Gerritsen AAM, de Vet HCW, Scholten RJPM, Bertelsmann FW, de Krom MCTFM, Bouter LM. Splinting vs Surgery in the Treatment of Carpal Tunnel SyndromeA Randomized Controlled Trial. JAMA. 2002;288(10):1245–1251. doi:10.1001/jama.288.10.1245
Author Affiliations: Institute for Research in Extramural Medicine, Vrije Universiteit Medical Center, Amsterdam, the Netherlands (Drs Gerritsen, de Vet, and Bouter); Dutch Cochrane Center/Department of Clinical Epidemiology and Biostatistics, Academic Medical Center, University of Amsterdam, Amsterdam, the Netherlands (Dr Scholten); Department of Neurology, Amstelveen Hospital, Amstelveen, the Netherlands (Dr Bertelsmann); and Department of Neurology, Maastricht University Hospital, Maastricht, the Netherlands (Dr de Krom).
Context Carpal tunnel syndrome (CTS) can be treated with nonsurgical or surgical
options. However, there is no consensus on the most effective method of treatment.
Objective To compare the short-term and long-term efficacy of splinting and surgery
for relieving the symptoms of CTS.
Design, Setting, and Patients A randomized controlled trial conducted from October 1998 to April 2000
at 13 neurological outpatient clinics in the Netherlands. A total of 176 patients
with clinically and electrophysiologically confirmed idiopathic CTS were assigned
to wrist splinting during the night for at least 6 weeks (89 patients) or
open carpal tunnel release (87 patients); 147 patients (84%) completed the
final follow-up assessment 18 months after randomization.
Main Outcome Measures General improvement, number of nights waking up due to symptoms, and
severity of symptoms.
Results In the intention-to-treat analyses, surgery was more effective than
splinting on all outcome measures. The success rates (based on general improvement)
after 3 months were 80% for the surgery group (62/78 patients) vs 54% for
the splinting group (46/86 patients), which is a difference of 26% (95% confidence
interval [CI], 12%-40%; P<.001). After 18 months,
the success rates increased to 90% for the surgery group (61/68 patients)
vs 75% for the splinting group (59/79 patients), which is a difference of
15% (95% CI, 3%-27%; P = .02). However, by that time
41% of patients (32/79) in the splint group had also received the surgery
Conclusion Treatment with open carpal tunnel release surgery resulted in better
outcomes than treatment with wrist splinting for patients with CTS.
Carpal tunnel syndrome (CTS) is the most common entrapment neuropathy
and is caused by compression of the median nerve at the wrist. The prevalence
of electrophysiologically confirmed CTS in the adult general population in
the Netherlands is 0.6% in men and 9.2% in women.1
While CTS may be treated with conservative options such as wrist splints,
injections with corticosteroids, or both, evidence for the efficacy of most
conservative treatment options is limited.2
A recent systematic review has shown that among the various available surgical
techniques, open carpal tunnel release is the preferred method.3
However, only 1 randomized controlled trial (RCT) has compared conservative
treatment (splinting) with surgery,4 but provided
no information on comparability of the groups at baseline, cointerventions,
compliance with the treatment, or blinding of the outcome assessor. Furthermore,
only 1 follow-up assessment was performed after 1 year, showing that the 10
patients treated surgically had complete relief of symptoms while 2 of the
10 patients treated with a wrist splint had experienced relief only temporarily.
Due to limited evidence, there is no consensus on the preferred method
of treatment for CTS. Advocates of surgery refer to its safety and effectiveness
for electrophysiologically confirmed cases with no underlying reversible disorder,5 and point out that conservative treatment options
generally offer only temporary symptom relief. Advocates of conservative options
refer to the potential benefits and safety of these treatments and to the
potential complications of surgery.6 In the
Netherlands, 40% of the neurologists reported to prefer conservative treatment
options, 39% reported to prefer surgery, and 21% reported to have no preference.7 Splinting was the preferred treatment of 26% of the
neurologists and is the most common. The objective of this study was to compare
the short-term and long-term efficacy of splinting and surgery for relieving
the symptoms of CTS.
