Summary of findings
Carpal tunnel syndrome (CTS) is the most common compressive neuropathy of the upper extremity, with a prevalence of clinically and electrophysiologically confirmed diagnosis being 2.7% of the general population (Atroshi 1999). The incidence of newly diagnosed cases of CTS in the UK is 90 men and 193 women per 100,000 visits to primary care departments per year (Latinovic 2006). The equivalent figures in the Netherlands are 90 and 280 per 100,000 visits per year (Bongers 2007). Approximately 500,000 operations for CTS are performed every year in the US, at a cost of over USD 2 billion annually (Palmer 1995). According to US Department of Labour figures (2009), a sick leave of at least 30 days per year is recorded in approximately 45% of people with CTS, with a median of 28 days away from work, which suggests important insurance-related consequences (U.S. Department of Labor 2009).
Description of the condition
CTS is caused by median nerve neuropathy, where the nerve passes along the carpal tunnel at the wrist. Increased pressure on the median nerve between the transverse carpal ligament and the carpal bones dorsally is usually the trigger that compromises the nerve's blood supply and leads to oedema, causing functional impairment and clinically evident symptoms (Fuchs 1991). CTS can be secondary when there is an obvious pathology that puts pressure on the median nerve or that indirectly contributes to the median neuropathy (Stevens 1992). The vast majority of cases though are usually considered idiopathic and most commonly affect women between 40 and 70 years of age (Atroshi 1999; Phalen 1966).
The first symptoms that people with CTS notice, and which often lead them to medical services, are paraesthesia and numbness in the distribution area of the median nerve often accompanied by pain. The symptoms are typically more apparent during the night and usually disturb sleep. Atrophy of the thenar muscles due to insufficient innervation by the median nerve appears gradually in the longer term and the person eventually notices weakness.
Electrophysiological tests (nerve conduction studies) have been used to support the clinical diagnosis of CTS, and to distinguish CTS from other lesions of the peripheral or central nervous system. The tests usually reveal a decreased conduction velocity and increased latency in the part of the median nerve located along the carpal tunnel (Jordan 2002).
In the early stages of CTS, conservative treatment is recommended to improve quality of life. This consists of rest, splinting or anti-inflammatory medication either orally or in the form of perineural corticosteroid injections (O'Connor 2003; Piazzini 2007). About 20% of people with CTS might improve without treatment of any kind (Padua 2001), but if conservative or surgical treatments fail, chronic pressure on the median nerve can lead to irreversible nerve damage and permanent muscle weakness (Gelberman 1988), even if the person undergoes surgery at a later date. Surgical intervention is eventually recommended in 30% to 40% of people with CTS (Latinovic 2006; Wilson 2003).
Description of the intervention
Surgical treatment of CTS consists of cutting the transverse ligament of the palm, thus releasing the pressure on the underlying median nerve (Ablove 1994; Richman 1989). Cannon and Love first described carpal tunnel release in 1946. The surgery was performed under direct vision, with a skin incision along the axis of the palm, followed by dissection of the subcutaneous tissue and cutting of the underlying transverse ligament. Following the first description of the surgical technique, many modifications were published, mainly regarding the shape and the extent of the surgical incision. More recent literature usually suggests less extended surgical trauma with an incision no more than 2 cm to 3 cm in length (Higgins 2002; MacKinnon 2005). Additional interventions have also been suggested in the past in order to increase the efficacy of CTS surgical treatment. Epineurotomy or even internal neurolysis of the median nerve have been performed, but are not common and are not performed except for specific indications (Curtis 1973; Fissette 1979). Reconstruction of the transverse ligament has also been proposed but authors have not managed to demonstrate its superiority, as several studies have shown an increased recurrence rate with this procedure (Karlsson 1997).
All techniques described above have the common step of dividing the skin and underlying tissue in addition to the transverse ligament. Endoscopic carpal tunnel release (ECTR) is a relatively new procedure, first being described in 1989 by Chow and Okutsu (Chow 1989; Okutsu 1989). It requires the use of special instrumentation, including an endoscopic camera, optic fibre light source and a monitor. The procedure is performed with one or two small incisions (portals) proximal or distal to the carpal tunnel. Instrumentation is advanced through those portals, underneath the transverse ligament. With the aid of a camera, the surgeon obtains indirect access to the bottom surface of the transverse ligament. The ligament is cut from its lower surface with a knife, thus preserving the subcutaneous tissue and the overlying skin. Several variations of the endoscopic method have been subsequently developed, although the two more commonly used techniques are the one-portal technique described by Agee (Agee 1992; Agee 1994), and the two-portal technique described by Chow (Chow 1989; Chow 1993).
How the intervention might work
The proposed advantage of ECTR over open techniques is that by accessing and dividing the transverse carpal ligament from within the carpal tunnel, the surgeon leaves overlying structures intact. This is thought to decrease postoperative morbidity by reducing pain, providing faster trauma healing, shortening patients' rehabilitation time and allowing an earlier return to work. The skin and subcutaneous tissue palmar to the transverse ligament have also been considered to have a pulley effect over the digital flexor tendons. Thus, preservation of these overlying tissues might enhance the increase in grip strength of the hand postoperatively (Macdermid 2003; Vasiliadis 2010).
ECTR should also be studied from a financial point of view. ECTR has been attacked on the grounds of the increased cost of instrumentation and surgeons' training expenses (Lorgelly 2005). On the other hand, an earlier return to work and a shorter period of sick leave must also be included in any evaluation of the total economic impact of the operation (Saw 2003).
Finally, there is a controversy regarding the safety of ECTR compared to conventional open carpal tunnel release (OCTR). Given that it takes longer for a surgeon to master the ECTR technique, some authors suggest that it is a dangerous surgical option for patients.
Why it is important to do this review
Since it was first described in 1989, endoscopic treatment of CTS has become increasingly popular. Among the surgical options, it is considered to be less invasive and to lead to faster postoperative rehabilitation due to decreased surgical trauma.
Companies launch new or improved instrument for ECTR regularly, and subsequent marketing also contributes to wider use of the technique. However, endoscopic surgery is costly and requires specialised training and equipment.
There is therefore interest in, and a need for, an evaluation of the current endoscopic technique. The main questions that need to be answered relate to its efficacy and safety compared to OCTR, which remains the gold standard method for carpal tunnel release. Despite the first studies’ scepticism regarding the safety of ECTR, after a period of modifications to the method and growing experience, endoscopic and open methods appear to have comparable complication rates according to more recent studies and reviews (Boeckstyns 1999). With endoscopic surgery, the limited surgical trauma is believed to offer better rehabilitation and a faster recovery, removing all the complications of incision (Vasiliadis 2006).
The first review comparing surgical treatments of CTS was published in 2001 (Gerritsen 2001). Scholten et al. have since published updates of that review in The Cochrane Library in 2002, 2004 and 2007 (Scholten 2007).
Due to the increasing number of studies since Scholten 2007, the review of surgical treatment for CTS has been split into smaller reviews, of which this is the first. This review focuses on ECTR techniques. The Scholten 2007 review is the reference for other surgical interventions for CTS until it is superseded by new, focused reviews.
To assess the effectiveness and safety of the endoscopic techniques of carpal tunnel release compared to any other surgical intervention for the treatment of carpal tunnel syndrome. More specifically, to evaluate the relative impact of the endoscopic techniques in relieving symptoms, producing functional recovery (return to work and return to daily activities) and reducing complication rates.
Criteria for considering studies for this review
Types of studies
We considered any randomised controlled trial (RCT) and quasi-RCT comparing endoscopic carpal tunnel release (ECTR) with any other surgical intervention for the treatment of carpal tunnel syndrome (CTS). We did not apply any language restriction.
Measurement of particular outcomes was not used as an eligibility criterion for study inclusion.
Types of participants
We included studies with participants with clinical diagnosis of CTS with or without electrophysiological confirmation. We accepted the authors’ definition of CTS and their views of what constituted electrophysiological confirmation.
Types of interventions
We considered studies comparing ECTR with any other surgical intervention. This included open carpal tunnel release (OCTR) and its variations, OCTR with mini-open technique and OCTR with concomitant interventions (such as lengthening of flexor retinaculum, internal neurolysis, epineurotomy or tenosynovectomy). We also included studies comparing different techniques of ECTR with each other.
Types of outcome measures
The primary outcome assessed was overall improvement of symptoms, considering any measure in which participants indicated the intensity of their complaints compared to the pre-operative status. We considered questionnaires measuring the overall improvement of symptoms with ratings of the kind 'improved' or 'not improved' or any patient-reported questionnaire assessing overall satisfaction.
We evaluated the following secondary outcome measures.
- Improvement of CTS symptoms, as measured by the Symptom Severity Score (SSS) (Levine 1993) or any other measure for improvement in pain, paraesthesiae, or nocturnal paraesthesia. If data for symptoms were presented separately for pain or paraesthesia they were used as long as they were measured using a validated instrument.
