Step 3. Search, selection, data abstraction and synthesis
All questions selected by the GSG and GDG were converted into a specific
question format using the PICO (patient, intervention, comparison and
outcome) model. Selection criteria for each question were determineda priori based on the patient population and exposures in each
question. Only comparative studies (systematic reviews of comparative
studies, randomized controlled trials, cohort studies, case-control
studies, and cross-sectional studies) were eligible for inclusion. A
systematic search of PubMed through February 1, 2024, was performed to
find studies addressing the questions included in the CPG. Additionally,
we manually reviewed the reference lists of all relevant systematic
reviews and included studies to find additional eligible studies. The
titles and abstracts of all identified references were reviewed by a
clinician and a methodologist from the GSG. Studies identified for
full-text review were then reviewed by all members of the GSG, and any
reasons for exclusion were noted.
Data from the included studies were extracted by two members of the GSG.
The risk of bias in the included studies was assessed using the
appropriate tool for each study design. The Cochrane RCT tool was used
for randomized controlled trials. The Newcastle-Ottawa tool was used for
cohort and case-control studies. Data on outcomes from similar studies
were extracted and pooled when appropriate by using the random-effects
model. The main outcomes of interest were overall survival, mortality,
completeness of resection, local recurrence, complications, and
intensity of therapy, as important outcomes to inform recommendations.
All comparative analyses were performed using the RevMan software
package. Pooled estimated for each outcome and certainty of evidence
were summarized using Grading of Recommendations, Assessment,
Development and Evaluation (GRADE) Summary of Findings tables.