Bariatric surgery is considered the gold-standard treatment for inducing significant weight loss, which can alleviate obesity-related complications in people with severe obesity [1]. In recent years, the demand for bariatric surgery has increased with the rising prevalence of obesity [2]. From 2011 to 2019, the total number of bariatric surgeries performed worldwide rose from 158,000 to 256,000 [2]. However, weight non-response (insufficient weight loss) or weight recurrence (weight regain) are reported in as many as 1 in 2 patients at 2 years, and 3 in 5 patients at 12 years after surgery [3, 4]. These are associated with reduced quality of life, re-occurrence of obesity-related complications, deteriorated health, and ultimately, escalated health care costs and mortality [3,4,5,6,7].

Previous studies have explored the factors contributing to weight non-response or significant weight recurrence following bariatric surgery, and identified patients’ post-operative diet to be a key modifiable determinant of weight status post-surgery [5, 8]. In addition, Zarshenas and colleagues’ systematic review reported poor diet quality among patients at least one year after bariatric surgery [9]. These findings highlighted the role of dietary intakes in the management of weight after bariatric surgery, and the importance of nutritional interventions to improve the long-term diets of patients post-surgery.

At present, nutritional management guidelines for patients after bariatric surgery either focus on the diet texture progression within the first month post-operatively, or have based their overall diet recommendations, after texture progression (10–35% or >60 g/d protein, 30–70% or >130 g/d carbohydrates, 20–35% fats, and 5 serves of vegetables a day), on extrapolated evidence from non-surgical populations and/or small studies with weak evidence [10,11,12]. As bariatric surgery leads to significant changes in patients’ anatomy, physiology, and tolerance of specific foods and food volumes, dietary advice intended for the general population may not be suitable for patients post-bariatric surgery [13,14,15,16]. Depending on individual needs, stomach capacity, surgical outcomes and time after surgery, there may be a change in macronutrient requirements over-time. More evidence is needed to drive consensus and inform dietary recommendations for the medium to long-term post-surgery [9].

Over the past decade, there have been several studies examining the potential influence of dietary intake on weight loss and/or weight recurrence post-bariatric surgery. Therefore, the aim of this systematic review was to synthesize the available evidence regarding associations between post-surgery dietary intake (macronutrient composition and food patterns) and weight outcomes at least one year and longer after bariatric surgery.


The protocol for this review is registered in PROSPERO (CRD42021264120) and was conducted and reported as per the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) checklist [17].

Search strategy

A systematic search was conducted in electronic databases: CINAHL, Cochrane, Embase, MEDLINE and Scopus for human studies published in English language from all years up to and including June 2021. The basic search strategy was (post OR after OR following) “Bariatric surgery” OR “weight loss surgery” OR gastric bypass” OR “gastric sleeve” OR “sleeve gastrectomy” AND diet* OR nutrition* OR eat* OR macronutrient* OR ‘postoperative diet’. The full search strategy and subject headings used for each database are available in Supplementary Material 1.

Record screening and eligibility criteria

Title and abstract screening were conducted by HC. Full text screening was completed by members of the research team independently in groups of two (Group 1: HC and LR, Group 2: HC and ES, Group 3: HM and JM) using systematic review software, Rayyan (Rayyan Systems Inc., Cambridge) [18]. Any uncertainties around study inclusion were raised to the research team for discussion until consensus was reached. Eligible study designs were systematic reviews, meta-analyses, interventional and cohort studies of adults (18 years of age or above) without pre-existing life-threatening conditions who received bariatric surgery (any type). Outcomes included at least one post-surgery dietary variable (reported at least 1 macronutrient, food group or dietary pattern) that was compared to any post-surgery weight outcome(s). Studies were included if their analyses involved: (1) weight outcomes in response to a prescribed diet; (2) a comparison of weight change between participant groups and their concurrent diet; and/or (3) correlation analysis of association between weight outcomes and diet. Studies were excluded if: (i) any participants were pregnant or breastfeeding (ii) the participants were inpatients, (iii)a post-surgery progression diet (i.e. texture modified or less than 1 month post-surgery), (iv) measures of adherence to a specific diet or diet preferences without indicating the exact diet being followed, or (v) if intake analysis were limited to energy, micronutrients, test meals, single meals, or supplements only.