The medical ethics committees of the 13 participating hospitals approved
the study protocol of this multicenter RCT. A detailed description of the
design of this RCT has been previously published.8
All patients with clinically suspected CTS had been referred to one
of the participating neurologists and were examined for eligibility to participate
in the study. Inclusion criteria were (1) pain, paresthesia, and/or hypoesthesia
in the hand in the area innervated by the median nerve9;
(2) electrophysiological confirmation of the diagnosis8
(median nerve sensory conduction velocity of the index finger ≤41.9 m/s
in patients <55 years or ≤37.3 m/s in patients ≥55 years, or median
nerve distal sensory latency of the index finger ≥3.5 ms; or median-ulnar
distal sensory latency difference of the ring finger >0.4 ms; or median nerve
distal motor latency ≥4.34 ms); (3) age of 18 years or older; and (4) ability
to complete written questionnaires (in Dutch). Exclusion criteria were (1)
previous treatment with splinting or surgery; (2) a history of wrist trauma
(eg, fracture) or surgery; (3) a history suggesting underlying causes of CTS
(eg, diabetes mellitus, pregnancy); (4) clinical signs or symptoms or electrophysiological
findings suggesting conditions that could mimic CTS or interfere with its
validation (eg, cervical radiculopathy, polyneuropathy); and (5) severe thenar
Patients, who were eligible according to their neurologist and were
interested in participation, were referred to one of the research physiotherapists
in the same hospital, who verified that all selection criteria were met. After
they had provided written informed consent, patients were included in the
study. Subsequently, potential prognostic indicators and the baseline values
of the outcome measures were assessed.
Patients were randomly allocated to receive either splinting or surgery
(Figure 1). If bilateral symptoms
were present, the hand with the more severe symptoms (according to the patient)
was treated. By preparing a list for each hospital, the randomization was
stratified by center. Permuted blocks of 4 patients were formed to ensure
near-equal distribution of patients over the 2 treatment groups in each hospital.10 Despite the small block sizes, the potential for
unmasking was considered to be low because the neurologists did not know the
method of randomization and different neurologists selected patients in each
hospital. Even if a neurologist were to know the allocation scheme, the chance
was high that when he/she selected a patient, this patient would not be assigned
the next treatment on the list because a patient selected by another neurologist
has an earlier appointment for the trial. The random sequence of the permuted
blocks was generated by using random number tables. The principal investigator
(A.A.M.G.), who was not involved in the selection and allocation of patients,
prepared, coded, and sealed opaque envelopes containing the treatment allocation.11 After the baseline assessment, the next envelope
was handed to the patient by the research assistant to ensure concealment
Attempts were made to keep the research physiotherapists unaware of
the allocated treatment by encouraging the patients not to reveal any information
regarding their treatment during the examination. Furthermore, before each
examination, the research assistant placed a bandage over the wrist and palm
of all patients to conceal the potential surgical scar. Afterward, the research
physiotherapists were asked to indicate the type of treatment that they thought
the patient had received and to give reasons for this assumption.12
Depending on the usual procedures of the hospital, patients allocated
to splinting received either a custom-made splint (made of soft cast) or a
prefabricated splint (Tricodur, BSN Medical, Hamburg, Germany) that immobilized
the wrist in a neutral position.13 The patients
were instructed to wear the splint during the night for at least 6 weeks,
but could wear it during the day. No other types of treatment were permitted
during the intervention period, except pain medication if necessary. After
6 weeks, the neurologist discussed with the patient whether any further treatment
was necessary, including continued splinting, other conservative treatment
options, or surgery. The decision to undergo surgery could also be made at
a later stage.
Patients allocated to surgery were referred to a general surgeon, neurosurgeon,
plastic surgeon, or orthopedic surgeon, depending on the usual procedures
of the hospital for an outpatient standard open carpal tunnel release. Because
there are waiting lists for surgery in hospitals, efforts were made to make
an appointment within 4 weeks after randomization. Before surgery, no other
types of treatment were permitted, except pain medication. In all surgical
cases, the transverse carpal ligament was released and no concomitant procedures
were performed (eg, flexor tenosynovectomy, internal neurolysis, epineurotomy).
Sutures were removed after 2 weeks. Patients were instructed to perform postoperative
active range-of-motion exercises and encouraged to use the hand as tolerated.
None of the patients received a splint following the surgical procedure. No
specific period off work was recommended.
Patients completed questionnaires and were examined by a trained research
physiotherapist in the hospital at baseline and 3, 6, and 12 months after
randomization. Although different research physiotherapists assessed the outcomes,
most patients were seen by the same therapist each time they visited the hospital.
