Regional Anaesthesia to Prevent Chronic Pain After Surgery
Regional Anaesthesia to Prevent Chronic Pain After Surgery
We searched the Cochrane Central Register of Controlled Trials, PubMed, EMBASE, and CINAHL from their inception to May 2012 without any language restriction. We used a combination of free-text search and controlled vocabulary search. We limited results to RCTs using a highly sensitive search strategy. We conducted a hand search in reference lists of included trials, review articles, and conference abstracts. We detailed our methods a priori in a published protocol and published the search strategies elsewhere in detail.
Types of Studies. We included RCTs. The effects of nerve blocks are obvious to patients and providers; therefore, we accepted single blinding. Blinding of the outcome observer was a prerequisite for inclusion.
Types of participants: we included studies in adults and children undergoing elective surgical procedures. We excluded trauma surgery.
Types of interventions: we included studies comparing local anaesthetics or regional anaesthesia vs conventional pain control. We included all routes of administration of local anaesthetics. We included studies providing regional anaesthesia during any time window in the perioperative period. We excluded studies comparing one regional technique vs another or investigating the effect of timing.
Types of comparators: we included studies which used conventional postoperative pain control such as opioids with or without concomitant nonsteroidal anti-inflammatory drugs or adjuvants.
Types of outcomes: we studied dichotomous pain outcomes as reported in the studies (pain vs no pain; pain or use of pain medication, or both, vs no pain). We included studies assessing differences in scores based on validated pain scales.
We chose the odds ratio (OR) as the summary statistic for our dichotomous primary outcome. We reported the ORs with 95% confidence intervals (CIs). We calculated the number needed (NNT) to treat for the subgroups of thoracotomy and breast cancer surgery.
The two authors independently assessed the methodological quality and extracted data in duplicate including on adverse effects using a standard data collection form, revised after a pilot run. The two authors checked and entered the data into the Cochrane Review Manager (RevMan 5.1, The Nordic Cochrane Centre, København, Denmark) computer software. We published a detailed table of characteristics of included studies listing trials design, participants, interventions, and reported outcomes elsewhere. We contacted the study authors about the missing data and reported data inconsistencies in the aforementioned table of included studies.
The two authors independently graded the quality of studies based on a checklist of design components. The main categories consisted of randomization, allocation concealment, observer blinding, and dealing with the missing data and are classified as high, low, or unclear risk of bias. We achieved consensus by informal discussion.
We grouped studies according to surgical interventions (thoracotomy, limb amputation, breast cancer surgery, laparotomy, and other) instead of pooling across different surgical interventions: each surgical intervention has a different natural history of chronic pain. We investigated study heterogeneity at the subgroup level using a χ test and the calculation of the I statistic. We considered an examination of publication bias using graphical and statistical tests (funnel plot, Egger's test).
We combined all groups using regional or local anaesthesia together and compared them against the groups using conventional pain control, if in a study several groups used variable timing of their regional anaesthesia interventions in different arms. For example, if the first study group received a regional anaesthesia intervention before incision and the second study group received it after incision, we pooled the (first and second) groups using local anaesthetics against the (third) control groups not using any local anaesthetics at all (i.e. using only conventional pain control instead).
If the follow-up varied only by weeks to 1 month, we pooled the results, for example data at 24 weeks or at 5 months with data at 6 months.
We used the inverse-variance approach, adjusting study weights based on the extent of variation among the varying intervention effects. A random-effects model will result in wider CIs for the average intervention effect as it accounts for any potential between-study variance. The result is a more conservative effect estimate. We provide the summary of findings in tables, after the process of GRADE assessment, elsewhere.
Methods
Search, Selection, and Inclusion Criteria
We searched the Cochrane Central Register of Controlled Trials, PubMed, EMBASE, and CINAHL from their inception to May 2012 without any language restriction. We used a combination of free-text search and controlled vocabulary search. We limited results to RCTs using a highly sensitive search strategy. We conducted a hand search in reference lists of included trials, review articles, and conference abstracts. We detailed our methods a priori in a published protocol and published the search strategies elsewhere in detail.
Types of Studies. We included RCTs. The effects of nerve blocks are obvious to patients and providers; therefore, we accepted single blinding. Blinding of the outcome observer was a prerequisite for inclusion.
Types of participants: we included studies in adults and children undergoing elective surgical procedures. We excluded trauma surgery.
Types of interventions: we included studies comparing local anaesthetics or regional anaesthesia vs conventional pain control. We included all routes of administration of local anaesthetics. We included studies providing regional anaesthesia during any time window in the perioperative period. We excluded studies comparing one regional technique vs another or investigating the effect of timing.
Types of comparators: we included studies which used conventional postoperative pain control such as opioids with or without concomitant nonsteroidal anti-inflammatory drugs or adjuvants.
Types of outcomes: we studied dichotomous pain outcomes as reported in the studies (pain vs no pain; pain or use of pain medication, or both, vs no pain). We included studies assessing differences in scores based on validated pain scales.
Summary Statistic
We chose the odds ratio (OR) as the summary statistic for our dichotomous primary outcome. We reported the ORs with 95% confidence intervals (CIs). We calculated the number needed (NNT) to treat for the subgroups of thoracotomy and breast cancer surgery.
Data Extraction
The two authors independently assessed the methodological quality and extracted data in duplicate including on adverse effects using a standard data collection form, revised after a pilot run. The two authors checked and entered the data into the Cochrane Review Manager (RevMan 5.1, The Nordic Cochrane Centre, København, Denmark) computer software. We published a detailed table of characteristics of included studies listing trials design, participants, interventions, and reported outcomes elsewhere. We contacted the study authors about the missing data and reported data inconsistencies in the aforementioned table of included studies.
Assessment of Risk of Bias
The two authors independently graded the quality of studies based on a checklist of design components. The main categories consisted of randomization, allocation concealment, observer blinding, and dealing with the missing data and are classified as high, low, or unclear risk of bias. We achieved consensus by informal discussion.
Stratification and Assessment of Heterogeneity and Reporting Bias
We grouped studies according to surgical interventions (thoracotomy, limb amputation, breast cancer surgery, laparotomy, and other) instead of pooling across different surgical interventions: each surgical intervention has a different natural history of chronic pain. We investigated study heterogeneity at the subgroup level using a χ test and the calculation of the I statistic. We considered an examination of publication bias using graphical and statistical tests (funnel plot, Egger's test).
We combined all groups using regional or local anaesthesia together and compared them against the groups using conventional pain control, if in a study several groups used variable timing of their regional anaesthesia interventions in different arms. For example, if the first study group received a regional anaesthesia intervention before incision and the second study group received it after incision, we pooled the (first and second) groups using local anaesthetics against the (third) control groups not using any local anaesthetics at all (i.e. using only conventional pain control instead).
If the follow-up varied only by weeks to 1 month, we pooled the results, for example data at 24 weeks or at 5 months with data at 6 months.
Data Synthesis
We used the inverse-variance approach, adjusting study weights based on the extent of variation among the varying intervention effects. A random-effects model will result in wider CIs for the average intervention effect as it accounts for any potential between-study variance. The result is a more conservative effect estimate. We provide the summary of findings in tables, after the process of GRADE assessment, elsewhere.