Evaluation of Dietary Assessment Tools Used in Bariatric Population

Severe obesity is associated with major health issues and bariatric surgery is still the only treatment to offer significant and durable weight loss. Assessment of dietary intakes is an important component of the bariatric surgery process. Objective: To document the dietary assessment tools that have been used with patients targeted for bariatric surgery and patients who had bariatric surgery and explore the extent to which these tools have been validated. Methods: A literature search was conducted to identify studies that used a dietary assessment tool with patients targeted for bariatric surgery or who had bariatric surgery. Results: 108 studies were included. Among all studies included, 27 used a dietary assessment tool that had been validated either as part of the study per se (n = 11) or in a previous study (n = 16). Every tool validated per se in the cited studies was validated among a bariatric population, while none of the tools validated in previous studies were validated in this population. Conclusion: Few studies in bariatric populations used a dietary assessment tool that had been validated in this population. Additional studies are needed to develop valid and robust dietary assessment tools to improve the quality of nutritional studies among bariatric patients.


Introduction
Obesity is a common, complex chronic disease and its prevalence has increased over the past several years, making it a major public health concern [1]. More importantly, the prevalence of severe obesity (BMI ≥ 35 kg/m 2 ) has increased dramatically in Canada [2]. Severe obesity is associated with major health issues such as an increased risk of hypertension, type 2 diabetes, sleep apnea and cancer [2]. Bariatric surgery is the only treatment for severe obesity to offer significant and durable weight loss as well as improvement of metabolic diseases [3]. Multiple types of surgery exist and are usually classified as restrictive, malabsorptive or mixed-procedures. Restrictive surgery limits the amount of food consumed by reducing stomach size, while malabsorptive surgery limits nutrient absorption by bypassing or reorganizing parts of the small intestine. Mixed-procedures, the most common surgeries, combine both gastric restriction and intestinal malabsorption [4,5]. Assessment of dietary intakes and eating behaviors are important components of the bariatric surgery process especially after surgery, since diet quality of bariatric patients is most likely to impact their risk of developing nutritional deficiencies [6] and their food preferences and choices could impact the success of their weight loss [7].
Many dietary assessment tools are used in nutritional research, especially self-report tools because they are often easier to use and less expensive as opposed to using controlled feeding environments, direct observation or measurements of biomarkers. The most common self-reported tools are food records (FR), 24-h dietary recalls (24HR) and food frequency questionnaires (FFQ). Over the last years, these tools have been adapted for a web-based use, such as self-administered web-based 24HR [8,9] or a web-based FFQ [10,11], to increase cost-efficiency and therefore the applicability to large cohort studies. The FR is a dietary assessment tool where respondents have to report all the foods and beverages consumed during the current day with as many details as possible (portion size, brand, method of cooking, time of the day, location of the eating occasion, etc.), for a variable number of days (often between 3 and 7) [12]. The 24HR consists of listing detailed information about everything the respondent ate and drank from midnight to midnight the previous day, or over the past 24-h period [12]. Finally, FFQ is a fixed-sequence questionnaire based on a predetermined series of foods and beverages consumed over a given period of time, which can be the previous week, month or year. The number and size of portions are often asked subsequently [12]. Of all these dietary assessment tools, the 24HR has been hypothesized as the least biased dietary assessment tool, since FR is more associated to reactivity biases such as a tendency to modify the usual diet for a more socially desirable manner or to simplify the recording task, and FFQ is known to encompass more important systematic biases than 24HR (does not capture the entire diet due to difficulty of the recall task) [12].
One reason explaining the difficulty to select the most appropriate dietary assessment tool with patients targeted for bariatric surgery and patients who had bariatric surgery is the relative lack of validation of these tools within those specific populations. Validity of an instrument is the degree to which an instrument measures what it is supposed to measure [12]. To determine the validity of an instrument, it is often compared with another instrument measuring the same concept and known to be accurate or considered as a gold standard [12]. Validation of dietary assessment is conducted to determine how accurately self-report instruments measure true dietary intakes [13]. It is crucial to develop and use tools that provide an accurate and precise measure of dietary intakes to optimize treatment and the nutritional care provided to patients targeted for or who had bariatric surgery [6,14]. Moreover, as patients who have undergone bariatric surgery have a higher risk of developing nutritional deficiencies [6], it is also essential to select dietary assessment tools validated for global intakes, particularly protein intakes since it is the major macronutrient deficiency after bariatric surgery [14,15].
The aim of this review was to document the dietary assessment tools that have been used in research involving patients targeted for bariatric surgery and patients who had bariatric surgery, and to explore the extent to which these tools have been validated.

