Article type
Year
Abstract
Background: Surgeons' expertise is a known confounder of outcomes in randomized controlled trials (RCTs).
Objectives:The aim of this work was to review the reporting and impact of surgeons’ expertise in RCTs.
Methods: A systematic review of surgeons’ expertise as reported in RCTs comparing total hip arthroplasty (THA) against hemiarthroplasty (HA), and comparing the different total knee arthroplasty (TKA) techniques of standard parapatellar (MP), midvastus (MV), subvastus (SV) and quadriceps sparing (QS). We searched MEDLINE, EMBASE, Science Citation Index, The Cochrane Library, and Conference Proceedings Citation Index; reference checking was also performed. Study selection, data extraction, critical appraisal (Cochrane risk of bias tool) and meta-analysis were undertaken by two reviewers.
Results: 75 relevant RCTs were identified: 65 RCTs comparing approaches to TKA and 10 for THA. The risk of bias due to surgeons' reported expertise was categorised as low, moderate or high. The majority of RCTs in this sample were assessed as being at high risk of bias on this variable: 42/65 TKA trials (56%) and 5/10 THA trials (50%). Approximately one quarter were deemed to be at low risk of bias: 15/65 TKA trials (23%) and 3/10 THA trials (30%). The reporting of this variable in both comparisons improved from 2009 onwards, e.g. the proportion of low risk of bias RCTs increased from 7% to 35% for the MV vs MP comparison, and from 0% to 53% for SV vs MP; the proportion of trials at high risk of bias also decreased exponentially, from 85% to 53% for MV vs MP, and from 83% to 26% for SV vs MP. Meta-analyses did not demonstrate any meaningful differences between the findings of trials at high risk of bias compared with trials at low/moderate risk of bias, except that the latter did tend to produce findings of relatively greater uncertainty across four of the five outcomes in this sample. This accords with other evidence that better-conducted trials can tend to produce less ‘positive’ findings.
Conclusions: Surgeons’ expertise is often poorly reported but has improved since 2009, possibly due to the dissemination of the CONSORT-NPT extension statement. However, it is not adequately captured by any other risk of bias domain and is therefore worthy of assessment.
Patient or healthcare consumer involvement: None; secondary research investigated procedural bias.
Objectives:The aim of this work was to review the reporting and impact of surgeons’ expertise in RCTs.
Methods: A systematic review of surgeons’ expertise as reported in RCTs comparing total hip arthroplasty (THA) against hemiarthroplasty (HA), and comparing the different total knee arthroplasty (TKA) techniques of standard parapatellar (MP), midvastus (MV), subvastus (SV) and quadriceps sparing (QS). We searched MEDLINE, EMBASE, Science Citation Index, The Cochrane Library, and Conference Proceedings Citation Index; reference checking was also performed. Study selection, data extraction, critical appraisal (Cochrane risk of bias tool) and meta-analysis were undertaken by two reviewers.
Results: 75 relevant RCTs were identified: 65 RCTs comparing approaches to TKA and 10 for THA. The risk of bias due to surgeons' reported expertise was categorised as low, moderate or high. The majority of RCTs in this sample were assessed as being at high risk of bias on this variable: 42/65 TKA trials (56%) and 5/10 THA trials (50%). Approximately one quarter were deemed to be at low risk of bias: 15/65 TKA trials (23%) and 3/10 THA trials (30%). The reporting of this variable in both comparisons improved from 2009 onwards, e.g. the proportion of low risk of bias RCTs increased from 7% to 35% for the MV vs MP comparison, and from 0% to 53% for SV vs MP; the proportion of trials at high risk of bias also decreased exponentially, from 85% to 53% for MV vs MP, and from 83% to 26% for SV vs MP. Meta-analyses did not demonstrate any meaningful differences between the findings of trials at high risk of bias compared with trials at low/moderate risk of bias, except that the latter did tend to produce findings of relatively greater uncertainty across four of the five outcomes in this sample. This accords with other evidence that better-conducted trials can tend to produce less ‘positive’ findings.
Conclusions: Surgeons’ expertise is often poorly reported but has improved since 2009, possibly due to the dissemination of the CONSORT-NPT extension statement. However, it is not adequately captured by any other risk of bias domain and is therefore worthy of assessment.
Patient or healthcare consumer involvement: None; secondary research investigated procedural bias.