Single-dose preoperative glucocorticoids in major gastrointestinal surgery: protocol and statistical analysis plan for a systematic review, dose response meta-analysis and trial sequential analysis of randomized clinical trials
Year: 2026
Type: study-registration
Abstract: Status at registration: the searches, study selection, data extraction, risk-of-bias assessment and analyses have been completed, and the manuscript is ready for submission. Statement on timing This protocol is registered retrospectively, after the analyses. It reproduces the review question, eligibility criteria, data extraction items and statistical analysis plan that the review team finalized before any pooled estimate was calculated. The statistical analysis plan was written after data extraction. Every amendment and deviation is listed in section 12, with its reason. 1. Review question and objectives In adults undergoing elective major gastrointestinal resection, does a single systemic glucocorticoid dose given before incision or during surgery, compared with placebo or no glucocorticoid, reduce postoperative complications, and does the effect depend on dose? Objectives: 1. To estimate the effect of a single preoperative glucocorticoid dose on postoperative complications, infections, mortality, length of stay and other clinical outcomes. 2. To assess whether the effect depends on dose, both across trials (dose in dexamethasone equivalents) and within trials that randomized different doses. 3. To rate the certainty of the evidence with GRADE and estimate, with trial sequential analysis, whether the accrued information is sufficient. 2. Rationale A single preoperative glucocorticoid dose attenuates the inflammatory response to surgery. Doses in randomized trials range from antiemetic doses of dexamethasone to methylprednisolone 30 mg/kg. Earlier meta-analyses pooled heterogeneous surgical populations, used odds ratios and compared dose only as a low or high category. The present review is restricted to elective gastrointestinal resection, models dose continuously and includes the trials that randomized dose directly. 3. Eligibility criteria Population. Adults undergoing elective resection of the oesophagus, stomach, small bowel, colon or rectum, liver or pancreas, including biliary–enteric reconstruction; open or minimally invasive; benign or malignant disease. Living-donor hepatectomy counts as elective major liver resection. Excluded: • emergency surgery; • transplant recipients; • non-gastrointestinal procedures; • minor abdominal procedures (cholecystectomy, appendectomy, hernia repair, anorectal and bariatric surgery, endoscopic procedures). Intervention. One systemic dose (intravenous, intramuscular or oral) of dexamethasone, methylprednisolone, hydrocortisone, prednisolone or prednisone, or betamethasone, given before skin incision (including at induction of anaesthesia) or during the operation before skin closure. Excluded: • non-systemic routes; • treatment of a disease rather than prophylaxis; • stress-dose or replacement regimens; • non-glucocorticoid steroids; • administration started only after the operation; • multi-dose regimens. Comparator. Placebo, saline or no glucocorticoid, with otherwise identical perioperative care (primary analysis set), or a different dose of glucocorticoid (dose-comparison set). Excluded: active non-glucocorticoid comparators alone, and co-interventions that differ between arms. Design. Randomized controlled trials with parallel groups, any blinding, any language and any publication date. Excluded: • quasi-randomized and non-randomized studies; • reviews, editorials, case reports and retracted articles; • conference abstracts, which are kept for sensitivity analyses. Outcomes. Outcomes are not a screening criterion. At full text, trials must report at least one clinical outcome. Analysis sets. • Primary set: glucocorticoid against placebo or no glucocorticoid. • Dose-comparison set: a higher against a lower glucocorticoid dose. A trial in which controls also received antiemetic dexamethasone enters this set. • Sensitivity analyses only: • trials in which fewer than 80% of participants underwent an eligible resection; • trials in which only a single-dose subgroup was eligible; • trials that fail the trustworthiness assessment. • Living-donor hepatectomy: eligible, with a sensitivity analysis that excludes it. 4. Information sources and search PubMed, Embase (Embase.com) and Web of Science Core Collection are searched from inception, without language or date restrictions. The full strategies are given in the appendix. Reviews and meta-analyses on the topic identified by the search are flagged, and their included trials are compared with the review's set. For trial protocols identified by the search, the registry record is checked and the results paper is sought. 5. Study selection 4. Deduplication: records are deduplicated with the Automated Systematic Search Deduplicator and checked by hand. 5. Pre-screening filters: the remaining records are managed in Rayyan. Keyword filters on publication type and title exclude, before screening, records that: • are not original randomized research; • do not involve a systemic glucocorticoid; • concern a medical indication or an ineligible operation; • study animals or children. 