Data extraction

Data extraction was performed by HC and cross-checked by LR. Information extracted included country of publication, participant characteristics including pre- and post-surgery health parameters, assessment timelines, dietary assessment method, dietary intakes (energy, macronutrients, food/dietary patterns) and weight outcomes at any timepoint: excess weight loss, odds of reaching >50% excess weight loss, total weight loss, average monthly weight loss, initial weight loss, BMI loss, risk of obesity remission i.e. BMI < 30 kg/m2, presence of weight recurrence as defined by the study authors as exceeding a nominated percentage of weight gain after nadir weight (lowest weight post-surgery), odds of weight recurrence, risk of weight recurrence. Key findings were those regarding any association between dietary intake variables and weight outcome(s), and/or comparisons between intervention and control groups or groups of participants achieving/not achieving pre-defined weight outcome(s) and dietary intake. Statistical analysis results reported in studies were extracted, including correlation co-efficient, odds ratio, hazard ratio and 95% Confidence Interval and interpreted according to conventional standards established by Cohen [19].

Quality assessment

Individual studies were matched to the Australian National Health and Medical Research Council (NHMRC) levels of evidence and grades for recommendation guidelines depending on study design [20]. Studies were first assigned a level based on the potential of the study design to adequately answer the defined research question(s): with level I being the highest level of evidence, followed by II, III-1, III-2, III-3, and IV (lowest) [20]. The Scottish Intercollegiate Guidelines Network (SIGN) Risk of Bias checklists for cohort, case-control and controlled trial study designs were then used to determine the risk of bias (low risk, acceptable risk, and high risk) [21]. The level of evidence and risk of bias of each study were assessed by HC and cross-checked by LR independently with blinding.

Data synthesis

To synthesize reporting differences between studies, dietary and outcome data were treated as follows: Intake assessments reported over different timeframes were grouped and presented as three main post-surgery time-dependent categories: up to one year (≤12 months); between one and two years ( > 12 to <24-months); and two years or longer (≥24 months). Studies were then further categorised within each assessment timeframe according to the dietary variables reported (macronutrient composition or food pattern). Within these dietary categories, all associated weight outcomes were included regardless of follow-up timeframes (equivalent or longer than the dietary timeframes) and grouped as two main outcome categories: (1) weight loss (including excess weight loss (EWL), initial weight loss (IWL), total weight loss (TWL), body mass index (BMI) loss), and obesity remission (i.e. reaching a BMI of <30 kg/m2); or (2) weight recurrence measures of odds ratio, hazard ratio, or presence of weight recurrence from nadir weight that had exceeded a study-specified percentage.

Data analysis

The bodies of evidence regarding the associations between post-surgery weight outcomes and individual macronutrients and food patterns were assessed and graded using the NHMRC Guidelines [20]. In accordance with these guidelines, the bodies of evidence were assessed based on five components, and each component has been graded based on a set of standard criteria [20].

Recommendations were then deduced from these bodies of evidence and graded based on the combined gradings from each graded component. The possible grades for recommendations were: Grade A (body of evidence can be trusted to guide practice); Grade B (body of evidence can be trusted to guide practice in most situations); Grade C (body of evidence provides some support for recommendation(s) but care should be taken in its application); and Grade D (body of evidence is weak, and recommendation must be applied with caution). The grading of evidence was conducted by HC, cross-checked by LR, then reviewed and achieved consensus with ES, JM, AB, and CL.


The screening process of this review is outlined in Fig. 1. A total of 5923 records were identified and title/abstract screened after the removal of 1495 duplicates. A total of 260 records were retrieved for full text screening, and 36 papers were included in this review. Reasons for exclusion were listed in Fig. 1.

Fig. 1: PRISMA Flow Diagram.
figure 1

Summary of the identified, screened, and included studies from databases and registers.

Study characteristics

The key characteristics of the 36 included studies are summarized in Table 1 [22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57]. The retrieved study designs included nine intervention studies (5 randomised controlled trials, 3 non-randomized controlled trials, 1 pre-post-intervention) [29, 32, 34, 36, 37, 47, 52, 53, 55]. Interventions delivered typically consisted of lifestyle modifications with or without a prescribed diet plan, where participants’ dietary intakes (macronutrient composition and/or food group pattern) were recorded and compared against pre-intervention values and/or between study intervention or control groups. In total, 27 observational cohort studies (8 prospective and 19 retrospective) were included [22,23,24,25,26,27,28, 30, 31, 33, 35, 38,39,40,41,42,43,44,45,46, 48,49,50,51, 54, 56, 57]. These studies compared the diets of participants grouped according to their weight status, and/or conducted direct tests of association between dietary variable(s) and weight outcome(s).