- Disability measured with the Disabilities of the Arm, Shoulder and Hand (DASH) questionnaire (Hudak 1996).
- Function measured with the Functional Status Scale (FSS) questionnaire (Levine 1993).
- Grip strength.
- Time to return to work or to resume activities of daily living.
We took both short-term (less than or equal to three months) and long-term (greater than three months) measures of overall improvement and improvement in CTS symptoms into consideration. In cases where multiple time points were reported, as the short-term measure we used the closest measure to three months. For long-term effects, we used the latest follow-up measurement (if at more than three months).
We also assessed the risk of complications as reported by the authors, which were measured as the proportion of patients with:
- major complications (for example, nerve, vascular or tendon injuries); and
- minor complications (for example, pain, scar disorders).
Search methods for identification of studies
On 18 November 2013, we searched the Cochrane Neuromuscular Disease Group Specialized Register, CENTRAL (2013, Issue 11 in The Cochrane Library), MEDLINE (January 1966 to November 2013) and EMBASE (January 1980 to November 2013). There were no language restrictions in the search strategy. We reviewed the reference lists of relevant articles and contacted trial authors.
This review fully incorporates the results of searches conducted up to November 2012. We updated the search in November 2013, to identify any additional studies to address in the next update.
Searching other resources
We searched reference lists of all primary studies and review articles for additional references. We also searched trial registers for ongoing trials: US National Institutes of Health ClinicalTrials.gov (www.clinicaltrials.gov) (June 2013), Current Controlled Trials (www.controlled-trials.com) (ISRCTN Register, Action Medical Research (UK), The Wellcome Trust (UK), UK trials (UK)) (June 2013), UK Clinical Trials Gateway (www.ukctg.nihr.ac.uk/default.aspx) (June 2013) and the World Health Organization Clinical Trials Registry Platform (www.who.int/ictrp/en/) (June 2013) (see Appendix 5).
Data collection and analysis
Selection of studies
Two review authors (HSV, IS) independently scanned records retrieved by the initial search. We included only RCTs and quasi-RCTs. We excluded obviously irrelevant studies and we retrieved for further evaluation the full text of studies chosen by at least one of the two authors. The authors resolved disagreements by discussion.
To be included, a study had to meet the following criteria:
- the study population consisted of people with CTS;
- ECTR was compared with an open surgical technique; and
- the study was designed as an RCT.
Data extraction and management
Two review authors (HSV, PG) extracted data independently using pre-standardised forms. Data extraction forms included information on methods, participants, interventions and outcomes. We compared extracted data and resolved differences by discussion. One author (HSV) entered the data into the Cochrane software Review Manager 5 (RevMan) (RevMan 2012), and another author (PG) checked the data entry on completion.
Assessment of risk of bias in included studies
Two review authors (HSV, PG) independently assessed the risk of bias for each trial using the Cochrane Collaboration's tool described in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011b, updated Higgins 2011a).
We assessed the adequacy of sequence generation, allocation concealment and blinding (of participants, personnel and outcome assessors) and we made judgements about the possible impact of incomplete outcome data, selective outcome reporting and other sources of bias. We evaluated each item as at low, high or unclear risk of bias.
The criteria for judging the risk of bias in each study are given in details in table 8.5.c of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011b). We have presented the bias items that we adapted to the context of our review below in more detail.
It is not possible to blind either surgeons or the participants to the performed operation. Surgical incisions are always obvious. Thus, we scored all studies as at high risk of bias for the item 'blinding of participants and personnel', unless otherwise reported by the authors. However, the outcome assessor could be blinded (for example, for assessing grip strength).
We gave the following judgements: 'low' when assessors were blinded to the performed operation technique, 'high' when they were not blinded and 'unclear' if the authors gave no information regarding the blinding of outcome assessment.
Addressing incomplete outcomes
We collected the number of dropouts and reasons for attrition or exclusion. We evaluated whether intention-to-treat (ITT) analysis had been performed and recorded differences in attrition between intervention groups.
The judgement was 'low risk of bias' when there were no missing values in the outcome data, when the numbers of and reasons for missing values were not likely to affect the outcome, or when imputations to achieve ITT analysis were appropriate. When the extent of missing outcome data and the reasons for missing data were likely to have affected the outcome, then the judgement was 'high risk of bias'. Our assessment was 'unclear' when trial authors did not provide enough information about the amount of attrition and the reasons for it.
We evaluated the possibility of selective reporting. We based our judgements primarily on comparing the study protocols (if these could be identified) with the published report. We searched in www.clinicaltrials.gov and www.controlled-trials.com (ISRCTN Register, Action Medical Research (UK), the Wellcome Trust (UK), and UK trials (UK)) to identify protocols of the included studies. In the absence of the protocols we evaluated whether reports presented all expected outcomes and whether there was agreement between the methods section and the results.
The judgement was 'low risk of bias' when it was clear from the protocol, the published report, or both that all outcomes were fully reported. We classified trials as at 'high risk of bias' when it was clear that the articles did not present results for some measured outcomes. We classified papers as 'unclear' when it was not clear whether the report presented results for all analysed outcomes.
We considered two additional sources of bias.
Trial sponsors (usually manufacturers of the instrumentation needed in ECTR) could have biased the results. Our judgement was 'high risk of bias' if there was a sponsor and 'low risk of bias' when there was a statement that the trial had not received any funding from a party with a vested interest; otherwise the judgement was 'unclear'.
As we anticipated that trials would have small sample sizes, we considered that the presence of baseline differences might have an impact on the results. We classified studies with baseline imbalance in important participant characteristics as at 'high risk of bias'. If there were no such differences or these differences at baseline were not clinically relevant, we classified the study as being at 'low risk of bias'. We reserved 'unclear risk of bias' for studies with insufficient information to form a judgement.
Measures of treatment effect
We described dichotomous data using the risk ratio with 95% confidence interval (CI).
For continuous outcomes measured with the same scale, we used the mean difference and 95% CI. When studies used different scales for the same outcome, we calculated the standardised mean difference. We collected results based on change scores only if final values were not available.
Unit of analysis issues
Bilateral CTS and surgical treatment of both hands are common in such trials. If results are reported for the first hand only, we used these to bypass the problem of dependency.
In the event of bilateral involvement where study authors analysed and presented data for hands rather than for participants, we had planned to extract effect sizes that account for the dependency of observations (such as effects calculated with generalised estimating equations or methods for cluster randomised trials). Many studies randomise participants in both groups: randomisation takes place for the first hand whereas the second hand is operated with the alternative technique. In such cases we extracted outcomes taking into account the paired nature of the data by seeking information on paired statistics and estimate standard errors as described in Section 16.4.6 of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011b, updated Higgins 2011a). When the correlation coefficient was not provided to derive the appropriate adjusted estimate we employed a correlation of 0.5 for the standard analysis and we used two other extreme values of 0.1 and 0.9 in a sensitivity analysis.
In some cases, we could not obtain adjusted estimates and in other cases, only a subset of the participants underwent operations on both hands and it was unclear whether randomisation took place for hands or participants. In these cases, we collected crude estimates based on outcomes pertaining to hands along with the number of randomised participants who contributed information from both hands to evaluate the degree of dependence in the outcomes.
In this latter case, we used sensitivity analysis to evaluate the extent to which the conclusions of the meta-analysis might be altered by failure to account for bilateral involvement in individual trials.
In the case of three-arm studies with more than two eligible study groups, the sample size and event rate of the ECTR group were divided by two, so that the participants randomised to ECTR were not double counted.
Dealing with missing data
With the purpose of including all participants randomised to any intervention, we made every effort to extract data according to the ITT principle; that is, to analyse participants as randomised. When outcome data were not available for some participants, we included the data as reported and we recorded the analysis method (for example, whether results pertain to per protocol or available cases analysis) and noted the lack of ITT as a risk of bias.
Assessment of heterogeneity
We evaluated the presence of clinical heterogeneity by comparing the participants' characteristics and the methodology across studies (see Data synthesis). We assessed statistical heterogeneity by visual inspection of the forest plots along with consideration of the test for heterogeneity and the I
Assessment of reporting biases
For outcomes with at least ten studies, we drew funnel plots to assess the association between study size and effect size. Where appropriate, we used contour enhanced funnel plots to distinguish between reporting bias and other causes of asymmetry.
We synthesised outcome data from studies sufficiently similar in participant characteristics (for example, age, sex, grip strength, distal motor/sensory latency) and methodology followed (length of follow-up, diagnostic criteria) using a random-effects model. We also calculated summary estimates according to fixed-effect models as part of the sensitivity analysis. We decided a priori that if the 95% CI for the random-effects summary estimate included the 95% CI for the fixed-effect summary estimate, we would report only the former as it appropriately conveys heterogeneity.
