Effect of weight‐loss diets prior to elective surgery on postoperative outcomes in obesity: A systematic review and meta‐analysis

Summary This systematic review investigated the effects of weight‐loss diets before elective surgery on preoperative weight loss and postoperative outcomes in people with obesity. Electronic databases were searched from inception to May 2021. Inclusion criteria were prospective cohort or randomised controlled studies that compared effects of weight‐loss diets to standard care on postoperative outcomes in adults with obesity awaiting surgery. Participants with cancer or undergoing bariatric surgery were excluded. Data on preoperative weight change, length of stay, postoperative complications and patient‐reported outcome measures were extracted and synthesised in meta‐analyses. One randomised controlled trial involving total knee arthroplasty and two that investigated general surgery were eligible that included 173 participants overall. Each study compared low‐calorie diets using meal replacement formulas to usual care. There is very‐low‐quality evidence of a statistically significant difference favouring the intervention for preoperative weight loss (mean difference [MD] −6.67 kg, 95% confidence interval [CI] −12.09 to −1.26 kg; p = 0.02) and low‐quality evidence that preoperative weight‐loss diets do not reduce postoperative complications to 30 days (odds ratio [OR] 0.34, 95% CI 0.08–1.42; p = 0.14) or length of stay (MD −3.72 h, 95% CI −10.76 to 3.32; p = 0.30). From the limited data that is of low quality, weight loss diets before elective surgery do not reduce postoperative complications.


Summary
This systematic review investigated the effects of weight-loss diets before elective surgery on preoperative weight loss and postoperative outcomes in people with obesity. Electronic databases were searched from inception to May 2021. Inclusion criteria were prospective cohort or randomised controlled studies that compared effects of weight-loss diets to standard care on postoperative outcomes in adults with obesity awaiting surgery. Participants with cancer or undergoing bariatric surgery were excluded. Data on preoperative weight change, length of stay, postoperative complications and patient-reported outcome measures were extracted and synthesised in meta-analyses. One randomised controlled trial involving total knee arthroplasty and two that investigated general surgery were eligible that included 173 participants overall. Each study compared low-calorie diets using meal replacement formulas to usual care. There is very-low-quality evidence of a statistically significant difference favouring the intervention for preoperative weight loss (mean difference [MD] À6.67 kg, 95% confidence interval [CI] À12.09 to À1.26 kg; p = 0.02) and low-quality evidence that preoperative weight-loss diets do not reduce postoperative complications to 30 days (odds ratio [OR] 0.34, 95% CI 0.08-1.42; p = 0.14) or length of stay (MD À3.72 h, 95% CI À10.76 to 3.32; p = 0.30). From the limited data that is of low quality, weight loss diets before elective surgery do not reduce postoperative complications. What is already known about this subject • The prevalence of obesity is increasing worldwide • Obesity is associated with greater risk of postoperative complications.
• Minimal research has been conducted on the effect of diet-based preoperative weight loss on postoperative outcomes.

What this study adds
• This review is currently the only systematic review in the English literature that has investigated the effects of diet-based weight-loss interventions in people with obesity undergoing elective surgery excluding bariatric surgery.
• Weight loss through diet interventions can be achieved before elective surgery.
• There is low-quality evidence that weight-loss diets do not reduce postoperative complications to 30 days.

