organisation, retention, retrieval and response. These factors have a detrimental effect on cognitive processing of food-related information, which is required for accurate recall and organisation of EI (Baranowski and Domel, 1994). This means that we can conclude a systematic difference in recall capacity between adults and children, which can cause bias when interpreting the results.

The search included the following terms, which were specified by finding synonyms using the Thesaurus function on EBSCOHost to capture as many applicable studies as possible.

(high or ⩾ 30) and (bmi or ‘body mass index’) and obes*
AND
Record* or diar* or assess* or measure* or report* or estimate*
AND
reliabl* or accura* or valid* or estimat* or overestimat* or underestimat*
or under-report* or correct*
AND
‘food intake’ or ‘calorie intake’ or diet* or ‘energy intake’
or ‘food consumption’

The search through PsychINFO, MEDLINE and CINAHL was run on the server EBSCOHost and repeatedly run on for Web of Science separately.

The search included four thematic buckets, so the process consisted of multiple steps. First, all buckets were entered into the search bar individually and subsequently merged to guarantee that all terms were combined appropriately. All databanks included in the search reported a total of 3475 papers, excluding duplicates. All eligible titles were screened in the next step. If the title itself did not provide sufficient information in terms of the inclusion criteria, the abstracts were screened simultaneously. After the full texts for the remaining 44 papers were retrieved, these were studied and assessed for eligibility. It was concluded that 11 studies failed to meet the full inclusion criteria due to various reasons. One paper did not report accuracy of the dietary reports; in two papers, under-reporting was not discussed, while the main reason for exclusion was a BMI < 30 (n = 8). While screening the reference list, an additional eligible paper was identified and included, resulting in a total number of 34 papers meeting the full criteria for the present review (Supplementary Figure 1). The final list of included papers was then confirmed by a second independent reviewer.

Inclusion and exclusion criteria

Studies were eligible for inclusion in the review if they met all the criteria below:

Published between 1 January 1982 and 18 March 2016;

Written in English language;

EI is measured via self-report;

Based primarily on adult samples (⩾18 years);

Participants with obese BMI (>30) are explicitly included in the data.

Quality of studies reviewed

The quality of the studies was reviewed using the Cochrane Handbook’s general guidance on non-experimental studies to inform the choice of quality indicators (2 indicating higher quality than 1). The reason for choosing this particular method derives from the nature of the studies that were included in the assessment. Due to the fact that the research question aims to identify the relationship between reporting accuracy and obesity, rather than the impact of manipulating any of the variables, the studies included in this review are not of experimental, but rather observational nature.

Therefore, this method was deemed adequate as its criteria are tailored to the characteristics of this type of study.

The examination criteria included sampling: non-random = 1, random = 2; representativeness: response rates: <60 per cent = 1, 60 per cent or more = 2; population definition: selected sample (e.g. school students) = 1, general population = 2; sample size: <100 = 1, >100 = 2. If any of the required details were not provided sufficiently, the underlying criteria were assumed as being ‘not described’, as this reflects poor reporting quality of elements in the study, rather than poor quality of the study.

To summarise the total quality score for each study, the scores for all four categories were determined and added up their total. The sum score was converted into the percentage numbers as a global indicator.

Once all studies had been reviewed in their methodological quality, the selected articles were collated; their characteristics and results are summarised in Tables 1 and 2 in Appendices 1 and 2. Additionally, the sample sizes were categorised in regard to their risk of bias based on a methodology checklist. Studies with up to 59 participants were rated poor/small, studies of 60–150 were rated adequate/average and studies with over 150 were good/large (SIGN, 2004).

Results

A total of 34 studies met the inclusion criteria and were included in the review. Tables with summaries detailing the relevant characteristics of the studies are outlined in Appendices 1 and 2.

All studies were carried out between 1982 and 2014. The vast majority of studies (79.5%) were based on samples from Europe (n = 6, 47.1%) or North America (n = 11, 32.4%). The sample size varied considerably across the studies, ranging from 9 to 23,289. From the papers identified, one sample was recruited from a nutritional study as a convenience subsample, and several samples were part of a weight loss treatment at the time of data collection (n = 5), while another sample derived from a clinical trial. In regard to the methodological design, a substantial proportion of the studies had a cross-sectional design (n = 25, 73.5%), including one observational study. The remaining designs included two case-control studies, two longitudinal studies, three case-control studies, two cohort studies, including one with a prospective design, one residential study and one covert study. While not all studies provided specific information about the number of obese participants, the sample sizes for this weight category, that were indicated, ranged from 9 to 155. In total, two samples consisted exclusively of obese participants. In regard to demographic characteristics, most studies represent both sex (70.6%), while the remaining focus on women. Two studies included older people (>65 and 70–80 years), and another one studied a middle-aged population. In terms of specific measurement tools, the most frequent tools comprised of diet records ranging from 3 to 9 days (n = 13), the 24-hour recall (n = 12) and the FFQ (n = 9). Four studies utilised food diaries varying between 3 and 7 days, and three samples were asked to complete the dietary history questionnaire. While nearly all studies aimed to measure the overall energy take of participants, one study only looked at salt intake (De Keyzer et al., 2015) and another one paid particular attention to self-reports of protein while still including all dietary sources in the analysis (Mossavar-Rahmani et al., 2013).

In general, researchers were interested in participants’ everyday eating behaviour in their usual environment. Poppitt et al. (1998) observed their participants during communal meals to test how menu selection impacts self-reporting accuracy. The reporting accuracy was frequently measured using Goldberg cut-offs (n = 10), and various authors stated using alternative cut-offs deriving from ratios between EI and EE (n = 13). Less common methods included the DLW technique (n = 4), correlation between recall and actual meals (n = 2), geometric means (n = 1) and Bland–Altman plots (n = 1). Two studies, however, did not specify how they evaluated self-reports (Lansky and Brownell, 1982; Lara et al., 2004).

The quality of the included papers generally ranged between 50 and 100 per cent. The most common result was 63 per cent (n = 11), which was often due to non-random sampling and using specific groups, such as clinical populations or intervention participants. Furthermore, several studies (n = 8) had small sizes that were unlikely to result in adequate power for the statistics applied. However, the majority of study samples were at least average (n = 7) or large (n = 19), which suggests higher generalisability. The results of the identified papers demonstrate a general bias for reporting accuracy, regardless of BMI, sex and other demographic factors. Average rates are generally poor and differ widely between studies, ranging between 11 per cent (Gnardellis et al., 1998), 33 per cent (Lutomski et al., 2011) and 40–50 per cent (Johansson et al., 1998).