Search Strategy
A literature search was conducted for all articles published on Pubmed up to January 2021 to identify studies that used dietary assessment tool with patients targeted for bariatric surgery or who had bariatric surgery. The search strategy was done using this keywords combination: "food intake"[All Fields] OR "food intake evaluation"[All Fields] OR "dietary intake"[All Fields] OR "dietary intake evaluation"[All Fields] OR "dietary assessment"[All Fields] OR "dietary assessment evaluation" [All Fields] OR "food assessment" [All Fields] OR "food assessment evaluation" [All Fields] AND bariatric [All Fields].

Selection of Studies
The literature search was performed independently by three authors (G.B.M., M.L., V.L.) and included all studies published on Pubmed up to 2021. Studies were found and retained in three stages: (i) the first stage was a screening done directly on Pubmed according to the title and abstract, (ii) the second one was the complete reading of the articles, and (iii) the third stage was a screening of the references of the retained articles. Inclusion and exclusion criteria that were used are presented in Table 1. Only original studies were included in this review, based on the inclusion and exclusion criteria.

Data Extraction
The following data were extracted by three authors (M.L., G.B.M., V.D.-L.) for each study: (a) bibliographical data (author, publication year, country); (b) sample characteristics (sample size, type of surgery, mean and standard deviation (SD) for age, sex and body mass index); (c) study design features (objective, study design and dietary assessment tool); (d) outcomes (self-reported energy and nutrient intakes, information on the validity of the dietary assessment tool, if available) ( Table 2). Information regarding the validity of the dietary assessment tool was also extracted, such as the reference method used for validation, the population in which the validation has been performed, and information about the validation process (Table 3).      Investigate the perception of hunger and satiety and its association with nutrient intake in women who regain weight in the postoperative period after bariatric surgery.  No correlations were found between serum/plasma concentrations and nutritional intake nor associations between low concentrations and inadequate intakes.

No No
Agreement between the two methods was high, although it may have been overestimated because the two assessments were consecutives to one another. The tool may be highly advantageous for large population-based surveys [125]. Pre-surgery Yes Comparing the results obtained for the modified Goldberg equations in this study, there was considerable variation in the proportion of underreporting (55% to 97%).

Pre-& post-surgery Yes
Values revealed that patients from both groups underreported their caloric intake by 8% pre-surgery.

No No
The data showed that there was substantial variability in the accuracy of the FFQ at the individual level. Furthermore, the results showed that the questionnaire was more accurate for groups than individuals [127]. On average, 39% of the men were classified in the same quartile with the two methods, and 3% in the opposite quartile. Very-long chain n-3 fatty acids in adipose tissue and total serum lipids reflect the dietary intake of very-long-chain n-3 fatty acids to the same degree. No associations were observed between intake of alpha-tocopherol and concentration in adipose tissue and serum [130]. Correlations between questionnaire and FR for percent of energy from fat were 0.67 and 0.65 respectively in the two groups; most correlations were similar to those achievable by a single 4-day FR [131].  [132]. Post-surgery Yes

Other dietary assessment methods
After SG, patients reported higher total energy intake and energy intake from carbohydrates compared to estimations using photographs. Digital photography appears reliable and accurate in adults in measuring energy intake in a cafeteria setting.

Food Records (FR)
Of the three studies having tested the validity of the FR per se in their bariatric population, two studies [27,63] used indirect calorimetry as a reference and one study [121] used plasma concentrations biomarkers (vitamin A, D, E and C) as reference (Table 3). Regarding the validity of the tools, Bobbioni-Harsh et al. [27] found that the mean selfreported energy intake from their 3-day FR was 17.2% lower than energy requirement

Food Records (FR)
Of the three studies having tested the validity of the FR per se in their bariatric population, two studies [27,63] used indirect calorimetry as a reference and one study [121] used plasma concentrations biomarkers (vitamin A, D, E and C) as reference (Table 3). Regarding the validity of the tools, Bobbioni-Harsh et al. [27] found that the mean self-reported energy intake from their 3-day FR was 17.2% lower than energy requirement evaluated with indirect calorimetry pre-surgery. Golzarand et al. [63] found that protein and carbohydrate oxidation were significantly decreased post-surgery. Wolf et al. [121] found no correlation between self-reported dietary intakes obtained from a 3-day FR pre-surgery and corresponding serum concentrations biomarkers of intake (25-hydroxycholecalciferol, retinol, ascorbic acid, tocopherol/cholesterol ratio, β-carotene, calcium, magnesium, phosphate).