Table 1 Characteristics of included studies.

A range of bariatric surgery procedures were reported among the included studies, with some individual studies reporting more than one type. The most common procedures were Roux en Y gastric bypass (RYGB) (N = 22 studies) [24, 25, 28, 30,31,32,33, 36, 37, 39, 40, 43,44,45,46, 48,49,50,51,52, 56, 57] and sleeve gastrectomy (SG) (N = 13 studies) [22, 25,26,27, 34,35,36, 40, 43, 46, 51, 53, 54]. Others included laparoscopic adjustable gastric banding (LAGB) (N = 4 studies) [29, 34, 52, 55], vertical banded gastroplasty (VBG) (N = 3 studies) [38, 41, 47], gastric bypass (did not specify type) (N = 3 studies) [23, 38, 41], gastric banding (did not specify type) (N = 1 study) [38], and anastomosis gastric bypass (OAGB) (N = 1 study) [25]. One study did not specify the type of bariatric surgery involved [42]. Dietary intake assessment tools (from most to least common) were food records (3–7 days), followed by 24-hour recalls, food frequency questionnaires, and other lifestyle or behavior surveys collecting dietary data as part of the tool [22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57]. Sample size ranged from 20 to 1610 participants (total N = 5065) with drop-out rate from 0 to 61%. The reported mean age of participants ranged from 32.7 ± 1.6 years to 57.3 ± 8.7 years. The proportion of female participants totaled across all studies was 75% (excluding the single study where gender ratio was not reported) [54]. The majority of studies did not mention the proportion of participants with insulin resistance or type 2 diabetes mellitus (T2DM). Even though some studies did report on the percentage of participants with at least one co-morbidity, they did not specify the exact proportion of participants with T2DM [33, 57]. Among studies that reported on them, the rate of participants with insulin resistance or T2DM mellitus ranged from 11.6%-52.5% [22, 25, 35, 36, 40]. Timepoint of measurement or assessment of post-surgery diets ranged from 6 months to 12 years. Significant post-surgery weight outcomes were reported in all studies from 1 to 12 years follow up.

Quality assessment

The level of evidence and risk of bias outcomes are reported in Table 1. There were no Level I studies, nine Level II studies [24, 37, 38, 43, 47,48,49, 52, 53], 22 Level III studies [22, 23, 25,26,27,28,29, 31, 33,34,35,36, 39, 40, 42, 44, 46, 50, 51, 54,55,56], and five Level IV studies [30, 32, 41, 45, 57]. Risk of bias assessment deemed 13 studies to be low risk [22, 23, 25, 28, 35, 38,39,40,41, 43, 48,49,50], 14 studies were acceptable risk [24, 26, 27, 29, 30, 33, 34, 36, 37, 44,45,46, 51, 57] and nine studies were high risk studies [31, 32, 42, 47, 52,53,54,55,56]. Studies’ level of evidence and risk of bias were both taken into consideration during quality assessment.

Study findings

Table 2 summarises the findings from individual studies, of which, 27 studies compared the concurrent diets of study groups (per intervention status or weight outcome) and 16 studies conducted direct tests of association between dietary variable(s) and weight outcome(s). The types of diet being assessed, key findings, and limitations for each included study. Study findings are presented according to the timeframe when dietary intakes were measured post-surgery ≤12-months, between 12 and 24 months, and ≥24 months.

Table 2 Study findings on the relationship between weight change and dietary intakes.

Diet ≤12-months post-surgery and weight loss

Sixteen studies described diets up to 12-month and observed weight losses up to ten-years [24, 25, 27, 29, 30, 34, 36, 38,39,40, 47,48,49, 52,53,54]. Regarding carbohydrate, significant inverse associations with weight losses were supported by three observational cohort studies (total N = 1888, up to 10-years follow-up) [30, 38, 40]. However, one RCT (prescribed protein enriched, low-carbohydrate diet vs prescribed normal protein and carbohydrate diet, with high reported adherence) and four observational cohort studies showed no significant associations (total N = 368, up to 8-years follow-up) [24, 25, 39, 48, 53]. Regarding protein, significant positive associations with weight loss was supported by five observational cohort studies (total N = 2232, up to 10-years follow-up) [30, 38, 40, 48, 49]. On the contrary, one RCT (prescribed protein enriched, low-carbohydrate diet vs prescribed normal protein- and carbohydrate diet, with high reported adherence), one pseudo-RCT (prescribed protein-enriched diet vs no prescribed diets, with poor reported adherence), and four observational cohort studies did not report significant associations with weight loss (total N = 444, up to 8-years follow-up) [24, 25, 27, 29, 39, 53]. Regarding fat, significant inverse associations with weight losses were supported by two observational cohort studies (total N = 1799, up to 10-years follow-up) [38, 40]. However, five observational cohort studies did not report any significant associations (total N = 400, up to 8-years follow-up). A single RCT that involved a lifestyle intervention did not result in any significant between-group differences in macronutrient intakes nor weight changes[52].