Subgroup analysis and investigation of heterogeneity
For outcomes with enough studies, we undertook pre-specified subgroup analyses to investigate differences in the effect sizes and heterogeneity across subgroups. The subgroups were: a) the open technique used (standard incision or modified incision including mini-open techniques, with or without concomitant procedures such as neurolysis or transverse ligament reconstruction); and b) the endoscopic technique (one or two portals).
We conducted sensitivity analyses to assess the robustness of the conclusions. We planned to exclude studies according to the following characteristics.
- High or unclear risk of bias for incomplete outcome data.
- Inappropriate adjustment for bilateral involvement.
- High or unclear risk of bias for allocation concealment.
Only complications were reported in a sufficiently large number of studies to allow sensitivity analysis and very few studies were at low risk of bias (nine for incomplete outcome data and two for allocation concealment). Therefore, we performed sensitivity analysis only when enough studies (three or more) per outcome were available.
'Summary of findings' table
We included the outcomes: overall improvement (main outcome), SSS, FSS, grip strength, time to return to work, reoperations, and major complications (for example, nerve, vascular or tendon injuries) in the 'Summary of findings' table.
For continuous outcomes (SSS, FSS, grip strength, time to return to work), we used the range of mean values in the control group (non-endoscopic intervention) as assumed risk.
For binary outcomes (overall improvement, re-operations and major complications) we calculated the assumed risk from the control intervention of the included RCTs by simply merging samples, as we did not expect important variations and we anticipated a small number of studies.
For both types of outcome, we used the summary estimate from the meta-analysis to calculate the corresponding risk for endoscopic surgery, using the open technique as the reference, according to Schünemann 2008.
The protocol of this review was published in the Cochrane Library (Vasiliadis 2010b).
Description of studies
Results of the search
From the searches to November 2012, the number of possibly relevant studies identified from each database were as follows: 58 from the Cochrane Neuromuscular Disease Group Specialized Register, 137 from CENTRAL, 294 from MEDLINE and 174 from EMBASE. We found a total of 663 publications from database searches and one from other sources. After removal of the duplicated abstracts, 545 were left for evaluation.
A total of 72 titles and abstracts regarding various surgical treatment options for carpal tunnel syndrome (CTS) seemed to fulfil the inclusion criteria and required further discussion between the authors. After discussion, we excluded 23; thus we evaluated 49 studies. We included seven studies reported only as abstracts and retrieved 42 full manuscripts for further evaluation. We finally judged seven abstracts and 26 manuscripts to fulfil the inclusion criteria for this systematic review. We have illustrated the study selection process in a flow diagram (Figure 1).
|Figure 1. Study flow diagram (does not include the results of search in November 2013, which will be fully assessed in the next update).|
Five papers were removed because of duplication. One study was published twice (in German and in English), so the results of both sets of papers were combined (Benedetti/Sennwald 1995). Atroshi 2006 and Atroshi 2009 presented short-term and long-term data respectively, from the same study. In one study (Foucher 1993), the results were duplicated in another publication in manuscript form (Braga 1996), and also in an abstract (Foucher 1994). We were unable to find Ugurlu 2009 in a full manuscript and we included it in the Studies awaiting classification.
Thus, we finally included 28 genuine studies in the review. Details of the participants, interventions and outcomes in these studies are presented in Characteristics of included studies.
Since the last update of Scholten 2007, we have identified four new studies (Incoll 2004; Malhotra 2007; Tian 2007; Tüzüner 2008). We also included Giele 2000; Koskella 1996; Sørensen 1997 and Werber 1996, which were awaiting assessment in Scholten 2007, and Schäfer 1996, which was previously excluded as a quasi-randomised trial.
Shortly before publication, in November 2013, we checked an updated search for additional studies and identified three further potentially eligible trials (Aslani 2012; Ejiri 2012; Kang 2013). These have not yet been incorporated into the results and will be addressed in the next update. See Characteristics of studies awaiting classification for details.
Twenty-eight studies were finally included in this review (see Characteristics of included studies). Five of the studies were presented only as an abstract (Giele 2000; Incoll 2004; Koskella 1996; Sørensen 1997; Werber 1996).
In total, 2586 hands were assessed, 1316 treated with endoscopic carpal tunnel release (ECTR) and 1270 with open carpal tunnel release (OCTR). Twenty-five studies compared ECTR with standard OCTR (Agee 1992; Atroshi 2006; Benedetti/Sennwald 1995; Brown 1993; Dumontier 1995; Eichhorn 2003; Erdmann 1994; Ferdinand 2002; Foucher 1993; Giele 2000; Hoefnagels 1997; Incoll 2004; Jacobsen 1996; Koskella 1996; Macdermid 2003; Malhotra 2007; Saw 2003; Schäfer 1996; Sørensen 1997; Stark 1996; Tian 2007; Trumble 2002; Tüzüner 2008; Werber 1996; Westphal 2000) ( Table 1), and five studies compared ECTR with OCTR using a modified incision (Eichhorn 2003; Mackenzie 2000; Rab 2006; Sørensen 1997; Wong 2003) ( Table 2). In Eichhorn 2003 and Sørensen 1997, both conventional open and mini-open techniques were compared with ECTR.
Different types of ECTR were applied. All techniques were aimed at dividing the transverse carpal ligament from within the carpal tunnel but differed in the way in which this was achieved. Eleven studies addressed Agee’s one-portal technique (Agee 1992; Benedetti/Sennwald 1995; Ferdinand 2002; Foucher 1993; Hoefnagels 1997; Mackenzie 2000; Malhotra 2007; Saw 2003; Schäfer 1996; Stark 1996; Trumble 2002), and five studies evaluated other one-portal techniques (Sørensen 1997; Tian 2007; Tüzüner 2008; Werber 1996; Westphal 2000). The other techniques evaluated included the Menon's one-portal technique (Tüzüner 2008), the Concept CTS Relief Kit (Sørensen 1997), the Okutsu technique (Tian 2007), the Endo-Cartris technique (Westphal 2000), with one paper not describing the technique adequately enough to be categorised (Werber 1996). In nine studies Chow’s two-portal technique was used (Atroshi 2006; Brown 1993; Dumontier 1995; Eichhorn 2003; Erdmann 1994; Jacobsen 1996; Macdermid 2003; Rab 2006; Wong 2003). Three studies did not describe the exact ECTR technique used (Giele 2000; Incoll 2004; Koskella 1996).
Nineteen studies solely addressed patients with electrophysiologically-confirmed CTS (Agee 1992; Atroshi 2006; Benedetti/Sennwald 1995; Brown 1993; Eichhorn 2003; Erdmann 1994; Ferdinand 2002; Hoefnagels 1997; Jacobsen 1996; Koskella 1996; Macdermid 2003; Mackenzie 2000; Malhotra 2007; Rab 2006; Sørensen 1997; Tian 2007; Trumble 2002; Tüzüner 2008; Wong 2003); one study addressed both patients with and without electrophysiologically-confirmed CTS (Stark 1996) and two studies addressed patients with clinical CTS where electrophysiological confirmation was not required (Foucher 1993; Saw 2003). In two studies it was not clear how CTS was diagnosed (Dumontier 1995; Westphal 2000).
One study also addressed patients with secondary CTS (Erdmann 1994). In nine studies the type of CTS was not mentioned (Eichhorn 2003; Foucher 1993; Hoefnagels 1997; Incoll 2004; Koskella 1996; Macdermid 2003; Schäfer 1996; Sørensen 1997; Werber 1996).
Only participants with unilateral CTS were included in nine studies (Atroshi 2006; Benedetti/Sennwald 1995; Dumontier 1995; Foucher 1993; Hoefnagels 1997; Macdermid 2003; Schäfer 1996; Werber 1996; Westphal 2000). Sørensen 1997 gave no information about unilateral or bilateral involvement.
In six studies only patients with bilateral CTS were included (Ferdinand 2002; Giele 2000; Incoll 2004; Rab 2006; Stark 1996; Wong 2003). In two of those studies the first hand was randomised to either ECTR or OCTR and, after full recovery of the first hand (Stark 1996), or after at least six months (Rab 2006), the other hand received the alternative treatment. In both studies the timing of the procedures was discarded and in one the analysis pertained to all hands, violating the assumption of independent observation (Stark 1996). In the other three studies, ECTR was randomly allocated to one hand only (Ferdinand 2002; Incoll 2004; Wong 2003). The other hand was treated with the alternative procedure in the same session in Ferdinand 2002 and Wong 2003. No information about the time of second surgery is given in Giele 2000 and Incoll 2004. One of the six studies with a matched design applied an appropriate statistical analysis (Ferdinand 2002). Two further studies provided data for which we were able to obtain relative treatment effects for pain adjusted for matching, assuming a correlation coefficient of 0.5 (Rab 2006; Wong 2003). We subsequently evaluated the impact of this assumption in a sensitivity analysis.
