Methodological issues in assessing change in dietary intake and appetite following gastric bypass surgery: A systematic review

Summary Gastric bypass surgery is an effective long‐term treatment for individuals with severe obesity. Changes in appetite, dietary intake, and food preferences have all been postulated to contribute to postoperative body weight regulation, however, findings are inconsistent. The aim of this systematic review was to evaluate the current literature on changes in dietary intake and appetite following gastric bypass surgery, in the context of the methodology used and the analysis, interpretation, and presentation of results. Four databases were systematically searched with terms related to “gastric bypass surgery,” “appetite,” and “dietary intake,” and 49 papers (n = 2384 patients after gastric bypass) were eligible for inclusion. The evidence indicated that only a reduction in overall energy intake and an increase in postprandial satiety are maintained beyond 6‐month post‐surgery, whereas relative macronutrient intake and premeal hunger remain unchanged. However, available data were limited by inconsistencies in the methods, analysis, presentation, and interpretation of results. In particular, there was a reliance on data collected by subjective methods with minimal acknowledgment of the limitations, such as misreporting of food intake. There is a need for further work employing objective measurement of appetite and dietary intake following gastric bypass surgery to determine how these mechanisms may contribute to weight regulation in the longer term.


| INTRODUCTION
Gastric bypass surgery is a successful established treatment for individuals with severe obesity (body mass index [BMI] ≥ 40 kg/m 2 or a BMI > 35 kg/m 2 with one or more associated comorbidities that would improve with weight loss), with weight loss superior to nonsurgical lifestyle interventions 1 and maintained by the individual in the long term. 2 In addition to weight loss, the surgery leads to improvements in comorbidities including type 2 diabetes mellitus (T2DM) 3 and adverse cardiovascular events. 4 The mechanisms behind weight loss and its maintenance following surgery are complex and not fully understood. 5 Previously, weight loss was thought to be mediated primarily through stomach restriction and nutrient malabsorption, but these physical changes do not fully explain the successful long-term body weight regulation. Postulated mechanisms contributing to the long-term success of gastric bypass surgery include changes in circulating gut hormones, 6 attenuated resting energy expenditure through the management and preservation of lean mass, 5,7 and changes in taste sensitivity thresholds, [8][9][10][11] which may affect food preferences and selection.
Postoperatively, patients typically report changes in appetite and dietary intake that may facilitate long-standing weight maintenance.
Overall daily energy intake (EI) is significantly reduced, with a recent meta-analysis estimating an average postoperative reduction of 4.39 MJ/day, 12 which is sustained in the longer term. [12][13][14] It is unclear whether these changes in EI are facilitated through changes in macronutrient contribution to the diet, and reported findings regarding dietary macronutrient contribution to EI, at least in the short term, are equivocal. 15 Animal studies have reported that rats avoid high-fat stimuli 16 and consume proportionally less fat in their diet following gastric bypass surgery. [16][17][18][19][20][21][22] Although some studies have reported similar changes in humans 23,24 these proportional shifts in relative fat intake are not a consistent finding. Proposed mechanisms accounting for these possible shifts in macronutrient contribution to the diet of patients post-gastric bypass surgery include changes in taste 25 and diminished preference for palatable (high-fat, high-sugar) foods. 26 Preserving or increasing dietary protein postoperatively may both increase satiety and protect against the loss of lean mass, 27 whereas reducing dietary fat may reduce overall dietary energy density (ED) and thus EI. 24 Changes in body fat set points in the longer term may also explain how reduced EI is maintained. 28 The regulation of appetite occurs through physiological changes in the neuroendocrine system, 29 as well as through sensory, cognitive, and physical processes that indicate feelings of hunger and satiety. Patients after gastric bypass surgery report improved postprandial satiety compared with control participants with obesity. 30 The proposed mechanisms include changes in appetite-regulating hormones, with postprandial appetite-suppressing glucagon-like peptide 1 (GLP-1) and peptide YY (PYY) secretion increased postoperatively. 31,32 The assessment of appetite may be through objective (EI), subjective (e.g., Visual Analogue Scales [VASs]), or biochemical (e.g., circulating gut hormones) measures. The VASs are the most commonly utilized measure, which, while easy to administer, can be difficult to quantify and compare. Appetite sensations are subjective, with no two individuals experiencing these feelings in the same way, 33 and so responses given on a VAS are not absolute values (e.g., a score of 40 mm on a hunger scale cannot be assumed to represent exactly half the hunger sensation of a score of 80 mm). Although new developments in scaling methodology have occurred over the last two decades such as the generalized labeled magnitude scale (gLMS), 34,35 which attempts to standardize for differential lifetime sensory experience, they are not yet widely adopted in bariatric surgery research with some exceptions. 11 Nevertheless, such scales still depend on subjective verbal report and would be reinforced by complementary implementation of more direct objective measures. 15 Studies often utilize subjective measures of both fasted and postprandial hunger and satiety in conjunction with objective measures of hormone levels to assess the impact of gastric bypass surgery.
Patients have reported changes in fasted hunger as early as 2 days postoperatively, which correlate with changes in gut hormone profiles in the short term, 36 but this is less consistent in the longer term. 31,37 Postprandial hunger and satiety are more variable and equivocal. The characterization of the effects of gastric bypass surgery on both fasted and postprandial appetite may lead to further insight into the underlying mechanisms of the success of bariatric surgery and help to predict which patients are likely to respond better to treatment.
Subjective measures (dietary recalls and dietary records) are also the method of choice when measuring free-living dietary intake in bariatric populations. There is a tacit acceptance of the validity of EI data despite extensive evidence that these methods are highly susceptible to misreporting, with underreporting particularly frequent in participants with obesity. 38 Although discrepant findings may be, in part, attributed to individual differences, it is likely that the current reliance on self-reported measures of appetite and EI are contributing to inconsistent findings. Previous reviews have also identified differences in the reporting of findings such as weight loss, 39 patient-reported outcomes, 40 and clinical outcomes. 41 Although there have been efforts to standardize the reporting of these outcomes, 42 the inconsistencies prevent quantitative synthesis and make results difficult to compare. Thus, disparities in the methods, analysis, and presentation of outcomes relating to both appetite and dietary intake in bariatric patients may, at least partially, contribute to the confusion and lack of clarity observed in the current literature.
The aim of this review is to evaluate the current literature on changes in dietary intake and reported appetite following gastric bypass surgery, specifically in the context of the methodology used and the analysis, interpretation, and presentation of results from these studies.