| INTRODUCTION
Worldwide, the rate of obesity has almost tripled since 1975 with over 650 million adults, or 13% of the population, having obesity in 2016. 1 The rising incidence of obesity has resulted in more people classified with obesity (defined as a body mass index [BMI] of 30 kg/m 2 or greater) before undergoing elective surgery. For example, in the United States, the rates of extreme obesity (BMI ≥40 kg/m 2 ) among those undergoing knee arthroplasty have doubled from 1993 to 2003. 2 Increasing obesity is problematic because people with obesity are at a greater risk of certain postoperative complications, with greater health and financial burdens to the patient and the health system. 3 Specifically, obesity is associated with elevated risks for revision surgery, 4 wound complications, 4-6 venous thromboembolism, 6 pulmonary emboli, 4 urinary tract infection, 5 with worse long-term patientreported functional outcomes, 7 reduced mobility 7 and inadequate activity levels postoperatively; all of which increase risk of chronic disease. Consequently, health professionals recommend preoperative weight loss to improve postoperative outcomes. That said, obesity is not always associated with worse postoperative outcomes because for instance, lower mortality rates may occur among people with obesity when compared to people who are underweight. 8 Thus, it remains unclear whether weight loss before surgery should be recommended.
Recently, some health services worldwide have restricted criteria for undergoing elective surgeries based on weight. For example, the National Health Service in England reported that 31% of Clinical Commissioning Groups, who are responsible for service provision, have at least one mandatory policy on BMI level and weight management before elective surgery. 9 Similarly, the surgeons at Logan Hospital in Queensland, Australia declined to operate on people with obesity unless they lost 10% of their body weight. 10 In this case, a dietitian-led presurgical weight management programme was implemented for their patients to achieve this target. Weight-loss diets are a safe option, with potential cost benefits to the individual and health-care system. However, little is known about the evidence in support of weight-loss diets before any elective surgery to decrease adverse events postoperatively in populations with obesity. Thus, the objective of this review is to determine the effect of preoperative weight-loss diets on postoperative clinical and service outcomes in people with obesity undergoing elective surgery.

| METHODS
This systematic review was performed using methods from the Cochrane Handbook for Systematic Reviews of Interventions 11 and according to the preferred reporting items for systematic reviews and meta-analysis (PRISMA) guidelines. 12 It was registered prior to commencement on Open Science Framework in February 2020 (https:// osf.io/dgf3t) and PROSPERO (CRD42020154074).

| Inclusion criteria
Eligibility criteria for included studies are summarised in Table 1

| Study selection
Search results were merged using EndNote and duplicate records were removed. Two reviewers (NP, RB) independently screened articles for relevance and excluded irrelevant articles based on titles and abstracts. Multiple reports of the same study were identified, with the most recent version included for review. Full texts of the remaining articles were independently assessed by NP and RB for inclusion against selection criteria. Disagreements were resolved through discussion with a third reviewer (JN).

| Risk of bias assessment
Selected studies were independently assessed for methodological validity by two reviewers (NP, RB). The Cochrane Handbook's Risk of Bias (RoB) Version 2 checklist 13 was used to assess individual outcomes from randomised controlled trials (RCTs) according to five domains of bias (randomisation process, deviations from intended interventions, missing outcome data, measurement of the outcome, and selection of the reported result). A priori, it was planned to use the Cochrane Handbook ROBINS-I tool 14 to assess the risk of bias in prospective nonrandomised trials. Discrepancies between reviewers were resolved through discussion. All studies, regardless of methodological quality, underwent data extraction and synthesis where possible.

| Data extraction
Standardised items in an Excel spreadsheet were used by two independent reviewers (NP, RB) to record the following extracted data from eligible studies: title; author; year of publication; journal; study design; setting; participant characteristics; recruitment procedures utilised; trial size; preoperative weight-loss intervention characteristics; details of the control; type of surgery; follow up or study duration; outcomes; outcome measurements; data analysis methods; details needed for risk of bias; author contact details; funding source. Data discrepancies were resolved through discussion. Three corresponding authors were contacted via email for further information and all responded. All analyses were performed on intention-to-treat data.