Regardless of their weight, three out of eight studies assessing the link between sex and reporting accuracy found that women are more likely to under-report their EI than men (Fereidoun Azizi, 2005; Johansson et al., 1998; Shaneshin et al., 2012; (p < 0.001)). All three studies were consistent in terms of their large sample size and cut-off to define under-reporting. A particularly interesting similarity is that the two most recent studies are based on urban Teheranian samples. They were conducted by a specific Iranian research group who had a specific interest in exploring food reporting accuracy in developing countries, since the majority of previous studies in this field have been carried out in affluent societies (Fereidoun Azizi et al., 2005; Shaneshin et al., 2012). It may be suggested that demographic influences, including country-specific characteristics as well as an urban environment may have had a moderating effect on these two samples. One of the researchers found that under-reporting is associated with attitudes towards people’s body weight, in particular the desire to lose weight, which was higher for women (Johansson et al., 1998). Consistent to this observation, Fereidoun Azizi et al. (2004) hypothesis a higher sensitivity to appearance and fitness for women, which could have, at least to an extent, contributed to the significant gender link. The fact that under-reporters have an overall lower intake of foods deemed unhealthy and high in sugar or fat mirrors this trend, as these types of food are generally avoided by people who are intending to lose weight (Johansson et al., 1998; Shaneshin et al., 2012). However, the sample distribution in regard to body weight suggests that being obese still may have a greater impact on under-reporting, as the majority of female participants were obese in one sample (Fereidoun Azizi et al., 2004).

In contrast to these findings, two additional studies either found slightly higher accuracy rates in women compared to men (Gnardellis et al., 1998) or non-significant differences, which supports a less significant gender link than originally suggested (Vyas et al., 2003). Moreover, the trend for female under-reporting tends to increase with age, as nearly half of older women under-report, while the rates for male subsamples tend to vary between studies in comparison to women (Cook et al., 2000; Johansson et al., 1998). Therefore, it is not surprising that age associated with increased risk factor for under-reporting (Fereidoun Azizi, 2005; Gemming et al., 2014), with the exception of one study (Samaras et al., 1999). Low education is another demographic factor that has been associated with food under-reporting independently from body weight (Gnardellis et al., 1998).

According to the present findings, the majority of papers clearly suggest that there is a consistent and clear link between under-reporting of food intake and an obese BMI in a considerable amount of included papers (n = 22). In contrast to this, Garriguet (2008), Westerterp-Platenga et al. (1996) and González et al. (2000) found that obese people do not differ significantly from other weight categories. Interestingly, one particular study found that the self-reporting of salt intake was even more accurate in obese participants compared to their normal weight counterparts (De Keyzer et al., 2015). One team of researchers interestingly concluded that obese people in a clinical setting overall were more likely to under-report their food intake than to over-report (Lara et al., 2004). In regard to the previously identified link between accuracy and sex, Cook et al. (2000) go as far as defining obesity as the sole cause of under-reporting in women. However, other findings suggest that under-reporting cannot be identified by just taking into account people’s BMI, as it only accounts for 6 per cent of changes in under-reporting. This observation is reflected by several studies that deemed under-reporting common in people with normal weight (Johansson et al., 1998).

Across the findings, several factors were found to be associated with under-reporting in obese people; hence, they may contribute to explaining why it is more common for this BMI category. In regard to demographic characteristics, higher education turns out to be one possible explanation (Abbot et al., 2008). Moreover, certain psychological factors tend to occur to a greater extent in obese people who under-report, mostly views and attitudes towards their own body. Specific drivers include desire to lose weight (Johansson et al., 1998), less realistic weight loss goals, body shape concerns (Abbot et al., 2008), depression (Kretsch et al., 1999) and cognitive restraint (Lichtman et al., 1992). One study compared obese individual’s reporting accuracy to people with anorexia, which may have led to higher visible difference than comparing this group to a population with a healthy BMI (Schebendach et al, 2012). Furthermore, dietary or eating habits add further clarity to the link between obesity and under-reporting. These include a higher consumption of foods with a high dietary glycaemic load (Murakami et al., 2013), and lower intake of bread, sweets, desserts and snacks (Svendsen and Tonstad, 2006).

Interestingly, under-reporting is particularly prevalent in obese and older individuals with high PALs (Meng et al., 2013). However, this link applies to over-reporters with an obese BMI to a similar extent (Pietiläinen et al., 2010). In the context of exercise, Abbot et al. (2008) emphasise that certain psychosocial influences lead to under-reporting, including higher perceived exercise competence and having access to social support for exercising.

Furthermore, overweight and obese people with a binge eating disorder are slightly less accurate in self-reporting dietary intake than others, however not significantly. It is suggested that processes during binge eating episodes impair people’s ability to remember their food consumption correctly. It is assumed that the commonly experienced less of control combined with increased EI during binge eating episodes in combination and reduced awareness of food consumption lead to under-reporting (Bartholome et al., 2013).

Considering that despite the high number of studies being included in the review no meta-analysis was performed, it needs to be mentioned that there were several reasons leading to the decision against one. First, there were significant differences regarding the quality of the studies, which generally ranged from 50 to 100 per cent. Based on the information provided by the authors, the most common weakness of several studies was failure to randomise samples (n = 11). Furthermore, there was substantial heterogeneity in terms of sample characteristics: the sample sizes of obese participants varied greatly from n = 9 to n = 155, while other studies (n = 8) did not explicitly state the exact number. Another aspect that makes a comparison questionable is the fact that a considerable amount of papers (~30%) derived their insights from exclusively female samples.

Discussion

The results of the reviewed studies are consistent in mirroring that under-reporting is more likely to occur in overweight and obese participants, especially in those reporting to be more physically active. According to the findings, sex is a key factor that is frequently found to be associated with under-reporting in obese participants, as the majority of studies investigating this link report significantly higher rates for under-reporting in women. A possible explanation is that obese women expose a greater drive for thinness, body dissatisfaction and disinhibited eating than men or women with a lower BMI. One limitation of the three studies exploring gender-specific reporting accuracy is a potential oversight of people with weight reducing diets that may explain high under-reporting rates, specifically for women (Fereidoun Azizi et al., 2004; Johansson et al., 1998; Shaneshin et al., 2012). Moreover, one study compared obese individual’s reporting accuracy to people with anorexia, which may have led to higher visible difference than comparing this group to a population with a healthy BMI (Schebendach et al, 2012). One study found that under-reporting persists into older age and can impact the accuracy of self-reporting food intake (Meng et al., 2013). As a consequence, self-reported EI data coming from obese people must be treated with extreme caution.