Food pattern, specifically fruit and vegetable intakes, were reported in three studies (one RCT, two non-randomized controlled trials) [34, 36, 47]. All studies reported significantly greater weight losses in intervention groups compared with controls after lifestyle interventions [34, 36, 47]. However, no significant differences in fruit and vegetable intakes were noted between intervention and control groups within the single RCT [47]. Although between-group intake differences were not measured within the two non-randomized controlled trials, both studies reported significant increases in fruit and vegetable intakes in their intervention groups when compared to pre-intervention intakes [34, 36]. As both studies included a physical activity component, it was not possible to attribute outcomes to diet alone [34, 36]. There was inadequate information provided on other food patterns including dairy, meat and grains, as they were not measured in the single RCT [47], whereas the two non-randomized controlled trials reported no significant changes in their intakes compared to pre-intervention [34, 36]. Lastly, a single cohort study observed the effect of varying purine contents of the diet (exact food group intakes not reported), which did not show any significant association with post-surgery weight [54].

Diet ≤12-months post-surgery and weight recurrence

No studies examined weight recurrence during this period, presumably due to the short timeframe to enable observation of weight recurrence.

Diet between 12- and 24-months post-surgery and weight loss

Six studies described diets between 12- and 24-months post-surgery and observed weight losses up to five years post-surgery [31, 37, 41, 51, 52, 55]. Diet was also described and compared with weight recurrence at two years post-surgery by a single cohort study [42]. Regarding carbohydrate, significant inverse associations with weight losses was supported by one observational cohort study (total N = 75, with up to 2-years follow-up), whereas a lack of association was suggested by one non-randomized controlled trial and one observational cohort study (total N = 290, with up to 5-years follow-up). Additionally, Lindroos et al. assessed the potential impact of different types of carbohydrate intakes and reported greater weight loss for participants with higher intakes of mono- or di-saccharides and lower intakes of polysaccharides, though the potential impacts of total carbohydrate intake was not assessed [41]. Findings for protein included one observational cohort study (N = 75, with up to 2-years follow-up) suggesting positive associations [31], one observational cohort study (N = 375, with up to 2-years follow-up) suggesting inverse associations [41], and two studies (one non-randomized controlled trial and one observational cohort) suggesting no significant associations with weight loss (total N = 290, with up to 5-years follow-up) [51, 55]. Regarding fat, only one observational cohort study (N = 375, up to two-years) [41] supported positive associations with weight loss whereas three studies (one non-randomized controlled trial, two observational cohorts) reported having no significant associations (total N = 365, up to 5-years) [31, 51, 55]. A single RCT that involved a lifestyle intervention did not result in any significant between-group differences in macronutrient intakes nor weight outcomes [52]. The relationship between post-surgery weight loss and food pattern was assessed by a single RCT [37], providing monthly home delivered meals with a personalized menu plan of 4 serves vegetables, 2–4 serves meat, and 1–2 serves grains per day. The intervention resulted in greater weight loss than a control group with no delivered meals or prescribed diet and may indicate some benefits for similar types of intervention or prescribed menu plans [37].

Diet between 12- and 24-months post-surgery and weight recurrence

Macronutrient composition, food pattern, and weight recurrence at up to 18-months post-surgery were described by a cohort study [42]. The findings of this study favored a diet with daily intakes of 3–5 fat, fruit, and vegetable exchanges for less weight recurrence, though the portion size of each food exchange was not reported [42].