| Search strategy
The Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines 43 informed the protocol for the systematic searching of the literature. This review was registered with the PROSPERO database (CRD42019126302).
Four databases (Embase, Medline, Scopus, and Cochrane) were systematically searched. Keyword searches were population based ("Gastric Bypass" OR "Roux en Y" OR "Bariatric surgery") in combination with appetite ("Appetite" OR "Eating" OR "Hunger" OR "Satiation") and dietary intake ("Energy Intake" OR "Dietary Fats" OR "Dietary Proteins" OR "Dietary Carbohydrates") including their cognates and synonyms. Gray literature searches of conference outputs and reference lists were also completed.

| Data selection and extraction
A single reviewer (TR) completed an initial screen of all titles and abstracts retrieved from the systematic searches. A second reviewer (AD) then completed an independent screen to ensure no eligible papers were excluded. These two reviewers independently assessed the selected papers for full text screening and determined which should be included for review. A third reviewer (a member of the authorship team) was available to adjudicate any differences in opinion between the two reviewers, of which there were none.
The electronic database searches initially returned 6757 studies.
After duplicates were removed, there were 5711 papers that underwent title and abstract screening. Gray literature searches returned an additional 10 papers to be included. Following this initial screen, 170 titles were left for full text review. Forty-nine papers were considered suitable, having met the inclusion criteria, and data were extracted for review. After completing the full screen, retained papers were assessed using the Critical Appraisal Skills Programme. 44 Figure 1 provides an overview of the literature searches.
Primary outcome data extracted for review from included papers were changes in fasted hunger, postprandial appetite, EI, and dietary composition following gastric bypass surgery. Secondary outcomes extracted (where available) were changes in circulating gut hormones, food selection, dietary ED, and micronutrient intake. Data extracted on methodology included measures used, sample size (and %follow-up), study duration, length of fast prior to appointment, composition of preload, influence of misreporting, and subject selection.
In total, 2384 participants were included, with sample size ranging from 5 to 294. Participant completion rate varied from 26% to 100%.
Approximately two thirds of participants were female (n = 1,657), with 10 studies measuring changes in females only. Recruitment for each study was from one site (e.g., one hospital), except for one study that recruited across two sites (Norway/Sweden). 45 No studies measured changes in participants who had undergone one-anastomosis gastric bypass and only one assessed postsurgical changes in adolescents.
Only four papers specified that medications that may affect intake/ metabolism/weight loss were an exclusion criterion or that data were collected on medications, with none presenting results on medication usage. The duration of follow-up ranged from 1-month to 8-year post-surgery.

