Article type
Year
Abstract
Background: Blinding, random sequence generation, and allocation concealment are established strategies to minimise bias in randomised clinical trials (RCTs). Meta-epidemiological studies of drug trials have demonstrated exaggerated treatment effects in RCTs where such methods were not employed. As blinding is more difficult in procedural trials, it is important to determine whether lack of blinding is a source of bias in surgical RCTs.
Objectives: To determine whether lack of blinding, and other potential sources of bias in RCT design, have a systematic effect on surgical RCT outcomes, using RCTs comparing laparoscopic and open surgery as an example.
Methods: The CENTRAL database was searched for systematic reviews of RCTs that compared laparoscopic and open abdominal surgical procedures. Each review was then scrutinised to determine whether at least one of the included trials was blinded. Eligible reviews were updated and individual RCTs retrieved. Extracted data included primary outcomes – length of stay (LOS) and complications – and a risk of bias assessment. A pairwise meta-analysis was performed for each procedure comparing laparoscopic and open surgery. Serial meta-regression was then used to determine how each bias-minimisation measure influences the size of the treatment effect within each procedure. The resulting coefficients were meta-analysed to obtain an overall difference between trials that employed bias-minimisation strategies, and those that did not.
Results: 440 RCTs were included, comparing laparoscopic and open surgery for nine different surgical procedures. Patient-blinded RCTs reported a smaller difference between laparoscopic and open surgery for LOS (difference in standardised mean differences (DSMD) -0.43 (95% confidence interval (CI) -0.81, -0.05)), and complications (ratio of odds ratios (ROR) 0.78 (CI 0.59, 1.02)). Blinding of carers (apart from the operating surgeon) also had a similar effect on reported LOS (DSMD -0.45 (CI -0.85, -0.05)) and complications (ROR 0.72 (CI 0.56, 0.94)). Other bias-minimisation measures did not show a significant difference.
Conclusions: Lack of blinding alters the effect estimates of RCTs comparing laparoscopic and open surgery, and may lead to erroneous conclusions. Blinding should be implemented in procedural RCTs wherever po
Patient or healthcare consumer involvement: No direct involvement
Objectives: To determine whether lack of blinding, and other potential sources of bias in RCT design, have a systematic effect on surgical RCT outcomes, using RCTs comparing laparoscopic and open surgery as an example.
Methods: The CENTRAL database was searched for systematic reviews of RCTs that compared laparoscopic and open abdominal surgical procedures. Each review was then scrutinised to determine whether at least one of the included trials was blinded. Eligible reviews were updated and individual RCTs retrieved. Extracted data included primary outcomes – length of stay (LOS) and complications – and a risk of bias assessment. A pairwise meta-analysis was performed for each procedure comparing laparoscopic and open surgery. Serial meta-regression was then used to determine how each bias-minimisation measure influences the size of the treatment effect within each procedure. The resulting coefficients were meta-analysed to obtain an overall difference between trials that employed bias-minimisation strategies, and those that did not.
Results: 440 RCTs were included, comparing laparoscopic and open surgery for nine different surgical procedures. Patient-blinded RCTs reported a smaller difference between laparoscopic and open surgery for LOS (difference in standardised mean differences (DSMD) -0.43 (95% confidence interval (CI) -0.81, -0.05)), and complications (ratio of odds ratios (ROR) 0.78 (CI 0.59, 1.02)). Blinding of carers (apart from the operating surgeon) also had a similar effect on reported LOS (DSMD -0.45 (CI -0.85, -0.05)) and complications (ROR 0.72 (CI 0.56, 0.94)). Other bias-minimisation measures did not show a significant difference.
Conclusions: Lack of blinding alters the effect estimates of RCTs comparing laparoscopic and open surgery, and may lead to erroneous conclusions. Blinding should be implemented in procedural RCTs wherever po
Patient or healthcare consumer involvement: No direct involvement