| Assessment of heterogeneity
Clinical and methodological heterogeneity was assessed for diversity in participants, interventions, outcomes, study characteristics, and risk of bias for included studies to determine whether meta-analysis was appropriate.
Statistical heterogeneity within each meta-analysis was assessed using the I 2 statistic. 15 Due to the small number of studies eligible for inclusion in The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).
b Downgraded by one level due to high risk of bias in one study due to deviation from intended intervention and missing outcome data and some concerns in the selective reporting of outcomes in the remaining studies.
c Downgraded by one level because the sample size does not meet the Optimal Information Size (OIS).
d Downgraded by one level due to high risk of bias in one study due to deviation from intended intervention and missing outcome data.
e Downgraded by one level due to inconsistency (high I 2 ).
f Downgraded by one level due to high risk of bias in the largest study due to selective reporting of outcomes and high risk of bias in one study due to deviation from intended intervention and missing outcome data. this review the decision was made to include all studies in the metaanalysis regardless of the I 2 statistic so long as the studies were similar with respect to clinical and methodological characteristics.

| Assessment of publication bias
Published reports were compared against trial protocols to evaluate potential for publication bias. While assessment of publication bias was planned, there were insufficient studies (<10 studies) to construct a funnel plot or perform Egger's regression test.

| Data synthesis
Data were analysed using Review Manager version 5.4.1 software from the Cochrane Collaboration. 16  3 | RESULTS

| Study inclusion
A total of 15 547 references were retrieved through searches of electronic databases with another 568 references from other sources.
After excluding duplicates and irrelevant articles from titles and abstracts, 16 articles remained (Figure 1). Thirteen articles were excluded (Table S1)  were unpublished with no data available for inclusion, 26,27 two were retrospective studies, 28,29 one was an editorial commentary 30 and one was an RCT with a weight-loss diet in both the control and intervention groups. 31 Three RCTs met the eligibility criteria. [32][33][34] No prospective non-RCTs were eligible for inclusion. No additional studies were identified by reference searching, yielding three articles that satisfied inclusion criteria and provided quantitative data for analysis.

| Methodological quality
Risk of bias assessments was completed for three outcomes: postoperative complications, preoperative weight change and length of stay (LOS) ( Figure 2). One study demonstrated some concerns across all outcomes due to selective reporting of outcomes. 32 Similarly, one study had some concerns for postoperative complications and LOS, and high risk of bias for preoperative weight change due to selective reporting of outcomes. 34 One study demonstrated high risk of bias across all outcomes due to deviation from the intended intervention, missing outcome data and selective reporting of outcomes for preoperative weight change. 33 Regarding the quality of the dietary interventions, no studies clearly stated a dietitian was involved in designing the intervention.
Whilst all studies assessed dietary compliance with various methods, none utilised a validated tool to assess participant dietary compliance.

| Characteristics of included studies
The three studies included were RCTs published in English. Two studies involved participants undergoing general surgery 32,33 and one examined participants undergoing total knee arthroplasty for knee osteoarthritis. 34

| Review findings
The details for all primary and secondary outcomes are described in Tables S2 and S3. Meta-analysis was undertaken where data were available as the included studies were similar with respect to participant and weight-loss intervention characteristics and study design.

| Postoperative complications
The primary outcome, postoperative complications to 90 days, was Burnand et al. 32 Hollis et al. 33 Liljensøe et al. 34 Study

| Length of stay
Acute hospital LOS was reported in all studies comprising data from 156 participants. The pooled effect estimate suggested that LOS was shorter in the intervention group, but the difference did not reach T A B L E 4 Characteristics of weight-loss diets from included studies Study Burnand et al. 32 Hollis et al. 33 Liljensøe et al. 34 Intervention Very-low-calorie diet Very-low-calorie diet Low-energy liquid diet   Limitations of the current review included the limited generalisability of findings due to the small number of identified studies. Second, the effect of weight loss, as opposed to weight-loss programmes, on postoperative outcomes could not be distinguished.
Preoperative weight change data for each participant was unavailable for subgroup analysis which was planned to compare participants who lost weight to those who did not regardless of group allocation.

DATA AVAILABILITY STATEMENT
Data relevant to this systematic review are found in the article and supplementary file. Further data will be made upon request from the corresponding author.