There are some limitations with the validity of dietary data that rely on participants’ self-reports, and it is often criticised that the results differ systematically from their usual eating patterns and amount of calories consumed. This assumption is derived from the common observation that the reported EI people with a high BMI, even if determined as accurate, are significantly lower than the number of calories required to maintain their body weight. One possible explanation is that participants who are aware that their food intake is being assessed and evaluated simply ate less than their usual intakes during the recording period. It is known that keeping a food record has an impact on eating behaviour, with participants changing their eating behaviour as a result of being observed (a ‘Hawthorne’ effect; McCarney et al., 2007). Furthermore, a recent study highlights that under-reporting has increased significantly since 1997. Considering the similarities between the surveys that inform this insight, the authors suggest a rising impact of psychosocial factors has affected accuracy, such as social desirability and cognitive restraint. This increase in influence may be due to the rising pursuit of thinness, which can encourage disordered thinking and low satisfaction with body shape and weight, leading to reluctance to report the total amount of food eaten, as it is perceived as socially desirable and more acceptable to eat less. This relationship is most prominent with people who are obese and female (Gemming et al., 2014).

Different dietary self-report methods consistently identify high under-reporting rates in obese participants, which means that there is currently no superior method for this population available to help increase reporting accuracy (Scagliusi et al., 2008). In regard to the methods used for assessing reporting accuracy, it is striking that although the DLW technique has been defined as the most accurate way to evaluate reporting accuracy, it was only used in a minority of the studies. The low utilisation is most likely due to limited financial resources available for conducting research, which shows that in practice the methodical quality tends to be compromised as a result.

In the past, authors have suggested several methods that aim to improve accuracy of calorie estimation. One option could be to simply inform the person about this bias by stating that their meal is twice that of his or her best guess; however, this is often ineffective in correcting them. Another option is to provide people with portion-size benchmarks that they can use in their daily lives as a reference point. Further evidence suggests a solution that asks participants to report calories item by item instead of a total estimate. Since this technique has been associated with higher accuracy, it may provide an effective approach; however, its validity needs to be explored further in future research (Wansink & Chandon, 2006). Martin et al. (2003) have observed that accuracy tends to improve over time and therefore requires practice. Future research needs to investigate whether improvements in reporting accuracy occur after several weeks, which often was the scope of present research. Moreover, it is important to clarify the source of errors in individuals, as this may play an important part in addressing barriers to reporting food intake correctly (Macdiarmid and Blundell, 1998).

Taking the overall findings into account, the current systematic review has been able to establish a clear link between under-reporting and certain demographic characteristics, such as body weight and age; however, no association with psychological, cognitive and social variables was made. Another limitation of this review is its failure to identify digital methods for reporting food intake. Considering the advent of the Internet and increased use of computer- and mobile-based trackers as well as wearable devices over the past decade in various health-related contexts, it is surprising that none of the studies have explored accuracy of digital assessment methods in obese populations and whether they may assist with decrease in under-reporting rates. This gap suggests that there may be scope for future research with added focus for this specific domain by expanding the search strategy.

Conclusion

Taking the studies from this review into account, it appears that some demographic factors other than obesity, also, are important in under-reporting, but most researchers agree that reporting accuracy of food intake decreases with BMI or adiposity.

In the past, research insights have suggested potential ways to overcome the accuracy bias in clinical practice with obese patients, although these are scarce and need to be explored in more detail. It may be beneficial to allow more time and provide additional guidance for people in order to gain confidence with the self-reporting tool, as well as additionally making them aware of the under-reporting bias. By increasing awareness along with providing potential ways to overcome the under-reporting bias, clinicians and dieticians may be able to get a more accurate understanding of their patients’ actual eating behaviours in regard to quantity and dietary sources. This added knowledge could be useful for selecting appropriate weight loss interventions that target relevant eating-related barriers contributing to the excessive weight. Future research needs to focus increasingly on exploring to what extent digital self-reporting methods, such as wearables or mobile apps, may be effective in reducing under-reporting bias particularly within the obese population, since the current review has not been able to identify such tools.

Footnotes

Appendix 1

Table 1.

Study characteristics.

Author	Sample size^a	Food intake assessment
Abbot et al. (2008)	Large	Record
Bartholome et al. (2013)	Small	Recall
Cook et al. (2000)	Large	Diary
De Keyzer et al. (2015)	Large	Recall
Emond et al. (2014)	Large	Recall
Fereidoun Azizi (2005)	Large	Recall
Garriguet (2008)	Large	Recall
Gemming et al. (2014)	Large	Recall
Gnardellis et al. (1998)	Large	Recall
González et al. (1999)	Large	Recall
Hutchesson et al. (2013)	Small	Record
Johansson et al. (1998)	Large	Recall
Kretsch et al. (1999)	Small	Record
Lansky and Brownell (1982)	Small	Record
Lara et al. (2004)	Large	Recall
Lichtman et al. (1992)	Large	Recall
Little et al. (1999)	Average	Record
Lutomski et al. (2011)	Large	Recall
Martin et al. (2003)	Average	Record and interview
McKenzie et al. (2002)	Average	Recall
Meng et al. (2013)	Large	Record
Mossavar-Rahmani et al. (2013)	Large	Recall and questionnaire
Muhlheim et al. (1998)	Small	Diary
Murakami et al. (2013)	Large	Record
Pietlänien et al. (2010)	Small	Diary and questionnaire
Poppitt et al. (1998)	Small	Recall
Samaras et al. (1999)	Large	Record
Schebendach et al. (2012)	Small	Record
Scagliusi et al. (2012)	Average	Recall and record
Shaneshin et al. (2012)	Large	Recall
Svendsen and Tonstad (2006)	Average	Recall and record
Vansant and Hulens (2006)	Average	Recall
Westerterp-Platenga et al. (1996)	Average	Diary
Vyas et al. (2003) (UK)	Large	Recall

Sample size classification based on SIGN50 Checklist defining small as <59, average as 60–150 and large as >150.