In the remaining months, and 18 months after randomization, questionnaires
General improvement was scored by the patient on a 6-point ordinal transition
scale, ranging from "completely recovered" to "much worse."12
Treatment success was defined a priori as "completely recovered" or "much
improved." The 2 other outcome measures were number of nights that the patient
awoke due to the symptoms during the past week and severity of the main complaint,
pain, paresthesia, or hypoesthesia at night and during the day during the
past week. Severity of the main complaint, pain, paresthesia, and hypoesthesia
were scored by the patient on an 11-point numerical rating scale (with 0 equaling
"no symptoms" and 10 equaling "very severe symptoms" used as anchors).12
There were 3 secondary outcome measures used. Mean (SD) scores were
collected using the Symptom Severity Scale (11 questions about symptoms experienced
during the past 2 weeks with 1 equaling mildest and 5 equaling most severe)
and the Functional Status Scale (8 items concerning difficulties in performing
various activities of daily living during the past 2 weeks with 1 equaling
no difficulty and 5 equaling cannot perform activity at all).14
After a standardized history-taking and a physical examination, the overall
severity of CTS complaints was scored by a research physiotherapist on an
11-point numerical rating scale with zero equaling no complaints and 10 equaling
very severe complaints.12 Results of nerve
conduction studies after 12 months were also used. At each follow-up assessment,
the patients were asked to record any treatment they had received and any
The groups were primarily compared at 3, 6, and 12 months because all
outcomes were assessed during the visits at the hospital. Furthermore, to
obtain a clear picture of the short-term effects, the follow-up assessment
time of 1 month was chosen. To determine the long-term effects, the assessment
time of 18 months after randomization was chosen. Differences in success rates
between the treatment groups along with 95% confidence intervals were calculated
using χ2 tests. Continuous outcomes were analyzed as change
scores (difference between baseline assessment and each follow-up assessment).
Subsequently, differences in improvement between the groups (mean change score
in surgery group minus mean change score in splint group) along with 95% confidence
intervals were calculated using t tests. Multivariate
analyses (logistic or linear regression) were performed to adjust for the
influence of eventual differences between the groups at baseline in prognostic
indicators (age, sex, duration of current episode of CTS complaints, bilateral
CTS complaints, dominant side more severely affected, previous episodes of
CTS complaints, and patients' preference for splinting or surgery).
All analyses were performed according to the following intention-to-treat
principle: the patients remained in the group to which they were allocated
at baseline.15 Two additional analyses were
performed. In the first analysis, patients who received surgery after splinting
were considered to be not improved. In the second analysis, patients in the
splint group who had received surgery were compared with patients in the same
group who had not. All patients who withdrew from the study were included
in the analysis until the time of withdrawal. For patients who occasionally
missed a follow-up assessment, the available data from completed assessments
Sample-size calculation was based on the ability to detect a clinically
important difference in success rates of 20% or more 3 months after randomization.
A total sample size of 190 patients was required (2-sided α = .05, β
= .20). Values of P<.05 were considered statistically
significant. Data were analyzed using SPSS statistical software (Version 10.1;
SPSS Inc, Chicago Ill).
During a period of 18 months (October 1998 to April 2000), 326 patients
were examined for eligibility by the neurologists. There were 111 patients
who were either ineligible or not interested in participation (Figure 1). The remaining 215 patients were referred to one of the
research physiotherapists who verified that all selection criteria were met.
Of these patients, 39 were not randomized for various reasons (Figure 1), and the remaining 176 were allocated to either splinting
(89 patients) or surgery (87 patients). By October 2001, 147 patients had
completed the follow-up assessment 18 months after randomization, resulting
in a final follow-up rate of 84%. Reasons for withdrawal are presented in Figure 1. In both of the groups, some patients
occasionally missed a follow-up assessment, but some of them completed the
questionnaire at home.
Table 1 shows the frequency
of potential prognostic indicators and the baseline values of the outcome
assessments for each group. There were small differences between the groups
with regard to sex, duration of current episode of CTS complaints, and patients'
preference for treatment.
All 89 patients allocated to splinting received a splint, either custom-made
(28 patients [31%]) or prefabricated (61 patients [68%]), within a median
period of 2 days after randomization (interquartile range, 0-5 days). However,
13 patients (15%) did not receive the treatment according to protocol: 2 received
additional treatment, 1 received physiotherapy, 1 received Mensendieck (exercise)
therapy, and 11 did not wear the splint every night for at least 6 weeks.