Diet at ≥24-months post-surgery and weight loss

Fourteen studies described dietary intakes after 24-months post-surgery and weight loss up to 12 years post-surgery [23, 26, 32, 33, 35, 39, 43,44,45, 47, 50, 51, 56, 57]. The absence of association was supported by 11 out of 11 observational cohort studies regarding carbohydrate and fat intakes (Total N = 1305, with up to 12-years follow-up) [23, 26, 33, 35, 39, 43,44,45, 50, 51, 56], and by 12 out of 12 observational cohort studies for protein intake (Total N = 1402, with up to 12-years follow-up) [23, 26, 33, 35, 39, 43,44,45, 50, 51, 56, 57]. A pre-/post-interventional study of individuals who had experienced weight recurrence at 3 years post-surgery reported significant weight loss from a three-month diet with 45% carbohydrate, 35% protein, and 20% fat [32]. Despite high adherence, the researchers did not compare to pre-intervention intakes, and an incentivized physical activity component made attribution of results to diet alone impossible [32]. Regarding food pattern, the absence of association between weight loss and any core food group, except fruit and vegetables, was supported by 2 out of 2 observational cohort studies (total N = 147, with up to 5 years follow-up) [23, 33]. A positive association between weight loss and fruit and vegetables was suggested by one RCT, where a lifestyle intervention resulted in significantly greater weight loss and concurrent significantly higher fruit and vegetable intakes in the intervention group compared to controls (N = 30, with up to 3 years follow-up) [47]. However, the component of physical activity prevented the outcomes being attributed to diet alone [47]. Despite this, three cohort studies reported no associations between fruit or vegetable intakes with weight loss up to 9-years (total N = 244, with up to 9-years follow-up) [23, 33, 57].

Diet at ≥24-months post-surgery and weight recurrence

Eight observational cohort studies described dietary intakes after 24-months and observed weight recurrence up to 9 years [22, 23, 26, 28, 33, 35, 46, 57]. For macronutrient intake composition, the lack of association between weight recurrence with carbohydrate and protein intakes was largely consistent across studies: Carbohydrate (6 out of 6 cohort studies, total N = 410, with up to five-years follow-up [22, 23, 26, 28, 33, 46]); Protein (5 [22, 23, 28, 33, 57] out of 6 cohort studies [22, 23, 26, 28, 33, 57], total N = 366 out of 403, with up to 9 years follow-up). The findings regarding fat intakes were mixed, though the absence of association was supported by 4 [22, 23, 28, 33] out of 6 cohort studies [22, 23, 26, 28, 33, 35] (total N = 269 out of 392, with up to 7-years follow-up [22, 23, 28, 33]). Differences in the definition of significant weight recurrence across studies (i.e. >2% to >25% recurrence of weight [22, 23, 26, 28, 33, 35]) may partially account for the mixed results. Notably, in the cohort study with the longest follow-up (7-years) and strictest criteria for Group 1 (>50% excess weight loss at first year and maintaining <25% weight recurrence until 7-years), higher fat intakes, in addition to total energy intake, were linked to a 4-fold increased risk of weight recurrence within a cox hazard regression model. Group 1 participants also reported significantly lower intakes of fat than Group 2 participants (>25% weight recurrence at 7-years).

Regarding food patterns, a lack of association between weight recurrence and any core food group was reported by 4 out of 4 observational cohort studies (total N = 326, with total follow-up of 8.9 years) [23, 28, 33, 47], except for fruit, where 1 out of the 4 studies was suggestive of an inverse association between fruit intakes and weight recurrence [28]. Diet quality score, measured using the Brazilian version of the Healthy Eating Index, was assessed by one cohort study that showed a weak inverse association with the odds of weight recurrence with a higher diet quality score (achieved by having a balanced daily servings of grains, vegetables, fruits, beans, meats, dairy products, fats and oils, sugar and sweets, restricted intakes of saturated fats and cholesterol, and having a high variety of foods within the diet) [28].

Grading of evidence and recommendations

Tables 3 to 6 provide a summary of the results assessed using the NHMRC body of evidence framework [20].

Table 3 Grading of Evidence for associations between dietary intakes ( ≤ 12 months) and weight loss (up to ten-years) (15 studies)*.
Table 4 Grading of Evidence for associations between dietary intakes (Between 12 and 24months) and weight loss (up to five-years – 4 studies) or weight recurrence (up to 18-months – one study)*.
Table 5 Grading of Evidence for associations between dietary intakes ( ≥ 24 months and beyond) and weight loss (up to twelve-years) (13 studies)*.
Table 6 Grading of Evidence for associations between dietary intakes ( ≥ 24 months and beyond) and weight recurrence (up to 9-years) (6 studies)*.