Appendix 2

Table 2.

Summary of findings.

Author	Participants	EI method	SR accuracy	Main findings	Quality assessment
Abbott et al. (2008) (USA)	N = 155 OB (BMI: x = 31), middle-aged	3d diet records	Goldberg cut-off (EI:BMR < 1.36)	46% UR (~401.6kcal/day) Factors for UR: years of education (p = 0.001), less realistic weight loss goals (p = 0.02), higher perceived exercise competence (p = 0.07), social support for exercise (p = 0.04), body shape concerns (p = 0.01)	63%
Bartholome et al. (2013) (USA)	N = 15 with BED; n = 17 non-BED F, OW and OB (BMI = 27–35) Age: x = 30.1 ± 6.7	24-hour dietary recall interview	Correlation between recall and laboratory meals	Reporting of actual intake: 90% in BED, 98% in non-BED (p = 0.086) Factors for UR in BED: subjective loss of control, increased EI during BED episodes, less awareness of EI	63%
Cook et al. (2000) (UK)	N = 1097 (n = 539 F; n = 558 M) Age > 65 years	4d food diaries	Goldberg cut-off (EI:BMR < 1.35) with LERs below 1.2 times estimated BMR	LERs have significantly higher BMI than non-LERs in both sex (27.5 vs 25.7 in M, 27.99 vs 25.4 in F) UR significantly higher in F (48% vs 29% in M) Obesity is highest predictor (p < 0.01), no other factors predict UR in F, in M social class and home ownership are additional factors	100%
De Keyzer et al. (2015) (Belgium, Norway, Czech Republic)	N = 365 (healthy adults, convenience subsample of European Food Consumption validation study) Age: 45–65 years BMI: x = 24.8–27.8 OB: n = 87	Questionnaire on salt use Dietary recalls using EPIC soft	Geometric means	Overall, AR for Na intake was highest among NW participants; however, OB women from Norway and Belgium had higher accuracy than OW and NW in these countries As Na⁺ density in diet increases, SR accuracy decreases	87.5%
Emond et al. (2014) (USA)	N = 250 (OW/OB: 51.2%; African-American: 49.2%; 34.4% M)	Web-based 24-hour dietary recall assessment	Reported total caloric intake within 25% of total EE per DLW	Participants who under-reported their total energy intake were more likely to be overweight/obese (61.8%) compared to those who over-reported their total energy intake (37.8%; p = 0.032)	75%
Fereidoun Azizi (2005) (Iran)	N = 947 (OB = 23; 12 M and 23 F); BMI = 24.8 ± 4.4 in M; 25.9 ± 5.4 in F Age = 37.3 ± 14.6 in M; 32.9 ± 13.6 in F	24-hour recalls	Goldberg cut-off (EI:BMR < 1.35)	Older age, higher BMI (p < 0.01), female sex (p < 0.001), but not educational level, are associated with UR	87.5%
Garriguet (2008) (Canada)	N = 16,190	24-hour recalls	EI:EE (ratio of true reporters 0.7–1.42)	Obese subjects were accurate reporters (EI:EE 0.79). UR% in OB subjects (30.3%) not significantly different from NW (31.1%) and OW (37.8%) subjects	87.5%
Gemming et al. (2014) (New Zealand)	N = 3919 (n = 1715 M; 2204 F) OB: n = 807 F; n = 595 M Age: x = 44.3 ± 0.28 M; x = 45.3 ± 0.25 F	Three pass 24-hour dietary recall (computer-based and interviewer-assisted)	EI: RMR_est < 0.9 = LERs	Mean values: M: 1.34 ± 0.02 F: 1.23 ± 0.02 LERs in M: 21% and in F: 25% 30% of LERs are OB, and 25% OW UR greater among priority ethnic groups, older age, OW and OB UR has increased since 1997 (from 6.1%–14.7% in M, from 14.4%–18.6% in F)	100%
Gnardellis et al. (1998) (Greece)	N = 9262 Urban and rural Greek population Sex: 58.1% F, 41.9% M Age: 30–82 years	Semi-quantitative FFQ	UR = EI < 1.14 × BMR	Average UR: 11.8% (13.5 M, 10.5% F) UR more common for men, lower education levels (Odds of 0.76 vs 0.60) and OB (16.8% vs 11.5% in OW and 7.3% in NW)	100%
González et al. (1999) (Spain)	n = 23,289 F; healthy volunteers n = 14,374 M Age: x = 51.5 ± 7.9 OB: n = 6999 Educational level: primary level (39.1%)	Dietary history questionnaire	EI:ER ratio ER = BMR × 1.55 (PA frequency used for Western countries)	Estimated UR for OB: 17.5% in F, 5.5% in M EI:ER ratios for OB F: 82.49:26.02, and for OB M: 94.50:27.86 No significant difference in accuracy to non-OB	100%
Hutchesson et al. (2013) (Australia)	N = 9 F (OB and OW) Age: x = 34.5 ±11.3 BMI: x = 29.2 ± 1.4	9d web-based food record	EI:TEE < 1; assessed by DLW	Mean reporting accuracy: 79.6% (UR in 4 participants) Web-based SR is consistent with other methods in accuracy	63%
Johansson et al. (1998) (Norway)	N = 3144 Age: x = 42.7 (M), x = 41.6 (F) BMI (M): UW: n = 4, NW: n = 54, OW: n = 37, OB: n = 5 BMI (F): UW: n = 15, NW: n = 59, OW, OB: n = 5	FFQ	Goldberg cut-offs (EI:BMR < 1.35)	UR was more common in OB participants with desire to lose weight (p < 0.001; desire for weight change in F: p < 0.05) Average proportion of UR: 40%–50% Link between BMI and UR can be explained by higher desire for weight loss in OB participants; however, most UR cannot be identified by BMI and change of weight Factors associated with UR: Female sex, obesity (9%), desire to lose weight (41%), less consumption of fat and sugar, attitude about body weight, older age, fibre and vitamin C intake	100%
Kretsch et al. (1999) (USA)	N = 22, healthy F NW (BMI: x = 21.3) and OB (50%, BMI: x = 34.2) 16 Whites, 2 Blacks, 2 Hispanics, 2 missed Educational level: 10–16 years	7d food record using household measures	EI – estimated records Negative value = UR	UR rates for NW: −9.7%, for OB: −19.4% BMI correlates inversely with EI difference for NW (r = −0.67, p = 0.02) Depression correlated positively with EI difference for OB	75%
Lansky and Brownell (1982) (USA)	N = 25 F in 27-week behavioural weight loss programme patients (F) OW: 27% Age: 22–55 years (x = 42)	Daily food records for 3 weeks (time and place of food eaten, quantity and calories), calorie guide was provided	Accuracy of participants’ conversion from quantities into calories	58% of drop outs of weight loss programmes UR (vs 23.3% in non-drop-outs) SR does not predict weight loss (r = −0.22–0.30), findings support use of direct observation instead 46% make errors in report: 26% OR, 20% UR	63%
Lara et al. (2004) (UK)	N = 184 F, Age: 18–65 years, seeking help for weight loss in primary care (comparison with non-clinical control group) OB: n = 37, OW: n = 45, BMI < 25: n = 52	3-FFQ	SR via item assessing intention to misreport (options: high OR, moderate OR, high UR, moderate UR)	Overall misreport: 68%, BMI > 30: 46% (both clinical and non-clinical groups) High OR rates (32%) in clinical OB After confrontation with UR, 46% admitted misreporting	63%
Lichtman et al. (1992) (USA)	N = 224, participants in weight loss treatment OB in Group 1: 9F, 1M (diet resistance) OB in Group 2: 67F, 13M BMI: 33.8 ± 4.1 and 36.4 ± 7.3	Self-reported food intake recall over 14 days	Evaluation of recall about test meal the following day via investigator	One day after test meal subjects in group 1 recalled having eaten approximately 20% less than they actually ate (p < 0.05), which means that UR is likely to account for diet resistance Group 1 subjects also displayed higher cognitive restraint, so this could further explain UR	75%
Little et al. (1999) (UK)	1. High risk group (n = 61): risk for cardiovascular disease (56% M, age < 50: 52%, OB: 54%) 2. Random population group (n = 50) (38% M, age < 50: 56%, OB: 18%)	Brief dietary assessment tools 7d weighed dietary record	EI:BMR < 1.2	UR is common (40%) and more likely in OB participants (29% difference, p < 0.001) UR rates for OB: 60%, non-OB: 30% (χ² = 10.9) Simple SR tools show acceptable agreements with standard measures	75%