6. Title and abstract screening: two reviewers independently screen titles and abstracts in Rayyan against the criteria of section 3. 7. Full text: they then assess full texts. 8. Disagreements: resolved by discussion or by a third reviewer. 9. Linked reports: reports of the same trial (substudies, follow-up, protocols, conference abstracts, letters) are linked under one trial. 10. Retrieval: reports that cannot be retrieved are listed with their likely relevance. 6. Data collection Data are collected with a standardized form, piloted on three trials. The form covers: • Trial and publication: registration and timing, registered and reported primary outcomes, design, sample size calculation, funding, ethics and retraction status. • Population and baseline, per arm. • Surgery and perioperative care, including enhanced recovery pathways and antiemetic glucocorticoid use. • Intervention and comparator: drug, dose, weight-based dose, timing and route. • Comparisons; binary, continuous and time-to-event outcomes, per arm and time window. • Risk-of-bias inputs and trustworthiness checks. • Linked reports and an author contact log. Rules: • Location: every value is recorded with its page, table or figure. • Copying and conversion: numbers are copied as printed, and conversions are kept in derived fields. • Conflicting numbers: tables take precedence over text and text over the abstract, and conflicts are flagged. • Patients, not events: patients with an outcome are counted, not events. Event-based counts enter sensitivity analyses only. • Estimate selection: when a trial reports several estimates, 30-day outcomes are preferred over 90-day and in-hospital outcomes, and randomized denominators over modified intention-to-treat and per-protocol ones. • Definitions: non-standard or derived definitions enter sensitivity analyses only. • Registries: registry records are checked for registration date and primary outcome. • Retractions: retraction and correction notices are searched in PubMed and Crossref. • Author contact: corresponding authors may be contacted for missing data. 7. Outcomes Co-primary outcomes: • any postoperative complication (patients with at least one complication, Clavien–Dindo grade I to V, or overall morbidity counted per patient); • major complications (Clavien–Dindo grade III or higher). Secondary binary outcomes: • infectious complications; • surgical site infection (any, superficial and organ/space); • anastomotic leak; grade B/C pancreatic fistula; post-hepatectomy liver failure; bile leak; • pulmonary complications; pneumonia; cardiac complications; • ileus; wound dehiscence; gastrointestinal bleeding; • reoperation; readmission; mortality; • postoperative nausea and vomiting; vomiting; hyperglycaemia; delirium. Continuous outcomes: • length of hospital stay; • C-reactive protein and interleukin-6 on postoperative day 1; • peak blood glucose; • bilirubin, alanine aminotransferase and international normalized ratio; • pain and time to flatus. Outcomes are pooled when at least two trials contribute and are otherwise reported descriptively. 8. Dose Doses are converted to dexamethasone equivalents by anti-inflammatory potency: dexamethasone 1, betamethasone 1.25, methylprednisolone 0.1875, prednisolone 0.15 and hydrocortisone 0.0375. Weight-based doses are converted with the arm mean body weight, or with 70 kg when weight is not reported. Doses are grouped as: • low: 10 mg or less; • intermediate: more than 10 and less than 50 mg; • high: 50 mg or more. 9. Risk of bias and trustworthiness Risk of bias. Two reviewers independently assess risk of bias with RoB 2 for the effect of assignment. The assessment is made separately for each outcome domain: complications, infections, mortality, length of stay, nausea and vomiting, and laboratory end points. Domain and overall judgements follow the algorithms of the official RoB 2 guidance (version of 22 August 2019). Trustworthiness. Every trial is assessed on: • prospective registration; • ethics approval; • plausibility of baseline characteristics and of reported results (spreads, arithmetic); • overlap with other publications. Trials that fail the assessment enter sensitivity analyses only. 10. Synthesis Effect measures and models • Binary outcomes: pooled as risk ratios. • Primary model: random-effects inverse-variance meta-analysis, with the restricted maximum likelihood estimator of between-trial variance and Hartung–Knapp–Sidik–Jonkman confidence intervals. An ad hoc correction prevents an interval narrower than the unadjusted one. • Prediction intervals are reported when at least three trials contribute. • Zero cells take a 0.5 correction in the treatment arm. Trials without events in either arm are excluded from relative measures but kept in mixed m
Source: Open Science Framework
Author: Juan Moreno
Language: English
Subfield: Surgery
Field: Medicine
Domain: Health Sciences
Open Access status: green
OpenAlex ID: W7223567100