In 10 studies some (but not all) of the participants had bilateral CTS (Agee 1992; Brown 1993; Erdmann 1994; Jacobsen 1996; Koskella 1996; Mackenzie 2000; Saw 2003; Tian 2007; Trumble 2002; Tüzüner 2008). In Malhotra 2007, one participant (out of 60) had a bilateral open surgery. In Agee 1992, randomisation of participants with bilateral CTS was discarded because participants who were randomised to ECTR refused to undergo OCTR as a second procedure. Therefore, the 25 participants with bilateral CTS were omitted from further analysis. For the other nine studies that included some participants with bilateral CTS, the articles provided no further details regarding the analysis.
We excluded 16 trials from this systematic review (see Characteristics of excluded studies). We excluded 10 studies because the participants were not randomised (Dimitriou 1997; Flores 2005; Futami 1995; Hallock 1995; Povlsen 1997; Uchiyama 2002; Uchiyama 2004; Vasiliadis 2010; Worseg 1996; Zhao 2004), and three studies assessed the validity of scores (Atroshi 2007; Katz 1994b), or responsiveness of measures (Katz 1994a). Bal 2008; Cellocco 2005 and Lorgelly 2005 compared open with mini-open techniques.
In Agee 1992, inadequate randomisation applied to the 25 participants with bilateral involvement, but not to the remaining 97 participants with unilateral involvement. Data regarding return to work were presented separately for those 97 participants and these data were included in our review.
Risk of bias in included studies
The results of the 'Risk of bias' assessment are presented in the Characteristics of included studies and summarised in Figure 2. Additionally, we have provided a brief descriptive account of the studies below.
|Figure 2. 'Risk of bias' summary: review authors' judgements about each 'Risk of bias' item for each included study. Green (+) = low risk of bias; yellow (?) = unclear risk of bias; red (-) = high risk of bias|
Appropriate sequence generation to ensure randomisation seemed likely in seven studies (Atroshi 2006; Benedetti/Sennwald 1995; Brown 1993; Ferdinand 2002; Saw 2003; Tüzüner 2008; Wong 2003). Schäfer 1996 was a quasi-randomised trial, as the treatment was allocated according to the day of the week (odd or even). None of the other trials adequately described the method of randomisation.
Allocation concealment was adequate in three studies (Atroshi 2006; Brown 1993; Tüzüner 2008). The method of allocation concealment was judged to be inappropriate, resulting in a high risk of bias, in six trials (Agee 1992; Dumontier 1995; Mackenzie 2000; Rab 2006; Schäfer 1996; Trumble 2002). The method of concealment was not clearly described in 19 studies.
Owing to the type of intervention, the participants and personnel could not be blinded and, therefore, we scored this item 'high risk of bias' for all studies.
In Atroshi 2006, the authors say that "Before each postoperative examination, the patients were instructed not to discuss the type of operation and had their palm and distal forearm covered with a stockinette (an elastic, sleeve-like dressing) concealing the scars. The assessor was thus blinded to the surgical method." However, there was no reference to blinding in the five-year follow-up (of Atroshi 2009). Given that most of the outcomes were patient-assessed questionnaires and that complications and the long-term outcomes were assessed in the latest follow-up of Atroshi 2009, we concluded that there was a high risk of performance and detection bias for this study.
Incomplete outcome data
In nine studies the risk of attrition bias was considered to be low (Atroshi 2006; Benedetti/Sennwald 1995; Hoefnagels 1997; Jacobsen 1996; Rab 2006; Saw 2003; Schäfer 1996; Stark 1996; Tüzüner 2008). None of the participants were lost to follow-up in Jacobsen 1996, Rab 2006, Schäfer 1996, Stark 1996 and Tüzüner 2008. The number of participants lost to follow-up or converted to another treatment was equally distributed between the groups, or in three studies was too small to qualitatively affect the final outcome (Atroshi 2006; Benedetti/Sennwald 1995; Hoefnagels 1997). In Malhotra 2007, six participants out of 36 and four out of 34 were lost to follow-up from the OCTR and ECTR groups respectively, at both one and six months. Although the number was comparable between groups, the incidence was quite large (15%) and the trial authors provided no explanation. Therefore, we judged the risk of attrition bias to be unclear. In Saw 2003, with respect to measures repeated over time, the investigators used a 'last observation carried forward' strategy to impute missing values. The review authors judged this study to be at low risk of attrition bias.
Three studies had a high risk of attrition bias (Dumontier 1995; Eichhorn 2003; Mackenzie 2000). Many participants did not provide outcomes in Dumontier 1995 (27 of 85 at three months and 65 of 85 at six months). In Eichhorn 2003, ECTR participants that intraoperatively went to open surgery were excluded from the analysis. In Mackenzie 2000 there was no information about the number of participants initially enrolled. In Agee 1992, the authors reported that only one to two participants in each group were missing for the activities of daily living outcome, but only said "a small number of observations was missing" when referring to other variables. Participants with bilateral involvement were also excluded from the analysis. We judged the risk of bias in this study to be unclear.
For the rest of the trials, insufficient information was provided to draw a safe conclusion.
Some but not all pre-specified outcomes and time points were reported in an adequate way in Brown 1993, Eichhorn 2003, Jacobsen 1996, Rab 2006, Saw 2003, Schäfer 1996, Stark 1996, Trumble 2002 and Wong 2003. Also, all the trials presented as abstracts provided insufficient information (Giele 2000; Incoll 2004; Koskella 1996; Sørensen 1997; Werber 1996).
No numerical data were provided for any of the outcomes in Foucher 1993. Agee 1992, Ferdinand 2002, Mackenzie 2000, Malhotra 2007, Stark 1996, Tian 2007 and Westphal 2000 gave no standard deviations (SDs) for any of the outcomes and we were not able to extract them from other statistics (for example, P values). Only diagrams, with no further information (definite outcomes, SDs, etc), were provided in Erdmann 1994, and Macdermid 2003. Overall, poor reporting characterised the majority of the included trials.
Funnel plots for all outcomes with at least ten studies appeared reasonably symmetric. However, the small differences between fixed-effect and random-effects models for complications might suggest that small studies give different results compared to large studies.
Other potential sources of bias
Only six of the studies were judged to be free of other bias (Atroshi 2006; Brown 1993; Malhotra 2007; Trumble 2002; Tüzüner 2008; Wong 2003). They clearly did not have baseline differences and the trials were not sponsored by a party with vested interests. Atroshi 2006, Hoefnagels 1997, Malhotra 2007 and Trumble 2002 had a form of financial support, but from an academic source.
In Agee 1992, the authors declared a conflict of interest as the study was supported in part by the manufacturer of the device used for the release. Nine studies clearly reported no conflict of interest (Atroshi 2006; Brown 1993; Ferdinand 2002; Incoll 2004; Macdermid 2003; Malhotra 2007; Trumble 2002; Tüzüner 2008; Wong 2003).
None of the other studies provided sufficient information to draw a safe conclusion regarding baseline differences or financial support. Therefore, we judged their risk of bias as unclear.
Effects of interventions
Endoscopic versus open and modified open carpal tunnel release
Short-term efficacy results (three months or less)
Out of 25 studies that compared ECTR with OCTR, 18 presented results on the short-term effects (Agee 1992; Atroshi 2006; Brown 1993; Dumontier 1995; Erdmann 1994; Ferdinand 2002; Giele 2000; Hoefnagels 1997; Incoll 2004; Macdermid 2003; Malhotra 2007; Saw 2003; Sørensen 1997; Stark 1996; Tian 2007; Trumble 2002; Tüzüner 2008; Westphal 2000). In 11 of the 18 studies, no differences were found between the groups for the outcomes assessed (Agee 1992; Atroshi 2006; Brown 1993; Dumontier 1995; Erdmann 1994; Hoefnagels 1997; Macdermid 2003; Saw 2003; Stark 1996; Tüzüner 2008; Westphal 2000). Seven studies concluded a superiority of ECTR over OCTR (Ferdinand 2002; Giele 2000; Incoll 2004; Malhotra 2007; Sørensen 1997; Tian 2007; Trumble 2002) ( Table 1; Table 2).