Lutomski et al. (2011) (Ireland)	N = 7521; 1.3% UW, 38.9% NW, 43.9% OW, 15.9% OB	FFQ	Goldberg cut-off (EI:BMR < 1.35)	33% classified as UR (with men significantly more likely to UR), 11.9% are OR (most common in normal BMI and UW women) Odds of UR greatest among obese (2.16 OR)	100%
Martin et al. (2003) (USA)	N = 56, diabetes mellitus type II patients with prescription of low-fat diet n = 28 intervention group n = 28 control group n = 25 M, n = 31 F BMI ⩾ 30: n = 22, x = 31.28	Interview-administered diet history; 3d food record	Goldberg cut-off for diet history interview EI:BMR < 1.14, for 3d food record EI:BMR < 1.06	UR more common in BMI > 30: 91% for diet history and 82% for 3d food record Increasing accuracy over intervention time of 12 months (p < 0.05), suggesting that lack of reporting skills is initial barrier	75%
McKenzie et al. (2002) (USA)	N = 88 F (OW and OB) Age: 37.2 ± 6.1 BMI: 31.6 ± 3.9	2 dietary recalls (multiple-pass 24-hour recall): 1. Telephone 2. In-person	Goldberg cut-off (EI:BMR) adjusted to PAL, lower cut-off ratios (0.90, 0.94; 1.04 for low, medium and high activity)	Interviewer body mass index has no impact on accuracy of self-reported EI (p = 0.19) UR occurred in both telephone and in-person recall data (~26%) with no significant difference (p = 0.57)	50%
Meng et al. (2013) (Australia)	N = 219 F, age: 70–80 years OB: n = 50, BMI; OW: n = 95, NW: n = 72, other: n = 2 (community-dwelling and recruited for dietary trial)	3d weighed food record	Goldberg cut-off (EI:BMR < 1.35)	BMI ⩾ 25 most significant cut-off for UR Highest likelihood of UR in obese category (32% of OB participants UR) Combination of high activity levels (p < 0.001) and high BMI (p = 0.001) is linked to highest chance of UR UR report fewer food items Prevalence of UR in elderly women: 49%	75%
Mossavar-Rahmani et al. (2013) (USA)	N = 450 F (postmenopausal) n = 156 NW, n = 121 OW; n = 173 OB	FFQ; 4DFR; 24-hour recall with focus on EI and protein intake	Goldberg cut-off (EI:BMR < 1)	Small influence of BMI on UR, higher for EI (r = 8.1%) than protein intake (r = 4.1%) Psychosocial factors play small role in UR: social desirability (increased for UR), meals at home (decreased for UR)	87.5%
Muhlheim et al. (1998) (USA)	N = 28 unsuccessful dieters (OW and OB)	1-week food diary, N = 17 continued 2 more weeks while being told that researcher could verify their report	EI:EE (measured by DWL technique)	Subjects in experimental group improved in their reporting accuracy compared to control group (52% vs 48% of actual intake), thus, the belief that researcher can verify reports reduced UR Experimental group continued to UR (p < 0.0005), showing that UR cannot be changed easily	75%
Murakami et al. (2013) (UK)	N = 1487 (678 M, 809 F), random adult sample Age: 19–64 years, BMI: x = 27.3 ± 4.4	7d weighed dietary record with support from trained interviewers	EER:EI = 1 (95% confidence interval)	Obesity has significant influence on UR (p < 0.001) 33% of male UR are OB (vs 18.5% in AR) and 31.8% of female UR are OB (vs 13.2% in AR) High dietary glycaemic load may explain UR-OB link (only present in OB under-reporters)	100%
Pietlänien et al. (2010) (Finland)	N = 24 MZ twins (N = 14 OB (BMI difference: 5.2 ± 1.6 kg/m²) N = 10 control pairs (MZ twin) Age: 24–28 years (x = 25.7)	3d food diary and eating behaviour questionnaire	TEE, measured by DLW Wicoxon test was used to analyse whether UR was significantly different from zero	UR is considerable for OB twins (3.2 ± 1.1 MJ/day, p = 0.036), OR in participants with high PAL (1.8 MJ/day, p = 0.04), but only in OB (25% of total EE vs 8% in non-OB) Both co-twins agree that OB twin consumes more food/snacks more than non-OB Results highlight that non-shared environment, or lifestyle behaviours, account for UR	63%
Poppitt et al. (1998) (UK)	N = 33 F (18 OB (x = 40.5 kg/m²), 15 non-OB (x = 23.7 kg/m²), recruited for long-stay metabolic facility Age: x = 42 ± 14 years (20–65 years)	24-hour recall of communal meals (selected from menu with options incl. high/low-fat and high/low sugar)	(Reported intake/ actual intake) × 100 Analysis with paired t-tests (p < 0.05)	Total reported daily EI was inaccurate (87.5%, p < 0.01), incl. 25 UR and 8 OR. Meals were accurately reported in both OB and non-OB participants, but snacks (between meals) sign. UR (p < 0.001). Reported EI from carbohydrates and added sugar <actual intake (p < 0.001), but accurate for protein and fat (p > 0.05). Overall no significant Differences between OB and non-OB	63%
Samaras et al. (1999) (UK)	N = 436, twin study in hospital setting Age: x = 58 ± 6 years BMI: x = 24.3 ± 3.6 (5% OB; 30% OW; 65% NW)	7d food record (n = 197) ‘Oxford type’ FFQ	EI – BEE × 1.35 < 0	Age was not found to be linked to UR UR occurred more commonly in OB participants (44% vs 39% in OW and 18% in NW; p < 0.001) Reported EI in UR sign. lower in fat, similar in carbohydrates and higher in proteins than AR	87.5%
Schebendach et al. (2012) (USA)	N = 40 (10 OB: 2M and 8F; 18 AN, 12 NW) BMI: 33.9 ± 2.4)	4d food record during laboratory meal study	Bland-Altman plots	Obesity is linked to UR (p = 0.016); average UR of 19% (160 kcal/day)	63%
Scagliusi et al. (2012) (Brazil)	N = 65F (28 NW, 10 OW, 27OB) Age: 33.7 ± 10.8 BMI: 27.9 ± 6.7 Education (years of study): 12.9 ± 2.3	3 × 24 hours recalls 3d food record FFQ	EI: TEE < 0.69 (24-hour recall) EI:TEE < 0.68 (food record) EI:TEE < 0.82 (FFQ) Total EE measured by DLW	UR in OB: EI:TEE in OB is 0.70 ± 0.24 for 24-hour recall, 0.68 ± 0.22 for food records and 0.76 ± 0.34 for FFQ For all 3 measures sign. more UR than in lower BMI categories (p < 0.05) Spearman correlation for BMI and EI:TEE: −0.47 (p < 0.001) UR varied across methods, FFQ having the lowest accuracy	63%
Shaneshin et al. (2012) (Iran)	N = 187 F BMI ⩾ 30: 35.8% (x = 27.7)	Semi-quantitative 168-item FFQ	Goldberg cut-offs (EI:BMR ⩽ 1.34)	UR rate across all weight categories was 35.5% OB: 48% UR, 30% AR, 21% OR higher resting metabolism is linked to UR (p < 0.05)	88%
Svendsen and Tonstad (2006) (Norway)	N = 50 (n = 23 M, n = 27 F) OB with metabolic risk factors BMI: x = 35.7	FFQ (90 days) Dietary records (3 days)	EI:TEE < 1.0	56% of the participants are identified as UR UR have lower intake of bread, sweets, desserts and snacks (100 vs 161 g/d in OR; p = 0.008) UR have higher scores in dietary restraint (p = 0.028)	63%
Vansant and Hulens (2006) (Belgium)	N = 137 F Age: 40 ± 12 BMI: 38.2 ± 6.0 Attendants of outpatient obesity clinic	Dietary history (obtained by trained dietician)	EI:EE (EI > 10% lower than EE)	16% were over-reporters, 66% were under-reporters (mean level of UR: 18.0±29.1%) Restrained eating was associated with UR (p < 0.01)	63%
Westerterp-Platenga et al. (1996) (Netherlands)	N = 68 F BMI: n = 34 OB, n = 34 non-OB Age: 20–50 years	Food intake diary with household measure incl. Harris Benedict equation	UR = reported EI < 10% of actual intake	Non-OB: 8.7% UR; OB: 8.8% UR OB consumed significantly more energy, esp. from highest energy density category (24% vs 13% in non-OB), and sign. fewer calories from lowest energy density category (24% vs 38%)	75%
Vyas et al. (2003) (UK)	N = 416 adult Pakistanis and Europeans in Manchester Age: 25–79 years BMI: x = 27.0–30.2	FFQ	EI:BMR < 1.2	UR had higher BMI in all ethnic groups and lower reported EI UR in M: 73.6%; UR in F: 77.5% Ethnic group and sex had no influence on UR rates	75%