The body of evidence for diet ≤12 months post-surgery and weight loss is summarized in Table 3. Overall, the body of evidence was inconsistent regarding the presence or absence of any positive or negative associations between weight loss and macronutrient composition. While the findings for food group patterns appeared to be consistent, co-variables related to study design prevented the effect from being attributed to dietary intake alone. Therefore, no recommendations could be drawn.

The body of evidence for diet between 12- and 24-months post-surgery and weight loss and weight recurrence is summarized in Table 4. The body of evidence was inconsistent regarding the presence or absence of any associations between weight loss and diet, and there was inadequate information to draw conclusions on whether, and in what way, dietary intakes were associated with weight recurrence. Therefore, no recommendations could be drawn.

The body of evidence for diet ≥24 months post-surgery and weight loss is summarized in Table 5. The overall body of evidence was consistent for the absence of any significant associations between weight loss and all macronutrient intakes and food patterns. This body of evidence contributed to the recommendation that long-term diets post-bariatric surgery can be individualized with flexibility as to macronutrient and food pattern composition (Grade B –the body of evidence can be trusted to guide practice in most situations).

The body of evidence for diet ≥24 months post-surgery and weight recurrence is summarized in Table 6. The body of evidence was consistent for the absence of any significant associations between weight recurrence and carbohydrate, protein, and food patterns. However, the study findings for the association between weight recurrence and fat intakes was inconsistent, and an inverse association with diet quality was reported by just one cohort study. Therefore, two recommendations can be made: long-term diets post-bariatric surgery can be individualized with flexibility to macronutrient and food group composition due to lack of association (Grade C); and that a high diet quality can be encouraged in long-term post-surgery diets for reduction of weight recurrence (Grade D).

Study findings for the dietary components not able to be graded

It was not within the scope of this review to have an in-depth analysis of the associations between energy intake and weight outcomes, due to not being included in the search term for systematic retrieval, and food patterns beyond core food group intakes due to not being consistently reported (e.g. specific food intakes like sandwiches, packaged foods or sweets that were defined differently across studies). These findings are presented as part of Table 2.


To our knowledge, the current review is the first to systematically synthesize existing literature reporting on associations between macronutrient composition, food patterns and weight outcomes post-bariatric surgery. Our review found that current evidence is related to the assessment of diet in distinct timeframes: 12 months, between 12- and 24-months and ≥24 months after surgery. Relationships between macronutrient intake and weight loss up to 24 months were inconclusive due to inconsistent findings between several studies of varying quality. Very few studies reported on food patterns, where the lack of studies and poor study design also contributed to inconclusive evidence. However, at 24 months and longer the evidence was consistent across several study findings for no association between macronutrient intake or food group pattern and weight loss, and between carbohydrate, protein and food group pattern and weight recurrence. An inverse association between weight recurrence and diet quality was reported from a single cohort study. Therefore, the existing body of evidence, overall, does not support a specific macronutrient composition or food pattern for optimal weight outcomes after bariatric surgery.

Previous research has suggested or attempted to establish the best macronutrient composition to support post-surgery weight loss [9, 40]. However, our review found the evidence base for associations with macronutrient composition and food pattern relied on less robust study designs and many contained confounding factors resulting in mixed results over different timeframes. The interpretation of association, therefore, was limited due to the lack of high-quality study designs. Despite this, the most notable phenomenon was that any relationship between macronutrients, food group pattern and weight outcomes, seemed to have lost significance after 24-months post-surgery, as studies that assessed dietary intakes after 24-months post-surgery found no associations whilst there was a mix of studies reporting the presence or absence of associations before 24 months [22, 23, 26, 28, 30, 31, 33, 35, 38, 39, 41,42,43,44,45,46, 50, 51, 55,56,57]. More well designed RCTs or prospective cohort studies are required to explore the potential associations between post-surgery weight outcomes with short- to medium- term dietary intakes (<24-months).

Although studies reporting on energy intake post-surgery were not systematically searched as part of this review, an inverse association with weight loss and positive association with weight recurrence was largely supported by those included studies that assessed energy intake [22,23,24,25,26, 28, 30, 31, 33, 35, 38,39,40,41, 43,44,45,46, 50, 51, 56]. Previous studies have debated the roles of long-term dietary restriction and food malabsorption in post-bariatric surgery weight outcomes [58,59,60,61,62,63,64]. However, there are known contributing factors that make it difficult to ascribe weight outcomes solely to energy intake, for example, the known prevalence of under-reporting dietary intakes among this population [59, 62, 63] and/or differences in energy balance as a result of individual basal requirements and physical activity levels [62, 65]. In Novais et al., only those with >50% excess weight loss reported an energy intake that was significantly lower than their estimated energy requirements. Furthermore, Benson-Davies et al. [59]. found participants who maintained their weight loss achieved a 2100-kilojoule deficit in energy balance (through a combination of lower energy intake and higher step counts) when compared to participants who experienced weight recurrence. Similarly, Forbush et al. suggested energy expenditure, rather than energy intake alone, to be a predictor of weight loss [62]. Given the limited reporting of physical activity levels in this population [65], energy balance may be an important focus for future research and practice to support long-term weight loss/maintenance post-bariatric surgery.