Meta-analysis was possible for five studies that assessed the Symptom Severity Scale (SSS) and the same five had assessed the Function Status Scale (FSS) (Atroshi 2006; Hoefnagels 1997; Rab 2006; Trumble 2002; Westphal 2000). SSS and FSS as described in the original study of Levine correspond to a scale from one to five, with one being the most favourable outcome (Levine 1993). Westphal 2000 reported a modification of SSS and FSS, which necessitated the use of standardised mean difference (SMD) as the summary estimate. Summary estimates showed no statistically significant differences between ECTR and OCTR either in SSS (five studies, 551 participants, SMD -0.13, 95% confidence interval (CI) -0.47 to 0.21) ( Analysis 1.3; Figure 3) or in FSS (five studies, 551 participants, SMD -0.23, 95% CI -0.60 to 0.14) ( Analysis 1.4; Figure 4). In both meta-analyses there was large heterogeneity, with I
|Figure 3. Forest plot of comparison: 1 Endoscopic versus open or mini-open carpal tunnel release, outcome: 1.3 Symptom Severity Scale (Levine) at 3 months or less.|
|Figure 4. Forest plot of comparison: 1 Endoscopic versus open or mini-open carpal tunnel release, outcome: 1.4 Function Status Scale at 3 months or less.|
Meta-analysis of four studies assessing a pain score ( Analysis 1.5) showed that pain did not differ significantly between ECTR and OCTR (four studies, 358 participants, SMD -0.21, 95% CI -0.72 to 0.30). A similar conclusion is supported by the five studies assessing pain on a dichotomous scale (Agee 1992; Dumontier 1995; Malhotra 2007; Stark 1996; Wong 2003) ( Analysis 1.8): a difference in pain between ECTR and OCTR could not be demonstrated nor refuted (five studies, 348 participants, risk ratio (RR) 0.69, 95% CI 0.33 to 1.45). A large heterogeneity was found (I
No statistically significant difference in numbness was found when synthesizing five studies comparing ECTR with OCTR (5 studies, 435 participants, RR 1.14, 95% CI 0.76 to 1.71) ( Analysis 1.9).
Regarding grip strength, the summary estimate from the six studies included in the meta-analysis favoured ECTR (6 studies, 560 participants, SMD 0.36, 95% CI 0.09 to 0.63) ( Analysis 1.10; Figure 5). This demonstrates a statistically significant difference. Assuming an SD of 11 (as in Atroshi 2006), this corresponds to a mean difference (MD) of 4 kg (95% CI 1 to 6.9 kg) favouring ECTR when compared with OCTR. This difference is relatively low and probably not clinically significant.
|Figure 5. Forest plot of comparison: 1 Endoscopic versus open or mini-open carpal tunnel release, outcome: 1.10 Grip strength at 3 months or less.|
Long-term efficacy results (more than three months)
Eleven studies reported long-term symptom outcomes, comparing ECTR with conventional open release (Agee 1992; Atroshi 2006; Dumontier 1995; Eichhorn 2003; Erdmann 1994; Ferdinand 2002; Mackenzie 2000; Malhotra 2007; Schäfer 1996; Stark 1996; Trumble 2002). No significant differences in symptoms were found in any of the studies, except in Ferdinand 2002, which favoured ECTR, and Macdermid 2003, which favoured OCTR. Atroshi 2006 also found a slight superiority of ECTR for pain at one year, which however was not evident at five-year follow-up (Atroshi 2009). Only Wong 2003 reported long-term symptom outcomes comparing ECTR with mini-open release.
Four studies assessed overall improvement, reporting no significant difference between ECTR and open release (four studies, 317 participants, RR 1.04, 95% CI 0.95 to 1.14, I
Return to work
Twenty studies assessed the time to return to work, expressed in many different formats. In 10 of them, ECTR-treated participants had a significantly earlier recovery (Agee 1992; Atroshi 2006; Benedetti/Sennwald 1995; Brown 1993; Erdmann 1994; Saw 2003; Schäfer 1996; Stark 1996; Tian 2007; Trumble 2002; Werber 1996). In one study OCTR participants returned earlier to work (Dumontier 1995), and seven studies recorded a non-significant difference (Foucher 1993; Hoefnagels 1997; Jacobsen 1996; Koskella 1996; Macdermid 2003; Sørensen 1997; Westphal 2000). In Atroshi 2006, return to work among participants who were on sick leave before surgery (n = 16) was eight days earlier for those who underwent OCTR (MD 8.00, 95% CI -62.59 to 78.59), but for participants not on sick leave before surgery (n = 112), it was five days earlier for the ECTR group (MD -5.00, 95% CI -11.49 to 1.49). Synthesizing the outcome from both subgroups yielded an MD which favoured ECTR by 4.9 days; however, not significantly (MD -4.89, 95% CI -11.35 to 1.57).
Meta-analysis was possible for four of the studies (Atroshi 2006; Benedetti/Sennwald 1995; Jacobsen 1996; Saw 2003). The mean estimate significantly favoured ECTR, revealing a faster return to work by on average of eight days (4 studies, 274 participants, MD -8.10, 95% CI -14.28 to -1.92) ( Analysis 1.20). This estimate did not significantly change with the addition of the group of participants on sick leave before surgery, assessed in Atroshi 2006 (MD altered to -7.99, 95% CI -13.93 to -2.05). However, the between-studies variation was important (I
Time to return to work is potentially subject to several confounding factors and may substantially differ between different national health systems or different patient groups (in terms of age, occupation, etc) (Cowan 2012). However, we assume that the arms in an RCT are similar in all the factors that might affect the recovery to work or activities of daily living. Therefore, we anticipate that despite the anticipated high heterogeneity in absolute values, the difference between groups is a reliable outcome.
Very few participants (14 out of 1508) reported major complications resulting in permanent damage or major impairments (for example, complex regional pain syndrome (CRPS), severe sympathetic reflex dystrophy, algodystrophy or severe pain). In one study, one mild case and one severe case of CRPS were recorded out of 25 hands in the OCTR group (Trumble 2002). There were two reports of CRPS (one in each group) in Benedetti/Sennwald 1995, and one report in the ECTR group of Foucher 1993. Malhotra 2007 reported two individuals with symptoms consistent with sympathetic reflex dystrophy, in the OCTR group. In one participant the symptoms were mild and resolved briefly, while in the other the symptoms were more protracted. Agee 1992 reported one injury to the deep motor branch of the ulnar nerve in an OCTR-treated participant. Atroshi 2006 reported no nerve, vascular, or tendon injuries, and no wound complications at one year; however, at five years' follow-up, they reported five ECTR and three OCTR participants with moderate or severe pain (Atroshi 2009). We classified these events as major complications. Meta-analysis of 15 studies (in six of which major complications occurred) did not reveal any difference between ECTR and conventional OCTR (15 studies, 1508 participants, RR 1.00, 95% CI 0.38 to 2.64, I
Upon synthesis of 19 studies from which data on minor complications could be extracted, ECTR appeared safer than open release (18 studies in total, 17 studies with events, 1786 participants, RR 0.55, 95% CI 0.38 to 0.81, I
|Figure 6. Forest plot of comparison: 1 Endoscopic versus open or mini-open carpal tunnel release, outcome: 1.24 Minor complications.|
ECTR was associated with more transient nerve problems (for example, neurapraxia, numbness or paraesthesiae) and OCTR with more wound problems (for example, infection, hypertrophic scarring or scar tenderness). In a few participants, ECTR had to be abandoned and OCTR was performed instead. Thirteen hands randomised to ECTR were finally converted to OCTR owing to intraoperative difficulties (one in Atroshi 2006, one in Benedetti/Sennwald 1995, two in Foucher 1993 and nine in Saw 2003).
In 19 studies, the total number of participants with complications was given or this information could be extracted. ECTR gives a significantly lower rate of complications (19 studies, 1850 participants, RR 0.60, 95% CI 0.40 to 0.90), providing a 40% less risk (95% CI 60 to 10) ( Analysis 1.25; Figure 7).
|Figure 7. Forest plot of comparison: 1 Endoscopic versus open or mini-open carpal tunnel release, outcome: 1.25 Total complications.|
Data on participants with recurrence of symptoms could be extracted from 12 of the studies. Meta-analysis also favoured ECTR but this difference was not significant (nine studies with events, RR 0.81, 95% CI 0.46 to 1.42) ( Analysis 1.21).
The need for repeated surgery was assessed in 11 studies. There was no statistical significant difference between ECTR and OCTR release (nine studies with events, 1116 participants, RR 1.06, 95% CI 0.54 to 2.08, I
Subgrouping to different open release techniques (conventional OCTR/mini-open techniques)
For pain assessed in the short term, in the studies comparing ECTR to conventional OCTR (Atroshi 2006; Saw 2003), ECTR was less painful (two studies, 278 participants, SMD -0.41, 95% CI -0.65 to -0.18) ( Analysis 1.5). Assuming an SD of 23 for a score on a range from zero (no pain) to 100 (severe pain) (extracted from Atroshi 2006), this SMD corresponds to an MD of -9.43 (95% CI -14.95 to -4.14). Two additional studies comparing ECTR to modified OCTR were highly heterogeneous (I
Regarding grip strength assessed in the short term ( Analysis 1.10), five studies compared ECTR with conventional OCTR, and showed a significantly stronger grip for ECTR participants (five studies, 540 participants, SMD 0.40, 95% CI 0.10 to 0.71) (Atroshi 2006; Benedetti/Sennwald 1995; Brown 1993; Dumontier 1995; Saw 2003). Assuming an SD of 11 (as in Atroshi 2006), this corresponds to an MD of 4.4 kg (95% CI 1.1 to 7.81 kg). This difference is relatively small and probably not clinically significant. Only one study compared ECTR with modified open release and it showed a nonsignificant difference at 12 weeks (20 participants, SMD 0.14, 95% CI -0.48 to 0.76) (Rab 2006).