| Daily EI
Postoperative changes in EI were reported in 30 (61%) studies.
Change was mostly expressed as total EI/day, although it was also reported as EI adjusted for BMI (n = 1) and EI/kg body weight (n = 2).
All studies that presented change in EI/day reported a reduction in EI up to 8-year post-surgery. When presented in terms of body size, findings were more inconsistent. In adolescents, changes in EI (adjusted for BMI) decreased up until 3-month post-surgery, although changes were no longer significant by 1-year post-surgery. 46 In adults, two studies observed a decrease in EI/kg up to 6-month post-surgery. 47,48 However, at 1-year post-surgery, they diverged, with one study observing a consistent reduction in EI/kg, 47   included studies specifically excluded fluid intake, and only Laurenius et al. 24 were specific about not measuring water intake in their calculation of ED (kJ/g).
Approximately three-quarters of the studies that presented change in macronutrient intake as g/day or g/kg/day reported a Only three studies measured sugar independently of total carbohydrate, but sugar was variously defined as sucrose, 50 added sugar, 51 or dietary sugar. 52 When presented as g/day, sugar intake decreased. 52 However, when presented as %EI, findings were inconsistent and reported as either a decrease from pre-surgery intake 50 or no different 51 at 1-year post-surgery.
Dietary fiber intake (g/day) was assessed in three studies, [51][52][53] and all reported a significant decrease in dietary fiber from pre-to 1-year post-surgery.
Dietary ED was measured in only one study. 24 Intake of high-ED foods (>16.7 kJ/kg) decreased at 6 weeks postoperatively, with a reciprocal increase in very-low-ED foods (<2.5 kJ/g). Only the increase in consumption of very-low-ED foods was maintained at 1-year post-surgery, but changes were no longer evident at 2 years.
This was not reflected in overall dietary ED (kJ/g), which remained significantly decreased at 2-year post-surgery.

| Food selection
Alongside the assessment of EI and macronutrient intake, food selection was also assessed in several studies. Most frequently examined were intakes of sweet foods, categorized as "desserts" 37,54 or as "sweets/sodas" and "milk/ice cream," 23,55 with a decrease in the consumption of these foods observed in most studies (n = 3) up to 2-year post-surgery. When followed up at 36 months, Brolin et al. 23 reported that the decrease in intake was no longer evident. This initial decrease was also observed by Miller et al. 52 when assessing intake of "sweets" and "sweetened beverages," but the return to preoperative intake levels was observed earlier (3 months postoperatively).
Different food groups were also assessed, but because of the lack of homogeneity between studies on how foods were grouped and  42 . Data sourced from papers summarized in Table 2 defined, it is difficult to draw meaningful conclusions from the data.
For example, fruit/vegetable intakes were variously categorized as fruit and vegetables, fruits, or vegetables or as "healthy" foods.

| Micronutrient intake
Eight studies measured micronutrient intake and status. Most frequently measured were iron, 52,56,57 calcium, 46,56,57 and potassium. 52,58 All studies reported a decrease in the intake of these micronutrients in line with reported decreases in EI.
When micronutrient status was assessed, rates of deficiency did not change post-surgery with several studies reporting that supplementation stabilized nutrient status following surgery. 49,59,60

| Subjective assessment of appetite
Nineteen studies measured subjective appetite in 362 (range:  patients. Studies measuring subjective appetite were smaller and shorter in duration compared with studies evaluating dietary intake. In addition, all patient follow-up was in the short term only, and the outcomes were considered at 3-month intervals.

| Premeal hunger
Twelve studies used VAS to assess premeal hunger ( No study observed an increase in premeal hunger. In the early postoperative stages, over half reported a decrease in premeal hunger, but from 6 months onwards, studies were reporting no change in premeal hunger. In the absence of long-term follow-up studies, the significance of these findings is unclear.

