Inclusion criteria

Exclusion criteria

1. Studies that included tumours from other colorectal locations but did not analyse the left hemicolectomy separately or for which data from the left hemicolectomy were not extractable were not included.

2. Benign colorectal disease or emergency surgery will be excluded.

3. No splenic flexure mobilisation will also be excluded.

4. Non-comparative studies and non-English publications were excluded.

Study selection

We systematically searched the PubMed, Web of Science, Medline, EMBASE and Cochrane Library databases for all literature comparing laparoscopic and open surgical approaches for left hemicolectomy from inception to 5 November 2021. Searches were carried out using medical subject headings and free text words in combination with the search strategy. We used the following keywords: “colon cancer”, “left hemicolectomy”, “laparoscopy” and “open”. All possible forms of these keywords will be used to ensure the comprehensiveness of the search. Additionally, we enriched our retrieval results with several methods, such as the similar articles function in PubMed, cross-checking references of the retrieved literature, searching ClinicalTrials (https://www.clinicaltrials.gov/), etc.

Search terms for PubMed

#1 ((((((((((((((((((((Colonic Neoplasms) OR (Neoplasm, Colonic)) OR (Neoplasms, Colonic)) OR (Colon Neoplasms)) OR (Colon Neoplasm)) OR (Neoplasm, Colon)) OR (Neoplasms, Colon)) OR (Cancer of Colon)) OR (Colon Cancers)) OR (Colon Cancer)) OR (Cancer, Colon)) OR (Cancers, Colon)) OR (Cancer of the Colon)) OR (Colonic Cancer)) OR (Cancer, Colonic)) OR (Cancers, Colonic)) OR (Colonic Cancers)) OR (Colon Adenocarcinoma)) OR (Adenocarcinoma, Colon)) OR (Adenocarcinomas, Colon)) OR (Colon Adenocarcinomas)

#2 open surgery

#3 ((((((((((((((((Laparoscopy) OR (Celioscopy)) OR (Peritoneoscopy)) OR (Surgical Procedures, Laparoscopic)) OR (Laparoscopic Surgical Procedure)) OR (Procedure, Laparoscopic Surgical)) OR (Procedures, Laparoscopic Surgical)) OR (Surgery, Laparoscopic)) OR (Laparoscopic Surgical Procedures)) OR (Laparoscopic Surgery)) OR (Laparoscopic Surgeries)) OR (Surgeries, Laparoscopic)) OR (Laparoscopic Assisted Surgery)) OR (Laparoscopic Assisted Surgeries)) OR (Surgeries, Laparoscopic Assisted)) OR (Surgery, Laparoscopic Assisted)) OR (Surgical Procedure, Laparoscopic)

#4 (left hemicolectomy) OR (left colectomy)

#5 #2 AND #3

#6 #1 AND #4 AND #5

The management of the literature search records will be carried out in EndNote V.X9.1. Two authors (QD and JZ) independently performed an initial screening of the titles and abstracts of the search results and assessed the eligibility of the articles. After removing duplicates and irrelevant literature, the two authors will assess the eligibility of the articles according to the inclusion criteria after reading the full text of the remaining articles separately. Any controversial points arising during this process will be referred to a third author (LY) and discussed until the dispute is resolved. The specific literature screening process will be summarised in a flow diagram.

Data extraction

Data to be collected, such as study details (first author, year of publication, study design, follow-up period, type of outcome), patient demographics (age, sex, American Society of Anesthesiologists score, tumour stage, etc), and the outcomes of interest mentioned above will be consolidated into a piloting spreadsheet. Additionally, we will extract the effect estimates of the outcome of interest for statistical analysis. If there were multiple representations of the data, we preferred to use the data after adjusting for confounding factors. To reduce bias and reduce errors in data extraction, the same two investigators (QD and JZ) independently extracted data from the included literature, cross-checked after extraction and disagreements were resolved by discussion and, if necessary, by asking a third author (LY) to resolve. Because this analysis was based on the intention-to-treat principle, all patients who were converted from the laparoscopic group to the conventional open surgery group remained in the laparoscopic group for analysis. We will also use sensitivity analyses to assess the impact of including studies that do not report intention-to-treat on overall outcomes.

There are currently several RCTs, such as COST, CLASICC, ALCCaS and COLOR, comparing laparoscopic and open colectomy, and we believe that inclusion of their data would enhance the quality of our evidence for this study. We will be sending emails to the authors of these trials asking for stratified data on left hemicolectomy. Meanwhile, for the missing data of other studies, we will also send an email to ask for.

Statistical analysis

Statistically, it is not possible to combine the median with the mean value, and only data expressed as the mean and SD can be used for meta-analysis. In this study, we will not use the median to estimate the mean, as other studies have done, because we believe this would not be worth the cost. The weighted mean difference or standardised mean difference and corresponding 95% CI were used for the analysis of continuous variables. The dichotomous variables were analysed using risk ratio values with 95% CIs. Considering the characteristics of survival analysis, we will first attempt to extract survival analysis-related data from the included studies and then calculate the pooled HR. HR and 95% CI will be extracted directly from the article, and if not reported in the article, we consider using software such as Engauge Digitizer to obtain the required data from Kaplan–Meier curves following the method provided by Parmar et al.26 Finally, the obtained data will be integrated into the spreadsheet designed by Tierney et al27 to calculate the HR and 95% CI. If the data were insufficient or the HR was not available for other reasons, then the pooled OR values of OS and DFS were calculated separately.

Statistical heterogeneity among the studies was calculated by χ² test and I² test.28 We considered that high heterogeneity existed if the value of p<0.1 or I² >50%. When the heterogeneity was 0, the fixed-effects model was used, and when the heterogeneity was between 0% and 50%, the random-effects model was used. We will conduct subgroup analyses, based on different study design types and meta-regression so that we can explore the potential causes of heterogeneity and reduce it as accurately as possible when heterogeneity exceeded 50%. If the heterogeneity is too high, then qualitative analysis was performed. Sensitivity analysis will be performed to determine the robustness of the results by sequentially excluding one study at a time. A p<0.05 was considered statistically significant. Software such as RevMan V.5.4 and STATA V.16 will be used for statistical processing. Publication bias will be estimated by visual assessment of funnel plots if ≥10 studies are available. If the extracted data are not suitable for pooling, a systematic narrative synthesis will be presented in textual form.

Risk of bias assessment

Quality assessment will be carried out by two authors (QD and JZ), and discrepancies will be resolved through discussion. If consensus was not reached, then the third author (LY) was consulted for arbitration. The risk of bias in RCTs will be assessed using the Cochrane Risk of Bias Tool,29 which includes six aspects: randomisation, allocation concealment, application of blinding, integrity of outcome data, selective reporting, and other biases. For each, we will use high risk, low risk or unclear risk to assess the results. The methodological quality of non-randomised controlled trials will be evaluated using the Newcastle-Ottawa Scale,30 which consists of three aspects: patient selection, comparability of cohorts, and assessment of outcome. The total score is 9 stars, and each article is classified as low quality (0–5 stars) or high quality (6–9 stars). The final results will be summarised in a table.

Evidence quality evaluation

The quality of evidence for each outcome will be evaluated using the Grading of Recommendations Assessment, Development and Evaluation system,31 with four levels: high, moderate, low and very low.

Patient and public involvement

Since this study is a secondary study based on other studies, there will be no direct patient or public involvement in this study.

Ethics and dissemination

Because no patients were involved, ethical approval was not required. The final results of this research will be submitted to a peer-reviewed journal or presented at relevant conferences, and any deviations from this protocol will be recorded and explained in the final report.