AN: anorexic; AR: accurate report; BED: binge eating disorder; BEE: basal energy expenditure; BMI: body mass index; BMR: basal metabolic rate; d: day; DFR: day food record; DLW: doubly labelled water; EE: energy expenditure; EER: energy efficiency ratio; EI: energy intake; EPIC: Software for storing electronic medical patient redord;s ER: energy requirements; F: female; FFQ: Food Frequency Questionnaire; LER: low energy reporter; M: male; MZ: monozygotic; NW: normal weight; OB: obese; OR: over-report; OW: overweight; UW: underweight; TEE: total energy expenditure; PA: physical activity; PAL: physical activity level; RMR: resting metabolic rate; SR: self-report; UR: under-report.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship and/or publication of this article.

Supplementary Material

Supplementary material for this article is available online.

References

Abbot

Thomson

Ranger-Moore

, et al. (2008) Psychosocial and behavioral profile and predictors of self-reported energy underreporting in obese middle-aged women. Journal of the American Dietetic Association 108(1): 114–119.

Baranowski

Domel

(1994) A cognitive model of children’s reporting of food intake. American Journal of Clinical Nutrition 59(1): 212S–217S.

Barrett-Connor

(1991) Nutrition epidemiology: How do we know what they ate? American Journal of Clinical Nutrition 54(1): 182S–187S.

Barrie

Coward

(1985) A rapid analytical technique for the determination of energy expenditure by the doubly labelled water method. Biological Mass Spectrometry, 12(9): 535–541.