No subgroup analysis was possible for measurements taken at long-term follow-up, as only Wong 2003 from among the ECTR versus mini-open trials reported such information.
No differences were demonstrated between subgroups for open and mini-open techniques for any of the safety outcomes ( Analysis 1.21; Analysis 1.23; Analysis 1.24; Analysis 1.25). We found subgroup differences only in re-operation rate ( Analysis 1.22). However, the heterogeneity was very high (I
Subgrouping to different ECTR techniques (one-portal or two-portal ECTR)
We also assessed the outcomes presented above distinguishing between the two main ECTR techniques (one-portal and two-portal techniques) ( Analysis 2.1; Analysis 2.2; Analysis 2.3; Analysis 2.4; Analysis 2.5; Analysis 2.6; Analysis 2.7; Analysis 2.8; Analysis 2.9; Analysis 2.9; Analysis 2.10; Analysis 2.11; Analysis 2.12; Analysis 2.13; Analysis 2.14; Analysis 2.15; Analysis 2.16; Analysis 2.17; Analysis 2.18; Analysis 2.19; Analysis 2.20; Analysis 2.21; Analysis 2.22; Analysis 2.23; Analysis 2.24; Analysis 2.25). No important differences were found between the two endoscopic techniques. In five studies assessing pain as a dichotomous outcome less than three months postoperatively, one-portal ECTR (two studies) was better than two-portal ECTR (three studies) (test for subgroup differences: P = 0.0002, I
1. In the sensitivity analysis for attrition bias, we included only nine studies at low risk of bias (Atroshi 2006; Benedetti/Sennwald 1995; Hoefnagels 1997; Jacobsen 1996; Rab 2006; Saw 2003; Schäfer 1996; Stark 1996; Tüzüner 2008) ( Analysis 3.1; Analysis 3.2; Analysis 3.3; Analysis 3.4; Analysis 3.5; Analysis 3.6; Analysis 3.7; Analysis 3.8; Analysis 3.9; Analysis 3.10; Analysis 3.11; Analysis 3.12; Analysis 3.13). Pain assessed as continuous data in three studies, seemed now to favour ECTR at three months ( Analysis 3.3; Analysis 3.4; Analysis 3.5). However, in studies adequately addressing incomplete data, ECTR and OCTR do not seem to differ in grip strength and complication rates ( Analysis 3.6; Analysis 3.12; Analysis 3.13). This sensitivity analysis revealed no other differences when compared with the main results.
2. We also performed sensitivity analysis excluding the studies that did not adjust appropriately for participants with bilateral involvement. Thirteen studies in total were excluded from the analysis. In three studies only people with bilateral CTS were included (Giele 2000; Incoll 2004; Stark 1996). Nine studies included bilateral CTS but provided no further details (Brown 1993; Erdmann 1994; Jacobsen 1996; Koskella 1996; Mackenzie 2000; Saw 2003; Tian 2007; Trumble 2002; Tüzüner 2008), whereas Sørensen 1997 provided no information about unilateral or bilateral involvement ( Analysis 4.1; Analysis 4.2; Analysis 4.3; Analysis 4.4; Analysis 4.5; Analysis 4.6; Analysis 4.7; Analysis 4.8; Analysis 4.9; Analysis 4.10; Analysis 4.11; Analysis 4.12; Analysis 4.13; Analysis 4.14; Analysis 4.15; Analysis 4.16). From this analysis, minor complications still favour ECTR, but this is marginally insignificant (RR 0.54, 95% CI 0.32 to 0.94) ( Analysis 4.15). In total complications no differences are found (RR 0.72, 95% CI 0.45 to 1.14) ( Analysis 4.16). Grip strength remains on average greater for ECTR participants in the short term (four studies, SMD 0.52, 95% CI 0.03 to 1.02) ( Analysis 4.11), corresponding to an MD of 5.72 kg (95% CI 0.33 to 11.22 kg) assuming an SD of 11 (Atroshi 2006).
4. We performed a post hoc analysis using Peto's odds ratio for complications. The results do not materially change compared to the Mantel-Haenszel fixed-effect method with RR.
5. Two studies had bilateral involvement and results were not adjusted (Rab 2006; Wong 2003). Analyses of these studies were undertaken assuming that the correlation coefficient between the two groups was 0.5. Changing the coefficient to 0.1 and 0.9 did not materially change the conclusions on pain, SSS, functional status, or grip strength.
Summary of main results
This review included 28 studies that compared endoscopic carpal tunnel release (ECTR) with standard open carpal tunnel release (OCTR) or modified OCTR for carpal tunnel syndrome (CTS). In total, 2586 hands were assessed. Twenty-three studies compared ECTR with standard OCTR, three studies with modified OCTR and two studies compared ECTR with both standard and modified OCTR. Sixteen studies addressed a one-portal technique for ECTR and nine studies a two-portal technique. The exact endoscopic technique was not defined in three studies.
There was very limited evidence from assessment of overall improvement of overall satisfaction. Clinical scores (Symptom Severity Scale (SSS) and Function Status Scale (FSS)), assessed by five studies, did not indicate that any treatment option was superior to another. Pain after ECTR and OCTR (conventional or modified) was also equal, although ECTR was superior when compared with conventional OCTR. However, assessment of pain scales favoured ECTR at short-term follow-up, when only studies with a low risk of attrition bias were considered.
No differences were found in the incidence of residual numbness.
The meta-analysis revealed that participants treated with ECTR had an increased grip strength (standardised mean difference (SMD) 0.36, 95% CI 0.09 to 0.63), especially when compared with conventional OCTR. However, the difference was relatively small and probably not clinically significant, corresponding to a mean diffference (MD) of 4 kg. Grip strength was found to be equal between ECTR and OCTR when only studies with a low risk of attrition bias were considered.
From the studies assessing clinical outcomes at least three months after the surgery, there was no evidence that any technique was superior for any of the outcomes assessed, except for grip strength.
For grip strength, data from two studies indicated that ECTR was superior to conventional OCTR (SMD 1.13, 95% CI 0.56 to 1.71). We estimated the corresponding MD to be 11 kg, which is potentially clinically relevant.
Return to work
Meta-analysis of four studies assessing return to work showed a faster return to work for ECTR by eight days. This superiority was marginal when we removed studies with inappropriate adjustment for participants with bilateral involvement from the meta-analysis.
The incidence of complications was assessed in 26 studies.
Only a small number of major complications was found for both treatment options (0.9% for both ECTR and OCTR). These were mainly instances of complex regional pain syndrome. No difference was revealed from the meta-analysis of 15 studies (1508 participants).
Regarding minor complications, there was a lower incidence for ECTR treated participants. Meta-analysis assessed 18 studies (1786 participants) showing an incidence of 5% for ECTR and 10.2% for OCTR participants (11.3% for conventional OCTR). From only three studies comparing ECTR with modified OCTR, no difference was found. ECTR more often resulted in transient nerve problems (caused by intraoperative injury), whereas OCTR had more wound problems (for example, infection or scar tenderness).
Assessing the total number of complications as extracted from 19 studies (1850 participants), ECTR was superior. The incidence was 5.5% for ECTR and 8.7% for OCTR. The additional benefit with ECTR was marginal when we excluded from the meta-analysis studies with inappropriate adjustment for participants with bilateral involvement.
The incidence of recurrence (12 studies assessed, 1228 participants) was equal between ECTR and OCTR (3.2% and 4.6% respectively).
Reoperation rates were also equal (2.8% for ECTR and 2.5% for OCTR), according to data from 10 studies (1116 participants).
Overall completeness and applicability of evidence
In this review, we included open and different mini-open (modified) techniques. The extent of the skin incision may differ among surgeons, especially between the different modified open techniques. In order to minimise the effect of this potential bias, we performed subgroup analysis assessing separately open and modified open techniques compared with ECTR.
Different ECTR techniques could also potentially differ in outcome, in particular techniques with one rather than two incisions. However, there was no evidence from the subgroup analysis that either of the ECTR techniques (one-portal or two-portal) was superior to the other. There was no study directly comparing different ECTR techniques with each other.
Surgical treatment of CTS is generally performed under local anaesthesia, unless special reasons exist. Therefore, in most of the studies such information was not even provided. This is why we did not systematically extract or assess the type of anaesthesia in this review.
The primary outcome of this systematic review, being the overall improvement of symptoms at less than three months, was reported in only one of the included RCTs.
Five of the studies were presented only in an abstract (Giele 2000; Incoll 2004; Koskella 1996; Sørensen 1997; Werber 1996). Therefore, the reports of these studies provided little information regarding the exact methodology and only some information regarding the outcomes.
In some studies bilateral involvement was also assessed. However, this was not always clearly defined.