| Postprandial appetite
Eighteen studies measured postprandial appetite sensations using assessments of satiety, hunger, fullness, and/or desire to eat either as individual measurements or in variable combination (Table 5).
Postprandial satiety was the most frequently assessed sensation   Table 2 T A B L E 4 Summary of methods and outcomes of papers measuring subjective fasted appetite changes in patients after gastric bypass from pre-to post-surgery, organized by follow-up duration The format and formulation of the test meal given varied between studies. Either mixed meals (n = 6), liquid meals (n = 9), or snack foods (n = 2) were given to participants as a pre-load, with energy content ranging from 1035 to 4707 kJ. Only one study adjusted meal size according to time postoperatively, 37 and relative protein (6% to 41%), fat (0% to 45%), and carbohydrate (40% to 89%) composition of the meals showed large variation. To assess the possi- ( Figure 4C). All studies that employed low-protein loads demonstrated an increase in postprandial satiety, whereas there was no definitive effect following a high-protein load. There were also no clear differences between the administration of a high-versus low-fat or highversus low-carbohydrate load. The influence of the test meal is an important consideration when interpreting these data, but more information is required in bariatric patient groups.

| Objectively measured appetite
Three studies used direct measures of EI to objectively measure appe-

| DISCUSSION
The overall findings from this systematic review demonstrated that only a decrease in total EI and an increase in postprandial satiety remain different from preoperative values for more than 6-month post-surgery, whereas relative macronutrient intake and premeal hunger remain unchanged. Given the paucity of follow-up data on patients for longer than 3 years, conclusions cannot currently be drawn on the impact of these changes on the longer term weight trajectory following gastric bypass surgery.
As with other systematic reviews evaluating outcomes following gastric bypass surgery, the findings of this review are constrained by the quality and consistency of the available evidence. [39][40][41] Most Percentage of follow-up time points measuring changes in postprandial satiety from preoperative values in patients after surgery. Data grouped by relative macronutrient content of the test meal and sourced from papers summarized in Table 5 notably, an overreliance on subjective methodology, the lack of standardization in the measurement and presentation of data, and the shortage of longer term studies (>3 years) make it difficult to draw meaningful conclusions from the available data.

| Measurement of dietary intake
All studies used subjective self-reported measures to assess EI. These measures are based on the tacit assumption that data obtained by these methods are a valid representation of habitual intake. However, misreporting of EI, specifically underreporting, is a limitation of these methods and is particularly prevalent in individuals with obesity. 38 A major concern is the failure to acknowledge the phenomenon of mis- The reasons why individuals with obesity are more likely to underreport are not well understood, but it probably represents an interplay between cognitive and behavioral processes. 70,71 Underreporting can be conscious or subconscious and may be affected by negative social attitudes toward carrying excess weight and subsequent guilt about the quantity or type of food consumed. 72 Macronutrient-specific underreporting is difficult to assess as relative macronutrient intake is highly interrelated, but it is conceivable that foods and macronutrients with perceived negative health connotations (e.g., high-fat, high-sugar foods) are underreported, whereas those with perceived positive attributes (e.g., high-protein foods, fruits, and vegetables) are overrepresented. However, at present, there is no way of verifying whether EI underreporting is also associated with macronutrient-specific misreporting.
Available methods for calculating misreporting were developed for weight-stable populations. [73][74][75] Whereas previous work has attempted to assess the efficacy of these methods in preoperative bariatric surgery candidates, 76 38 The inability to measure accurately dietary intake in free-living individuals is an intrac- Another challenge when using subjective methods of dietary assessment is the inability to measure portion size accurately. The difficulties that face all individuals when documenting portion size are known, 78 and a significant reduction in postoperative portion sizes 79 means that methods that rely on predefined portion sizes (e.g., artificial food models) may not accurately represent reduced portion sizes post-surgery. Less than half of studies specified the use of a tool to improve the accuracy of portion size reporting, and although there is little research evaluating the efficacy of these tools in the bariatric surgery context, the methods employed should be flexible enough to accommodate fluctuating portion sizes.
A major focus of this review was to evaluate changes in relative macronutrient intakes as these may be more indicative of changes in food preferences, which may influence long-term weight loss and maintenance. Although some studies have also presented macronutrient data in absolute (g/day) values, 12 ideally, data should be presented as both absolute and relative values to aid the interpretation of findings.
Inconsistencies in the presentation of results were also evident in the measurement of food selection. In the limited studies that measured these changes, food groups were researcher defined, and both the definition and reporting of these changes were highly variable.
Intake was reported as either servings/day, %EI, or %patients classified as "frequent" (>4 days a week) consumers.
Finally, when sugar intakes were reported, the definition used was not consistent or comparable between studies. Only three studies measured sugar intake independently of total carbohydrate, [50][51][52] and by not considering sugar intake separately to carbohydrate, important dietary shifts may have been obscured. For example, Kenler et al. 55 reported a significant decrease in sugar-sweetened beverage intake that had no corresponding impact on relative total carbohydrate intake but did not report changes in relative sugar intake which may have been present.