Bartholome

Peterson

Raatz

, et al. (2013) A comparison of the accuracy of self-reported intake with measured intake of a laboratory overeating episode in overweight and obese women with and without binge eating disorder. European Journal of Nutrition 52(1): 193–202.

Black

(2000) Critical evaluation of energy intake using the Goldberg cut-off for energy intake: Basal metabolic rate. A practical guide to its calculation, use and limitations. International Journal of Obesity and Related Metabolic Disorders 24(9): 1119–1130.

Black

Prentice

Goldberg

, et al. (1993) Measurements of total energy expenditure provide insights into the validity of dietary measurements of energy intake. Journal of the American Dietetic Association 93(5): 572–579.

Buhl

Gallagher

Hoy

, et al. (1995) Unexplained disturbance in body weight regulation: Diagnostic outcome assessed by doubly labeled water and body composition analyses in obese patients reporting low energy intakes. Journal of the American Dietetic Association 95(12): 1393–1400.

Consultancy UK (2015) McKinsey: Obesity costs UK society 73 billion per year. Available at: http://www.consultancy.uk/news/1278/mckinsey-obesity-costs-uk-society-73-billion-per-year (accessed 20 June 2015).

10.

Cook

Pryer

Shetty

(2000) The problem of accuracy in dietary surveys. Analysis of the over 65 UK National Diet and Nutrition Survey. Journal of Epidemiology and Community Health 54(8): 611–616.

11.

Crawley

Summerbell

(1997) Feeding frequency and BMI among teenagers aged 16-17 years. International Journal of Obesity 21(2): 159–161.

12.

De Keyzer

Dofková

Lillegaard

ITL

, et al. (2015) Reporting accuracy of population dietary sodium intake using duplicate 24 h dietary recalls and a salt questionnaire. British Journal of Nutrition 113(03): 488–497.

13.

Emond

Patterson

Jardack

, et al. (2014) Using doubly labeled water to validate associations between sugar-sweetened beverage intake and body mass among White and African-American adults. International Journal of Obesity 38(4): 603.

14.

Fereidoun Azizi

(2005) Correlates of under—and over-reporting of energy intake in Tehranians: Body mass index and lifestyle-related factors. Asia Pacific Journal of Clinical Nutrition 14(1): 54–59.

15.

Garriguet

(2008) Impact of identifying plausible respondents on the under-reporting of energy intake in the Canadian Community Health Survey. Health Reports 19(4): 47.

16.

Gemming

Jiang

Swinburn

, et al. (2014) Under-reporting remains a key limitation of self-reported dietary intake: An analysis of the 2008/09 New Zealand Adult Nutrition Survey. European Journal of Clinical Nutrition 68(2): 259–264.

17.

Gibson

(2005) Principles of nutritional assessment. New York: Oxford University Press.

18.

Gnardellis

Boulou

Trichopoulou

(1998) Magnitude, determinants and impact of under-reporting of energy intake in a cohort study in Greece. Public Health Nutrition 1(2): 131–137.

19.

González

Pera

Quirós

, et al. (2000) Types of fat intake and body mass index in a Mediterranean country. Public Health Nutrition 3(03): 329–336.

20.

Heitmann

(1993) The influence of fatness, weight change, slimming history and other lifestyle variables on diet reporting in Danish men and women aged 35-65 years. International Journal of Obesity and Related Metabolic Disorders 17(6): 329–336.

21.

Hill

Davies

PSW

(2001) The validity of self-reported energy intake as determined using the doubly labelled water technique. British Journal of Nutrition 85(4): 415–430.

22.

Hutchesson

Truby

Callister

, et al. (2013) Can a web-based food record accurately assess energy intake in overweight and obese women? A pilot study. Journal of Human Nutrition and Dietetics 26(s1): 140–144.

23.

Johansson

Solvoll

Bjørneboe

, et al. (1998) Under-and overreporting of energy intake related to weight status and lifestyle in a nationwide sample. American Journal of Clinical Nutrition 68(2): 266–274.

24.

Kretsch

Fong

Green

(1999) Behavioral and body size correlates of energy intake underreporting by obese and normal-weight women. Journal of the American Dietetic Association 99(3): 300–306.

25.

Lafay

Basdevant

Charles

, et al. (1997) Determinants and nature of dietary underreporting in a free-living population: The Fleurbaix Laventie Ville Sante (FLVS) Study. International Journal of Obesity 21(7): 567–573.

26.

Lansky

Brownell

(1982) Estimates of food quantity and calories: Errors in self-report among obese patients. American Journal of Clinical Nutrition 35(4): 727–732.

27.

Lara

Scott

Lean

(2004) Intentional mis-reporting of food consumption and its relationship with body mass index and psychological scores in women. Journal of Human Nutrition and Dietetics 17(3): 209–218.

28.

Lincoln

(1972) Calorie intake, obesity, and physical activity. The American Journal of Clinical Nutrition 25(4): 390–394.

29.

Little

Barnett

Margetts

, et al. (1999) The validity of dietary assessment in general practice. Journal of Epidemiology and Community Health 53(3): 165–172.

30.

Livingstone

Prentice

Coward

, et al. (1992) Validation of estimates of energy intake by weighed dietary record and diet history in children and adolescents. American Journal of Clinical Nutrition 56(1): 29–35.

31.

Livingstone

Prentice

Strain

, et al. (1990) Accuracy of weighed dietary records in studies of diet and health. BMJ 300(6726): 708–712.

32.

Livingstone

MBE

Black

(2003) Markers of the validity of reported energy intake. Journal of Nutrition 133(3): 895S–920S.

33.

Lutomski

Van den Broeck

Harrington

, et al. (2011) Sociodemographic, lifestyle, mental health and dietary factors associated with direction of misreporting of energy intake. Public Health Nutrition 14(03): 532–541.

34.

McCarney

Warner

Iliffe

, et al. (2007) The Hawthorne Effect: A randomised, controlled trial. BMC Medical Research Methodology 7(1): 1.

35.

Macdiarmid

Blundell

(1998) Assessing dietary intake: Who, what and why of under-reporting. Nutrition Research Reviews 11(2): 231–253.

36.

McKenzie

Johnson

Harvey-Berino

, et al. (2002) Impact of interviewer’s body mass index on underreporting energy intake in overweight and obese women. Obesity Research 10(6): 471–477.

37.