For many, the ultimate goal of achieving and sustaining weight losses after bariatric surgery is to improve the management of obesity-related complications. With the modest direct associations between post-surgical weight and diet composition found in this review, it may be sensible to place higher emphasis on exploring the impacts of diet composition on non-weight parameters of health, and specifically, those that are associated with the management of obesity-related complications. In the current review, studies that have reported non-weight clinical parameters mostly focused on the associations between body composition, quality of life and protein intakes [29, 37, 49, 53, 55], due to protein’s role in muscle mass maintenance, believed to be beneficial for weight loss maintenance. However, the overall evidence from the current review is mixed and does not support a high-protein diet for weight loss outcomes long-term post-surgery. This finding aligns with those of two previous systematic reviews focusing on the associations of protein intakes and body composition, which resulted in inconclusive findings [66, 67]. Larger and higher quality studies that place greater emphasis on obesity-related health parameters, in addition to weight, are required to explore how post-surgery diets may influence these parameters. Also, a recent study found no significant correlations between the extent of post-surgery weight loss and improvements in cardiovascular risk factors [68]. While this information was from a single study, it is apparent that future studies would benefit from placing additional emphasis on the relationship between post-surgery dietary intakes and clinical parameters, such as cardiometabolic health, in patients with or without weight non-response or recurrence at long-term post-bariatric surgery.

With the need to explore the relationships between obesity-related health parameters beyond weight, and in the absence of associations between weight non-response or recurrence with individual macronutrients and food groups, focusing on the overall quality of individuals’ diets may provide a more holistic approach to improving patients’ post-surgery eating pattern. In this review, Da Silva et al. was the only study retrieved that assessed diet quality in a systemic manner using an established tool, and suggested an ongoing weak association with weight re-occurrence after two years post-surgery [28]. Other identified studies that mentioned participants’ diet quality either did not use any established tool to measure diet quality [23, 33, 42], or did not adequately report participants’ dietary intakes [69,70,71,72,73], thus preventing comparisons across studies and, therefore, not included in the review as part of the data analyses for diet quality [23, 33, 42] or excluded for not meeting eligibility criteria [69,70,71,72,73]. Additionally, it is not known how useful or valid are established diet quality tools for use in this population [74, 75]. At present, existing research undertaken to derive and validate diet quality indices does not specifically consider bariatric surgery population [28, 70,71,72,73,74,75,76]. The potential changes in physiology and/or dietary needs after bariatric surgery may reduce the generalizability of diet quality indices from the general population to the bariatric surgery population. Without an accepted definition or measurement of high diet quality following bariatric surgery, it is not possible to compare the potential influence of diet quality on different obesity-related health parameters between studies and over-time. Future studies are needed to establish a consensus or criterion for the definition of a high-quality diet after bariatric surgery and to develop and validate a bariatric-specific diet quality measurement tool.

The differences in surgical procedure are often speculated as a contributor to different weight outcomes. Only one included study compared outcomes between different surgical types (SG versus RYGB), with no significant differences observed for weight outcomes and macronutrient composition [43]. Although findings for LAGB studies may not be generalizable across different surgery types, as they did not involve the resection of the gastro-intestinal tract, only 4 out of the 36 included studies involved post-LAGB patients. However, the differences in dietary intakes across surgical procedures were not the main interest of the current review. Since most of the included studies in this review reported single procedures only, and key findings did not differ across single or mixed surgeries or different surgery types, separate reporting of findings per surgery type was not attempted by this review. More studies may be needed to assess whether differences in surgical procedure can moderate the association between diet and post-surgery weight, and if so, whether such differences are significant enough to warrant separate dietary advice per surgery type. Another commonly speculated contributor to different weight outcomes is concomitant medications that may influence body weight (e.g. anti-depressants, corticoids, insulin) and the presence of T2DM, which may result in less weight loss than participants without T2DM [77]. In our review, only four of the included studies excluded participants taking these medications, but it was unclear whether the type(s) of medications considered in each study were the same [22, 23, 27, 28, 37]. Similarly, the rate of participants with insulin resistance or T2DM were only reported in a few studies and was often combined with other co-morbidities for assessment of the effects of number of co-morbidities on post-surgery weight [22, 25, 35, 36, 40]. The single study that assessed the potential impact of T2DM on excess weight loss reported no significant associations [25]. Another study reported higher rate of weight recurrence among those with insulin resistance, but acknowledged participants’ higher body fat percentage or BMI may have contributed in part to the results [22]. Future studies may need to clarify whether participants with medications or conditions that may influence participants’ weight are included in their statistical analysis, in order for higher confidence that the recorded weight changes are attributable to participants’ lifestyle.