CTS can be treated by different specialties such as plastic surgeons or general orthopaedic surgeons, who may or may not be specialised in hand surgery. To our knowledge, there are no studies comparing outcomes of CTR based on the specialty of the surgeon. Furthermore, CTR is a common operation, and all specialities are expected to have adequate experience. Therefore, we did not perform a subgroup analysis analysing the studies separately, according to the specialty of the surgeon.
We did not include assessment of cost effectiveness among the aims of this review. However, two of the RCTs assessed the cost-effectiveness of ECTR and OCTR. Saw 2003 reported an increased cost of the equipment for ECTR versus OCTR, by GBP 98. However, this was accompanied with a faster return to work by eight days, resulting in an overall net saving of GBP 438 per employed patient treated with ECTR. Trumble 2002 reported no difference in cost between the two interventions.
Quality of the evidence
The quality of evidence in this review may be considered as quite low. Five of the studies were presented only in abstracts, with not enough information regarding the risk of bias. This might also increase the publication bias.
Funnel plots do not appear to be asymmetric, which suggestis that there are no systematic differences between small and large studies. The apparent symmetry could indicate the absence of clear evidence of publication bias, although we cannot exclude the possibility of such bias.
For selection bias, attrition bias or other bias (mainly from baseline differences but also from the financing of the studies), we could not reach a safe judgement regarding whether there was a high or low risk of bias. Therefore, in many cases we considered the risk of bias to be unclear.
Blinding of participants was considered not to be applicable in any of the studies. However, such a judgement was due to the nature of interventions (it was not possible to hide the skin incision from participants until the last follow-up), rather than reflecting a low quality of information.
Heterogeneity in SSS and FSS and statistical imprecision (low sample size, low number of events or both) limit further the credibility of the findings of this review.
Potential biases in the review process
Agreements and disagreements with other studies or reviews
Gerritsen 2001 found no evidence for a superiority of ECTR, based on seven studies. However, no pooling of the studies was performed in this systematic review. It seems that ECTR produces more transient nerve problems (neurapraxia, numbness or paraesthesia) and OCTR more wound problems (infection, hypertrophic scarring or scar tenderness). Gerritsen 2001 also stated that because OCTR is technically less demanding, it incurs a lower risk of complications and fewer added costs. However, the authors did not present any evidence for this, as neither complications nor cost effectiveness were assessed in the review.
Sanati 2011 also found an earlier return to work after ECTR. The authors highlighted the remarkable inconsistencies in how different randomised clinical trials (RCTs) examined return to work as an outcome measure.
Boeckstyns 1999 focused on complications, assessing 54 publications (from case reports to RCTs). They found no differences, except that ECTR resulted in a higher rate (4.3% versus 0.9%) of reversible nerve damage (that is, transient neurapraxias).
Benson 2006 (literature search up to 2001) assessed 68 articles that included complications as one of the outcomes, irrespective of the type of study (the review even included studies with no comparator). The authors focused on complications caused by damage to nerves, arteries or tendons. Complications not involving structural injury or which were subjective in nature (for example, CRPS, scar hypersensitivity and would healing problems) were not included. The review found the overall proportion of complications for the OCTR technique to be 0.74%, and for the ECTR technique to be 1.63% (P < 0.005). However, this difference was mainly the result of an increased incidence of transient neurapraxias after ECTR. Transient neurapraxias were reported in 1.45% of ECTR cases and in only 0.25% of OCTR cases. When transient neurapraxias were not encountered, ECTR seemed to be safer than OCTR (OCTR 0.49% versus ECTR 0.19%; P < 0.005). When only extrabursal ECTR was analysed, this superiority of ECTR was even more obvious. Transbursal ECTR was associated with an increased rate of transient neurapraxias. Finally, damage to median or ulnar nerves was not statistically different between open or endoscopic techniques (0.11% versus 0.13%). The findings of this study seem to be comparable with our findings. It seems from our data (on minor complications) that ECTR results in more frequent transient neurapraxias, while OCTR results in wound healing problems, infections or painful scars. Because Benson 2006 excluded complications closely related to OCTR, OCTR appeared safer than ECTR. Another major nonstructural complication, mainly found after OCTR, namely CRPS, was also not included in Benson 2006, which therefore potentially underestimates the major complications of OCTR.
Thoma 2004 (search from 1989 up to 2001/2002) assessed 13 RCTs. The authors found that grip and pinch strength favoured ECTR at three months. There was no difference in pain and return to work in the studies assessed (four and three studies, respectively). They found that ECTR was three times more likely to cause neurapraxia (reversible nerve injury) than OCTR (six RCTs, pooled odds ratio (OR) 0.336, 95% CI 0.117 to 0.908). However, ECTR was associated with less scar tenderness (pooled OR 3.78, 95% CI 2.16 to 6.59). Both neurapraxia and scar tenderness were included as minor complications in our systematic review. Thoma 2004 did not include summarised complications, so a direct comparison with their findings is not possible. However, it has also been obvious during our data extraction that most of the minor complications after ECTR (when mentioned in detail) were transient neurapraxias, while most of the OCTR minor complications were due to skin problems (for example, scar tenderness or skin infections). However, such outcomes (specific complications) were not included in our systematic review.
A main difference between this study and ours is the rate of complications per intervention. It seems that 5.5% versus 8.5% for total complications and 5% versus 9.5% for minor complications for ECTR and OCTR respectively are much higher that the incidence found in Benson 2006, or other similar studies. This is mainly due to the definition of complications as given by the authors and the complications included from each study. For example, we considered all minor complications in our review, including wound infections or painful scars, which increased the total number. In addition, our review included the additional studies by Tian 2007 and Atroshi 2006, which reported an increased rate of complications.
The transbursal approach for ECTR was associated with an increased rate of complications. This was the main reason why this approach was abandoned and the extrabursal technique finally prevailed. Benson 2006, from a higher number of studies (although not limited to RCTs), assessed 22,327 ECTR and 5669 OCTR cases. The authors concluded that the transbursal approach to the carpal tunnel, which was popular when the endoscopic technique was first developed, was associated with a significantly higher complication rate of neurapraxia when compared with the extrabursal approach (2.8% versus 0.9%, respectively). However, in our review we did not exclude participants treated with the transbursal approach. Had we done so, we would expect the rate of neurapraxias (that is, minor complications) after extrabursal ECTR to be even lower than that reported for both approaches in this current review.
From our review, ECTR seems to provide a lower rate of minor complications (Figure 7). It seems from the data that three of the studies provide a risk ratio that (not significantly) favours OCTR. All of these studies were published at a time when experience with the technique was just developing (Brown 1993; Jacobsen 1996; Werber 1996). For example, at least one was performed with a transbursal technique that was later abandoned due to a higher risk of complications (Jacobsen 1996).
From 1966 up to 2001, more than twice as many cases of ECTR were available in the literature than OCTR. However, ECTR was only first described in 1989 (Benson 2006), and the increased interest in the safety of the new ECTR approach would be expected to lead to increased publication of complications even if the complication rate were the same between the two approaches. Our finding that the complication rate in RCTs is the same between the two approaches strongly suggests that this was likely to have been the case.
Implications for practice
Grip strength was greater in endoscopic carpal tunnel release at three months or less after the surgery; however, only a few studies assessed this outcome and the difference is probably not clinically significant. At time points more than three months after the surgery, ECTR showed a more pronounced superiority in grip strength, which was probably clinically significant. However, the value of this finding is moderated by the fact that only two studies reported this outcome. Return to work was faster after endoscopic release, by eight days.
Given the results of this meta-analysis, endoscopic surgery might offer an advantage in time to return to work and recovery of grip strength. This might have implications for those who rely on hand function day to day, in whom early recovery of grip strength and early return to work or full daily activities is important.
Endoscopic carpal tunnel release is as safe as open release (that is, the number of major complications is similar). There is some evidence in favour of endoscopic release over open release in the rate of minor complications and the total number of complications.
The findings in this review should be considered with caution, as their credibility is limited by shortcomings in study design and in the reporting of the included trials.
Implications for research
There is still uncertainty about whether endoscopic carpal tunnel release produces a better outcome than open carpal tunnel release or modified open techniques in terms of pain, relief of symptoms and functional recovery, as this review found no clear differences for these important outcomes between the techniques. However, the few studies that assessed these outcomes were at high risk of bias, which prevents us from reaching a safe conclusion about the potential superiority of any of the techniques. There is a need for higher quality study design and reporting to increase the credibility of the findings. Studies should regularly use clinical questionnaires to measure outcomes. Moreover, investigators should consider collecting information about adverse events prospectively and report them in detail. Research questions regarding the added benefit of endoscopic carpal tunnel release in relieving pain and improving functional recovery in either the short or long term are not yet answered.
More studies should be conducted to assess the cost-effectiveness of endoscopic carpal tunnel release compared with conventional open carpal tunnel release.