McKinsey & Company (2014) The obesity crisis. Available at: http://www.mckinsey.com/mgi/overview/in-the-news/the-obesity-crisis (accessed 26 April 2016).

38.

Martin

Tapsell

Denmeade

, et al. (2003) Relative validity of a diet history interview in an intervention trial manipulating dietary fat in the management of Type II diabetes mellitus. Preventive Medicine 36(4): 420–428.

39.

Meng

Kerr

Zhu

, et al. (2013) Under-reporting of energy intake in elderly Australian women is associated with a higher body mass index. Journal of Nutrition, Health & Aging 17(2): 112–118.

40.

Morgan

Jain

Miller

, et al. (1978) A comparison of dietary methods in epidemiologic studies. American Journal of Epidemiology 107(6): 488–498.

41.

Morris

Beilharz

Maniam

, et al. (2014) Why is obesity such a problem in the 21st century? The intersection of palatable food, cues and reward pathways, stress, and cognition. Neuroscience & Biobehavioral Reviews 58: 36–45

42.

Mossavar-Rahmani

Tinker

Huang

, et al. (2013) Factors relating to eating style, social desirability, body image and eating meals at home increase the precision of calibration equations correcting self-report measures of diet using recovery biomarkers: Findings from the Women’s Health Initiative. Nutrition Journal 12(1): 63.

43.

Muhlheim

Allison

Heshka

, et al. (1998) Do unsuccessful dieters intentionally underreport food intake? International Journal of Eating Disorders 24(3): 259–266.

44.

Murakami

McCaffrey

Livingstone

MBE

(2013) Associations of dietary glycaemic index and glycaemic load with food and nutrient intake and general and central obesity in British adults. British Journal of Nutrition 110(11): 2047–2057.

45.

Myers

Klesges

Eck

, et al. (1988) Accuracy of self-reports of food intake in obese and normal-weight individuals: Effects of obesity on self-reports of dietary intake in adult females. The American journal of clinical nutrition 48(5): 1248–1251.

46.

Pietiläinen

Korkeila

Bogl

, et al. (2010) Inaccuracies in food and physical activity diaries of obese subjects: Complementary evidence from doubly labeled water and co-twin assessments. International Journal of Obesity 34(3): 437–445.

47.

Poppitt

Swann

Black

, et al. (1998) Assessment of selective under-reporting of food intake by both obese and non-obese women in a metabolic facility. International Journal of Obesity & Related Metabolic Disorders 22(4): 303–311.

48.

Poslusna

Ruprich

de Vries

, et al. (2009) Misreporting of energy and micronutrient intake estimated by food records and 24 hour recalls, control and adjustment methods in practice. British Journal of Nutrition 101(S2): S73–S85.

49.

Royal College of Physicians (2013) NHS must step up to obesity challenge. Available at: https://www.rcplondon.ac.uk/news/nhs-must-step-obesity-challenge (accessed 3 July 2017).

50.

Samaras

Kelly

Campbell

(1999) Dietary underreporting is prevalent in middleaged British women and is not related to adiposity (percentage body fat). International Journal of Obesity 23(8): 881–888.

51.

Scagliusi

Ferriolli

Pfrimer

, et al. (2008) Underreporting of energy intake in Brazilian women varies according to dietary assessment: A cross-sectional study using doubly labeled water. Journal of the American Dietetic Association 108(12): 2031–2040.

52.

Schebendach

Porter

Wolper

, et al. (2012) Accuracy of self-reported energy intake in weight-restored patients with anorexia nervosa compared with obese and normal weight individuals. International Journal of Eating Disorders 45(4): 570–574.

53.

Schoeller

(1990) How accurate is self-reported dietary energy intake? Nutrition Reviews 48(10): 373–379.

54.

Schoeller

Van Santen

(1982) Measurement of energy expenditure in humans by doubly labeled water method. Journal of Applied Physiology 53(4): 955–959.

55.

Sedghi

(2014) How obese is the UK? And how does it compare to other countries? The Guardian. Available at: http://www.theguardian.com/news/datablog/2014/may/29/how-obese-is-the-uk-obesity-rates-compare-other-countries (accessed 20 July 2015).

56.

Selassie

Sinha

(2011) The epidemiology and aetiology of obesity: A global challenge. Best Practice & Research Clinical Anaesthesiology 25(1): 1–9.

57.

Shaneshin

Rashidkhani

Rabiei

(2012) Accuracy of energy intake reporting: Comparison of energy intake and resting metabolic rate and their relation to anthropometric and sociodemographic factors among Iranian women. Archives of Iranian Medicine 15(11): 681–687.

58.

SIGN (2004) SIGN 50: A Guideline Developer’s Handbook. Edinburgh: Scottish Intercollegiate Guidelines Network.

59.

Svendsen

Tonstad

(2006) Accuracy of food intake reporting in obese subjects with metabolic risk factors. British Journal of Nutrition 95(3): 640–649.

60.

Thompson

Midthune

Williams

, et al. (2008) Evaluation of a short dietary assessment instrument for percentage energy from fat in an intervention study. The Journal of nutrition 138(1): 193S–199S.

61.

Tooze

Subar

Thompson

, et al. (2004) Psychosocial predictors of energy underreporting in a large doubly labeled water study. American Journal of Clinical Nutrition 79(5): 795–804.

62.

Vansant

Hulens

(2006) The assessment of dietary habits in obese women: Influence of eating behavior patterns. Eating Disorders 14(2): 121–129.

63.

Vyas

Greenhalgh

Cade

, et al. (2003) Nutrient intakes of an adult Pakistani, European and African-Caribbean community in inner city Britain. Journal of Human Nutrition and Dietetics 16(5): 327–337.

64.

Wang

McPherson

Marsh

, et al. (2011) Health and economic burden of the projected obesity trends in the USA and the UK. Lancet 378(9793): 815–825.

65.

Wansink

Chandon

(2006) Meal size, not body size, explains errors in estimating the calorie content of meals. Annals of Internal Medicine 145(5): 326–332.

66.

Westerterp-Plantenga

Pasman

Yedema

, et al. (1996) Energy intake adaptation of food intake to extreme energy densities of food by obese and non-obese women. European journal of clinical nutrition 50(6): 401–407.

67.

Westerterp

Saris

Van Es

, et al. (1986) Use of the doubly labeled water technique in humans during heavy sustained exercise. Journal of Applied Physiology 61(6): 2162–2167.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.03 MB