The strengths of this review include the emphasis on long-term outcomes after bariatric surgery, and being the first review to systematically examine any potential associations between weight and dietary intakes after bariatric surgery. Limitations include the lack of inclusion of energy intake as a search criterion and the inability to discriminate the results by the type of surgery. The application, interpretation and generalization of results required subjective judgements to arrive at final recommendations. The retrieved studies were limited by a generally poor evidence base and poor consistency due to their methodological flaws and heterogeneity in study design, including different units of measure and different methods of assessment of association. Most studies did not account for dietary changes made before or after their dietary assessments, which may have played a role in weight status measured at time of study. Furthermore, the definitions used for post-surgery weight outcomes (weight non-response or weight recurrence) lacked standardization. These limitations align with the findings of a recent review that called for a research-derived definition of clinically significant post-surgery weight recurrence [78]. This lack of standardization limited comparisons between studies and the grading of evidence to support specific recommendations, particularly those assessing short-term dietary intakes (up to 24 months). The unknown contributors to weight outcomes such as physical activity, genetics and individual motivation, especially in studies with longer durations, were poorly reported within the included studies. Although the treatment of known confounders by the study authors were taken into consideration during risk of bias assessment, their interference on the potential clinical impact of the synthesized evidence was inevitable. Generalizability and applicability of studies to other bariatric contexts was also affected by individual study designs that may have incurred risk of selection bias, or contained inadequate information (e.g. adherence to prescribed diets, or dietary intakes prior to study intervention), to enable reproducibility of study outcomes to external populations. Moreover, the retrieved studies had a high proportion of female participants. Lastly, the results were prone to inherent limitations of dietary assessment methodologies such as recall bias and under-reporting. While there are standardized methods for excluding implausible reporting [79], only two studies in our review attempted to address the known risk of under-reporting by comparing two methods of self-report (a 72 h recall versus a three day food record) to evaluate compliance to prescribed diets, which reported high compliance and similar energy intakes [53, 54]. The remaining studies simply acknowledged this limitation, but did not identify nor exclude potential under-reporters [24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52, 55,56,57]. Even though one study hoped to use the Goldberg cut-offs to assess implausible intakes among their participants within 12-months post-surgery, said method assumes weight stability and was deemed unsuitable for use for patients who had yet to reach weight maintenance stage after surgery [29]. Importantly, there are no standardized energy intake cut-offs for this population to assist with assessment of implausible intakes, which further complicated the exclusion of implausible reporters. As this limitation was discussed by some of the included studies as a potential contributor to the lack of association found between dietary intakes and weight outcomes, future studies on patients post-bariatric surgery may benefit from including the assessment of implausible intakes using estimated energy requirements, supplementing food diaries with weighed food records or photographed pictures of participants’ meals, adjusting for total energy intake based on two methods (e.g. a food frequency questionnaire and three 24 h food recalls), or other standardized methods, to help improve accuracy of dietary assessments.


In view of the overall finding that there is a lack of evidence to support the strong association of any particular diet composition or pattern with weight loss or recurrence, it is not unreasonable to suggest that long-term dietary advice post-bariatric surgery can be individualized with flexibility as to the macronutrient and food composition, or that a focus on diet quality may be beneficial for the reduction of weight recurrence over the longer term. However, these recommendations should be taken with care until higher quality studies further confirm and strengthen these statements. Well-designed prospective trials with standardized reporting and monitoring of dietary intakes including energy intake/expenditure, coupled with the considerations of potential moderators of weight, such as the type of surgery, may be beneficial to help clarify the relationships between weight outcomes and dietary intakes.