This systematic review will be conducted based on a review protocol registered with the International Prospective Register of Systematic Reviews (PROSPERO registration CRD42020199509). The protocol for this review has been prepared in accordance with the Preferred Reporting Items for Systematic Reviews and Meta- Analysis for Protocols (PRISMA-P) Guidelines.(11) (Table 1, additional file 1) The aim of this study is to compare the non- operative management of resectable colorectal cancer in elderly patients to surgical resection in respect to patient survival and quality of life through a systematic review of literature including randomized control trials, prospective and retrospective cohort studies, consecutive series, and cross-sectional studies.
The study selection criteria have been establish according to the PICOS (Population – Intervention – Comparator- Outcomes – Study design)
Primary outcome: life expectancy, survival
Secondary outcomes: quality of life, need for future treatments (e.g palliative surgery, emergency surgery, radiation, etc.), resource utilization, change in level of care (e.g need of nursing home, etc)
Covariates thought to be associated with disease outcomes will also be collected such as tumor factors (TNM staging, location of the tumor), patient factors (comorbidities, ASA score, smoking status, functional status, etc.), treatment factors (chemotherapy, radiation therapy, type of surgical treatment, postoperative hospital stay and morbidity)
Study designs: We will include randomized control trials, prospective and retrospective cohort studies, consecutive series, cross-sectional studies, case series. Case reports editorial letters and review articles will be excluded.
Search Strategy and Data Sources
We will search MEDLINE, EMBASE and the Cochrane Database of Controlled Trials from 2000 to April 2020 (inclusively) using a predetermined search strategy developed with the assistance of a health information specialist with expertise in systematic reviews and clinical expert in the field of colorectal surgery (R.M). The search strategy will be comprised of Medline subject headings and key words. The search strategy will be peer reviewed by a second health information specialist using the PRESS framework to ensure robust capture. (12)We will also search grey literature to include abstracts presented at relevant society meetings from the past three years and ongoing key websites (clinicaltrials.gov, etc.) to explore ongoing and upcoming studies relevant to our review.
Study Selection and Data Collection
Two authors (R.H and R. S) will complete abstract screening independently and in duplicate using Mendeley systematic review manager software. Full text screening will be completed in duplicate by two authors (R.H and R.S). To ensure consistent application of selection criteria, the two reviewers (R.H and R.S) will carry out a pilot exercise comparing their study selection. This will be done during each stage of selection, after review of the first 50 abstracts at stage 1 and the review of the 20 full texts at stage 2. All disagreements will be settled by a third party reviewer (R.M). The study selection process will be summarized in a PRISMA flow chart.
Two authors (R.H and R.S) will complete data extraction independently and in duplicate using a standardized data extraction form implemented in Excel. A pilot extraction exercise of three studies will be performed to ensure consistency in approach between reviewers. Data elements to be collected will include those related to basic publication characteristics (including year, journal, authorship list and country), study methods (including design and elements necessary for risk of bias appraisal), population studied (enrollment criteria and key demographic measures including age, gender, BMI, comorbidities, ASA, preoperative nutritional status, anemia, frailty index, smoking, TNM stage of the cancer, tumor location), intervention compared (non-operative – observation and palliative procedures, chemotherapy, radiation therapy, surgical intervention – laparoscopic or open) and outcome data (survival/ life expectancy, quality of life, need for future treatment, change in care needs, postoperative morbidity and mortality). Binary outcomes will be collected as n(%) and continuous outcomes will be collected as mean (SD).
Risk of Bias Assessment
The Methodological Index for Non- Randomized Trials (MINORS) criteria will be used to assess the risk of bias of all included studies. (13)The MINORS criteria is a validated bias assessment tool for non randomized studies. Comparative studies are evaluated on twelve criteria (Table 2) and each item is assigned a score between 0 to 2 (0 = not reported, 1 = reported by inadequate, 2 = reported and adequate). Studies are then scored out of a global ideal of 24. Two authors (R.H and R.S) will complete risks of bias assessments using the MINORS criteria. Disagreements will be settled by a third party reviewer (R.M). Findings will be narratively summarized in the text of the final study with full assessments available in an appendix. Sensitivity analyses based upon findings from the assessments (e.g focused upon high methodologic quality studies) will be considered in the event that studies are pooled for analysis.
Data Synthesis and Statistical Analysis
The literature search for this review may identify both comparative studies and single group studies. Single group studies will be focused upon reporting experience with either operative or non- operative treatment, whereas comparative studies will be assessed and analyzed separately from the latter. If multiple comparative studies are identified that directly compared (i.e simultaneously collected and compared endpoints) curative resection to non-operative management of colorectal cancer in the elderly, whether prospective or retrospective, a meta-analysis will be performed using random-effects, inverse variance approach. For group of studies presenting single arm study data related to outcomes for either operative or conservative management, random effects meta-analyses will be performed to assess the median survival of the individual study arm, in addition to evidence of benefits and harm; narrative summaries of these findings will be prepared with the realization that comparisons based on these data are weaker based upon the nature of the data. Clinical and methodologic heterogeneity will be assessed by the research team for the set of included studies to establish their degree of similarity in terms of patient populations and study methods. Statistical heterogeneity will be assessed for meta-analyses using the I2 statistic, with a value of 50% or higher suggesting the presence of important heterogeneity. Forest plots will be presented for all syntheses, and pooled estimates will be removed from these plots in cases where the presence of high statistical heterogeneity cannot be addressed.