| Measurement of appetite
Although the subjective assessment of appetite using VASs may, in theory, be easy to administer, their limitations need to be acknowledged, including difficulties in quantification as scores are not representative of absolute values. 80 Most studies reviewed combined subjective measures with either biochemical assessments or independently observed EI, but these objective measurements did not always correspond with subjectively reported changes suggesting that VAS may not be sensitive to changes in appetite sensations in post-surgery patients.
When measuring premeal hunger sensations, inconsistencies in the measurement protocols make cross-study comparisons difficult.
The duration of fast prior to measurement ranged from 5 to 12 h, with one study not stipulating a period of fasting at all. 63 Antecedent diet was also not standardized or recorded by any study, despite previous work in nonsurgical populations that observed a significant difference in the measurement of premeal hunger following consumption of different antecedent diets. 81 Physiological changes following gastric bypass surgery may further exaggerate the effect antecedent diet may have on premeal hunger, but this effect has yet to be evaluated.
In the measurement of postprandial appetite sensations, there were notable inconsistencies between studies in the format and formulation of the test meal presented. The effect of different test meals on both objectively and subjectively measured appetite has not been well documented in patients after bariatric surgery, but in nonsurgical populations, meal structure has been observed as impacting on subjectively assessed appetite. 82 The texture and formulation of test meals offered have also been observed to influence appetite hormone responses, 83

| Other methodological considerations
Other methodological issues that may influence the outcome of stud- Current dietary advice given to patients is not uniform, and so it is conceivable that differences in the information received may affect patient behaviors and, in turn, results.
Participant retention in studies was highly variable, with some studies reporting a completion rate of less than 30%. Lack of adherence to clinical follow-up is associated with poorer outcomes, 84 and although attrition of patients after gastric bypass has not been evaluated in the context of dietary intake or appetite measurements, it could be surmised that those who drop out are not adhering to postoperative recommendations. Those who are lost to follow-up may not experience comparable postoperative changes, and so findings may be obscured or exaggerated. There is a lack of transparency in the documentation of reasons for participant dropout. 85 As most studies in this review were observational, there are strong arguments that the development of study protocol guidelines specifically for bariatric research (similar to what exists for RCTs 86 and observational studies 87 ) would enhance reporting and ensure comparability of results.
The disproportionate representation of studies with short-term follow-up is highly likely to be contributing to inconsistencies in the current literature. Patients progress through diet stages (liquid, pureed, soft) in the first 3-month post-surgery, 88 and typically postoperative weight loss is initially large before plateauing at 18-24 months' post-surgery. 89 The assessment of appetite in particular is mostly carried out only up until 6-month post-surgery and is most unlikely to provide an informed evaluation of the impact on longer term body weight regulation.

| CONCLUSION
In conclusion, after gastric bypass surgery, patients have a profound decrease in EI, which appears to be accompanied by an increase in postprandial satiety. Moreover, these postoperative consequences appear to be robust and maintained up to 3 years after gastric bypass surgery. However, only tentative conclusions can be drawn from the available literature as there is a paucity of studies with objectively validated assessment of changes in dietary intake and appetite in patients after gastric bypass in the medium and long term. Inconsistencies in the methods, analysis, interpretation, and presentation of results are highly likely to be contributing to the confusion and remain major obstacles in bariatric research. The critical gaps in understanding the dynamics of food selection and intake following bariatric surgery will only be filled by the application of fit-for-purpose methodology and by reaching a consensus on reporting criteria.