Ms Angela Gunn, Trials Search Co-ordinator, Cochrane Neuromuscular Disease Group, MRC Centre for Neuromuscular Diseases, National Hospital for Neurology and Neurosurgery, London, UK developed the search strategies.
The Cochrane Neuromuscular Disease Group for editorial and methodological support.
The editorial base of the Cochrane Neuromuscular Disease Group is supported by the MRC Centre for Neuromuscular Diseases.
Data and analyses
- Top of page
- Summary of findings [Explanations]
- Authors' conclusions
- Data and analyses
- Contributions of authors
- Declarations of interest
- Sources of support
- Differences between protocol and review
- Index terms
Appendix 1. Cochrane Neuromuscular Disease Group Specialized Register search strategy
#1 MeSH DESCRIPTOR Carpal Tunnel Syndrome [REFERENCE] [STANDARD]
#2 "carpal tunnel" [REFERENCE] [STANDARD]
#3 ("nerve entrapment" or "nerve compression" or "entrapment neuropath*") and carpal [REFERENCE] [STANDARD]
#4 #1 or #2 or #3 [REFERENCE] [STANDARD]
#5 endoscop* or octr or ectr [REFERENCE] [STANDARD]
#6 #4 and #5 [REFERENCE] [STANDARD]
#7 (#4 and #5) AND (INREGISTER) [REFERENCE] [STANDARD]
Appendix 2. The Cochrane Library (CENTRAL)
#1 "Carpal Tunnel Syndrome"
#2 ("nerve entrapment" or "nerve compression" or "entrapment neuropathy" or "entrapment neuropathies")
#4 #1 or (#2 and #3)
#5 endoscop* or OCTR or ECTR or releas*
#6 #4 and #5
Appendix 3. MEDLINE (OvidSP) search strategy
Database: Ovid MEDLINE(R) <1946 to November Week 1 2013>
1 randomized controlled trial.pt. (389866)
2 controlled clinical trial.pt. (89904)
3 randomized.ab. (287333)
4 placebo.ab. (156850)
5 drug therapy.fs. (1767223)
6 randomly.ab. (199448)
7 trial.ab. (302482)
8 groups.ab. (1276425)
9 or/1-8 (3299027)
10 exp animals/ not humans.sh. (4060470)
11 9 not 10 (2809295)
12 Carpal Tunnel Syndrome.mp. or Carpal Tunnel Syndrome/ (7915)
13 (carp$ tunn$ or tunn$ syndrom$ or carp$ syndrom$).mp. (9575)
14 (nerve entrapment or nerve compression or entrapment neuropath$).mp. (11216)
15 median nerve entrapment.mp. (99)
16 nerve compression syndromes/ (9072)
17 or/12-16 (19390)
18 endoscop$.mp. (159054)
19 OCTR.mp. (35)
20 ECTR.mp. (59)
21 releas$.mp. (600120)
22 or/18-21 (757515)
23 11 and 17 and 22 (328)
Appendix 4. EMBASE (OvidSP) search strategy
Database: Embase <1980 to 2013 Week 46>
1 crossover-procedure/ (38971)
2 double-blind procedure/ (118651)
3 randomized controlled trial/ (360008)
4 single-blind procedure/ (18506)
5 (random$ or factorial$ or crossover$ or cross over$ or cross-over$ or placebo$ or (doubl$ adj blind$) or (singl$ adj blind$) or assign$ or allocat$ or volunteer$).tw. (1303033)
6 or/1-5 (1385895)
7 exp animals/ (19025289)
8 exp humans/ (14995220)
9 7 not (7 and 8) (4030069)
10 6 not 9 (1245034)
11 limit 10 to embase (962420)
12 Carpal Tunnel Syndrome.mp. or Carpal Tunnel Syndrome/ (11573)
13 (carp$ tunn$ or tunn$ syndrom$ or carp$ syndrom$).mp. (14487)
14 (nerve entrapment or nerve compression or entrapment neuropath$).mp. (13134)
15 nerve compression/ (11098)
16 or/12-15 (25116)
17 carpal tunnel release/ (61)
18 (endoscop$ or releas$ or OCTR or ECTR).mp. (1055110)
19 or/17-18 (1055110)
20 11 and 16 and 19 (201)
Appendix 5. Search for ongoing trials
Databases: http://www.clinicaltrials.gov, http://www.controlled-trials.com (ISRCTN Register, Action Medical Research (UK), The Wellcome Trust (UK), UK trials (UK)), http://www.ukctg.nihr.ac.uk/default.aspx and http://www.who.int/ictrp/en/
# "carpal tunnel"
Contributions of authors
Conceiving the review: Rob Scholten (RS), Haris S Vasiliadis (HSV)
Designing the first drafts of the title proposal and the review protocol: HSV
Feedback for the final title proposal and protocol: Georgia Salanti (GS), RS, Ian Shrier (IS)
Co-ordinating the review: HSV
Data collection for the review: HSV, PG
Undertaking manual searches: HSV
Screening search results:HSV, PG, IS
Organizing retrieval of papers: HSV, RS
Screening retrieved papers against inclusion criteria: HSV, PG, IS
Appraising quality of papers: HSV, PG
Abstracting data from papers: HSV, PG
Writing to authors of papers for additional information: HSV
Providing additional data about papers: RS
Obtaining and screening data on unpublished studies: HSV
Data management for the review: HSV, GS
Entering data into RevMan 5: HSV
Analysis of data: HSV, GS. RS, IS
Interpretation of data: HSV, GS, IS, RS
Writing the review: HSV, GS
Securing funding for the review: not applicable
Performing previous work that was the foundation of the present study: RS
Guarantor of the review (one author): HSV
Statistical analysis: HSV, GS
Declarations of interest
HSV received travel support from manufacturer of instrumentation for mini-open and endoscopic release to attend orthopaedic conference. Additionally, he is the Principal Investigator in an ongoing RCT comparing ECTR versus mini-open carpal tunnel release.
PG, GS, IS, RS: none known.
Sources of support
- Haris S Vasiliadis, Greece.none
- Petros Georgoulas, Greece.none
- Ian Shrier, Canada.
- Georgia Salanti, Greece.GS received funding from the European Research Council (IMMA 260559)
- Rob JPM Scholten, Netherlands.
- None, Not specified.
Differences between protocol and review
Although we planned to assess the analysis taking out studies with high or unclear risk of bias for allocation concealment, this was not done. Only two studies would have been included in the analysis (Atroshi 2006; Tüzüner 2008), and therefore no valuable information would have been found.
Some figures were not cited in the text, so are included only in the full version of this systematic review (Figure 8; Figure 9; Figure 10; Figure 11; Figure 12; Figure 13; Figure 14; Figure 15; Figure 16; Figure 17; Figure 18; Figure 19; Figure 20).
|Figure 8. Forest plot of comparison: 1 Endoscopic versus open or mini-open carpal tunnel release, outcome: 1.6 Pain at 3 months or less (corr = 0.9).|
|Figure 9. Forest plot of comparison: 1 Endoscopic versus open or mini-open carpal tunnel release, outcome: 1.8 Pain (dichotomous) at 3 months or less.|
|Figure 10. Forest plot of comparison: 1 Endoscopic versus open or mini-open carpal tunnel release, outcome: 1.9 Numbness (dichotomous) at 3 months or less.|
|Figure 11. Forest plot of comparison: 1 Endoscopic versus open or mini-open carpal tunnel release, outcome: 1.19 Grip strength at more than 3 months.|
|Figure 12. Forest plot of comparison: 1 Endoscopic versus open or mini-open carpal tunnel release, outcome: 1.20 Time to return to work.|
|Figure 13. Forest plot of comparison: 1 Endoscopic versus open or mini-open carpal tunnel release, outcome: 1.21 Recurrence.|
|Figure 14. Forest plot of comparison: 1 Endoscopic versus open or mini-open carpal tunnel release, outcome: 1.22 Reoperations.|
|Figure 15. Forest plot of comparison: 1 Endoscopic versus open or mini-open carpal tunnel release, outcome: 1.23 Major complications.|
|Figure 16. Funnel plot of comparison: 1 Endoscopic versus open or mini-open carpal tunnel release, outcome: 1.21 Recurrence.|
|Figure 17. Funnel plot of comparison: 1 Endoscopic versus open or mini-open carpal tunnel release, outcome: 1.22 Reoperations.|
|Figure 18. Funnel plot of comparison: 1 Endoscopic versus open or mini-open carpal tunnel release, outcome: 1.23 Major complications.|
|Figure 19. Funnel plot of comparison: 1 Endoscopic versus open or mini-open carpal tunnel release, outcome: 1.24 Minor complications.|
|Figure 20. Funnel plot of comparison: 1 Endoscopic versus open or mini-open carpal tunnel release, outcome: 1.25 Total complications.|
Medical Subject Headings (MeSH)
MeSH check words
* Indicates the major publication for the study