24-h Dietary Recall (24HR)
Four studies tested the validity of the 24HR per se in their bariatric population [67,92,97,115] (Table 3). In two of those studies, indirect calorimetry (resting metabolic rate, energy requirement) pre-surgery [97] and pre-and post-surgery [115] was used as a reference. Total daily energy intake assessed by 24HR was below measured resting metabolic rate pre-surgery by 8% in Verger et al.' study [115], while Quesada et al. [97] found that 55 to 97% of their participants underreported their intake compared to resting metabolic rate. Another study [67] tested the validity of their 24HR using 24-h urine recovery biomarker data as a reference for protein intake pre-surgery, and another one [92] used FR post-surgery as a reference (energy, macro and micronutrient intakes). Kops et al. [67] concluded that approximately 37% of bariatric patients underreported protein intakes pre-surgery assessed with 24HR compared to 24-h urinary recovery biomarker data, while 25% overreported it. Novais et al. [92] validated their 24HR by comparing it with a 3-day FR and found a high level of agreement between both tools for energy and nutrient intakes.

FFQ
One study [34] directly tested FFQ validity using a 24HR as a reference in post-surgery patients and found a difference of 150 kcal between the two methods (1230 kcal with the FFQ vs. 1083 kcal with the 24HR) ( Table 3).

Questionnaires
None of the studies that used a questionnaire to assess mean daily energy intake used a questionnaire validated in bariatric population. It is important to mention that little information was available about the form of questionnaires used. Five studies [66,69,71,93,120] used the Swedish Obese Subjects (SOS) study questionnaire [132] (Table 3), which was adapted from a simplified dietary history interview and was previously validated using a 4-day FR, nitrogen urinary excretion and 24 h energy expenditure measured by indirect calorimetry in obese and non-obese population, but not in bariatric population.

Other Dietary Assessment Methods
Al-Ozairi et al. [17] used a photo-assisted diet capture method to assess energy intake in post-surgery (Table 3). They found that after SG, patients reported a higher energy intake with the 24HR compared to estimations obtained using photographs, but they suggested that digital photography was more reliable and accurate for measuring energy intake in this specific population than 24HR [17].

Mixed Methods
Two studies validated the use of mixed methods to assess dietary intakes among bariatric population [28,33] (Table 3). Casagrande et al. [33] used both FFQ and 24HR to assess dietary intakes pre-surgery. Protein, cholesterol and sodium intakes were lower with the FFQ than with the 24HR, while calcium intake was higher [33]. To assess the accuracy of the estimated mean dietary intake found with the 24HR, Brolin et al. [28] used a 1-week FFQ to compare both dietary intakes pre-surgery. They found statistically significant correlations between the tools for total energy intake and intake of milk and ice cream products, sweet/soda and nonliquid sweets [28].

