2.2. Selection of studies and data extraction
Two investigators (WJX and HFC) reviewed all titles, abstracts, and then full texts sequentially. Finally, eligible trials were determined and eligibility, quality, and outcomes data were retrieved independently. Disagreements on eligibility between two reviewers were resolved via mutual discussion, when needed, a third reviewer (YCL) was requested for final decision. Relevant data were extracted from eligible literature with a standard extraction formula with subsequent cross-checking.
Data retrieved included: (1) first author, year of publication, study location, study design, sample size, gender, age, ASA status, types of surgery, premedication, anesthesia maintenance, intervention description, control description, dose of opioids, postoperative analgesic strategines, and (2) pain intensity in the form of the various pain scores during the 0 to 24 postoperative hours, pain threshold or normalized area of hyperalgesia during the 0 to 48 postoperative hours, cumulative morphine consumption at 24h after surgery, time to first rescue analgesic, and incidence of postoperative opioid-related side-effects, such as postoperative nausea and vomiting (PONV), shivering, dizziness and hypotension. Dichotomous data were extracted as the number of patients (%). Continuous data were extracted in the form of mean ± standard deviations (SDs).
When the target data in the article were incomplete, we attempted to contact the author via e-mail twice, but no responses were received. When the standard deviation was missing, range and median estimation[15] were used for the conversion.