Discussion
The objective of this review was to document the dietary assessment tools used among patients targeted for bariatric surgery and those who have undergone bariatric surgery. A total of 108 studies were included in this review; only 27 (25%) validated their dietary assessment tool or used a tool that had been previously validated, and only 11 (10%) were validated in bariatric population. Of these 11 studies, only 3 of them validated the dietary assessment tool before and after surgery, 5 validated it only before surgery, and 3 only after surgery.
The validation process of dietary assessment tools is complex but is imperative in order to evaluate usual dietary intakes and also provide an adequate estimation of nutrient intakes and potential deficiencies following bariatric surgery [6]. As previously mentioned, the dietary assessment tool of interest is often compared with another tool measuring the same concept and known to be accurate or considered as a gold standard to determine the validity [3,4]. Direct observation, which refers to objective assessment of foods and beverages consumed, is also frequently used in a clinical setting [5]. This method remains the best option to exclude risk of estimation bias, which could be present with another dietary assessment tool [2], but it is not representative of usual intakes and can cause other biases such as response bias since participants are being observed. No study using direct observation were found for this review. Most of validation studies included in this review used the comparison with another dietary assessment tool (n = 4; 1 FR, 2 24HR, 1 FFQ) or used indirect calorimetry (n = 4) to assess energy expenditure and macronutrient's oxidation. Indirect calorimetry is less biased than self-report dietary assessment tools [12], however the later are more commonly selected as they are more accessible [12]. In the general population, FR are the most commonly used self-report tools to validate dietary intakes [8]. In order to improve quality of the validation process, the dietary assessment tool needs to be tested and compared, by direct observation or with a reference method, within the same population [6,7]. In the current review, we found that only 10% of the validated tools were validated in a bariatric population, showing a clear lack of studies that used a tool validated in that specific population. Moreover, conclusions about validity of the tools varied considerably among studies (as seen in Table 3). However, in general, FR were found as acceptable as a dietary assessment tool [27,124]. Authors found underreporting of dietary intake while validating their 24HR [97,115], but it was still deemed appropriate [67,92], particularly when used within epidemiological studies [125]. Studies examining the validation of FFQs found almost the same conclusion, namely more accurate with groups than individuals [127,129] and with a reasonable validity [128,130,131]. The only validated questionnaire had the tendency to report higher dietary intake than FR or nitrogen excretion [132]. Finally, digital photography seems to be a reliable and accurate tool for dietary intakes assessment [17], but more studies are needed to confirm these results.
Factors characterizing the bariatric population such as bias and stigmatization, dietary requirements pre-versus post-surgery and type of surgery might influence the choice of the dietary assessment tool and need to be considered in the validation process. Inclusion of patients who will have bariatric surgery and patients who have undergone bariatric surgery in the same study can be questioned as characteristics of patients and susceptible biases in reporting dietary intakes can broadly differ. For instance, social desirability biases and stigmatization can be stronger prior to than after bariatric surgery [133] since patients want to be eligible for the surgery and do not want to be excluded based on some inadequate eating habits. In addition, because several types of bariatric surgeries exist and have different impact on energy restriction and nutrient's absorption, the need to categorize individuals according to the type of surgery, more specifically post-surgery, should also be considered in the validation process. Some studies included in this review evaluated a cohort longitudinally and assessed dietary intakes pre-and post-surgery using the same dietary assessment tool, but none of them differentiated the validity of the tool to measure dietary intakes prior to and after surgery.
This review has strength and limitations. It showed an important lack of studies that used a tool validated in bariatric population and the need to conduct research to address this concern. Indeed, a considerable number of studies used a dietary assessment tool that had been previously validated in a non-bariatric population, such as the Swedish Obese Subjects study questionnaire. Furthermore, only a few studies included in this review specifically aimed to validate the dietary assessment tool used to assess dietary intakes in bariatric population, another indicator of the lack of literature. The interpretation of the results remained difficult considering the limited availability of information regarding the validation process and conclusions about the validity in most studies, and the high level of methodological differences between studies.
Identification of the most relevant dietary assessment tools validated prior to and after bariatric surgery would allow to characterize dietary intakes more accurately while improving nutritional interventions among these patients. Validity of dietary assessment tools should be tested for total daily energy intake and in terms of diet quality. Indeed, quality of dietary intakes of patients targeted for bariatric surgery can impact their risk of developing nutritional deficiencies after the surgery [6] and the success of their weight loss [93]. Moreover, web-based and technology-assisted assessment methods have opened the way to a new wave of self-administered automatic tools [8,9]. Considering that the web-based 24HR has been associated with reduced desirability bias compared to standard administrated questionnaires at least in the general population [8], such tools could be an interesting approach to assess dietary intake in bariatric population. The potential benefits and risks associated with these web-based tools need to be evaluated in bariatric population. More studies about the validation of dietary assessment tools in bariatric population are needed, taking into account potential biases in this population.

Conclusions
In conclusion, few studies included in the review validated their dietary assessment tool. Additional studies are needed in order to develop valid and robust dietary assessment tools among bariatric population. These tools are essential in evaluating efficacy of nutritional interventions conducted in this population.