Twenty-four patients (27%) indicated that they had also worn the splint during
the day. Eleven patients (12%) used pain medication during the intervention
Of the 87 patients allocated to surgery, 73 (84%) actually underwent
this treatment within a median period of 35 days (interquartile range, 20-55
days) after randomization. Of the patients who completed follow-up assessments,
36% had undergone surgery after 1 month (29/80), 80% after 3 months (62/78),
and 92% after 6 (71/77), 12 (67/73), and 18 (63/68) months. Of the 14 patients
who did not undergo carpal tunnel surgery, 9 did not receive any treatment
and 5 were treated differently. Two patients received physiotherapy, 2 received
pain medication, and 1 received a splint. Seven patients (8%) used pain medication
In the splint group, 74 (83%) of 89 patients continued to wear the splint
after the intervention period for a varying period of time. Furthermore, 52
(58%) of 89 patients received 1 or more additional treatment options after
6 weeks. Twenty-nine patients received pain medication, 4 received physiotherapy,
1 received manual therapy, 2 received Mensendieck (exercise) therapy, 1 received
occupational therapy, 5 received 1 or 2 local corticosteroid injections, and
35 received surgery. Of the patients who had completed follow-up assessment,
7% (6/86) received surgery after 3 months, 31% (26/84) after 6 months, 39%
(32/83) after 12 months, and 41% (32/79) after 18 months.
In the surgery group, 26 (30%) of 87 patients received 1 or more additional
treatment options after surgery. Twenty-five patients received pain medication
primarily to relieve pain caused by the operation, 2 received physiotherapy,
1 received occupational therapy, and 1 received a local corticosteroid injection
and surgery to relieve pain caused by reflex sympathetic dystrophy.
Comparisons between the univariate and multivariate analyses showed
that adjustment for potential prognostic indicators and the baseline values
of the outcome measures minimally influenced the results. Therefore, only
the unadjusted analyses are presented.
After 1 month, more patients in the splint group had improved than in
the surgery group, but more patients in the surgery group improved after 3,
6, 12, and 18 months (Table 2
and Figure 2).
The results of the other outcome measure assessments are also shown
in Table 2 and Table 3. Scores for pain and hypoesthesia are not presented because
the findings were similar to those for paresthesia. After 1 month, the differences
in the primary outcome measures were mainly in favor of splinting. However,
after 3, 6, 12, and 18 months, surgery was found to be more effective than
splinting and with regard to the secondary outcomes. There was no difference
in outcomes between patients in the splint group using a custom-made splint
and those using a prefabricated splint.
An additional analysis was performed for the outcome measure success
rate, using last observation carried forward for the patients who withdrew
from the study. When this was done for the 10 patients who withdrew from the
splint group, the success rates for this group were 52% (46/89) after 3 months,
66% (59/89) after 6 months, 71% (63/89) after 12 months, and 72% (64/89) after
18 months. Four patients in the surgery group did not have any of the follow-up
assessments taken and were regarded as not improved. In addition, in the analysis
using the last observation carried forward for the other 15 patients who withdrew
from the surgery group, the success rates were 26% (23/87) after 1 month,
74% (64/87) after 3 months, 87% (76/87) after 6 months, 86% (75/87) after
12 months, and 85% (74/87) after 18 months.
Withdrawal was not related to general improvement. Of the 10 patients
who withdrew from the splint group, 5 indicated to be much improved at the
time of withdrawal, 2 indicated to be slightly improved, 1 indicated that
no change had occurred, 1 indicated to be slightly worse, and 1 indicated
to be much worse. Of the 19 patients who withdrew from the surgery group,
4 did not have any follow-up assessments, 10 indicated to be completely recovered
at the time of withdrawal, 3 indicated to be much improved, 1 indicated to
be slightly improved, and 1 indicated no change.
In the first additional analysis, patients who received surgery after
splinting were considered to be "not improved." After 18 months, only 29 of
79 patients indicated to be improved by splinting alone, resulting in a final
success rate of 37% for the splint group.
In the second additional analysis, patients in the splint group who
had received surgery were compared with patients in the same group who had
not. After 18 months, the success rate in the group of patients who received
surgery after splinting was 94% (30/32 patients) compared with 62% (29/47
patients) in the group who did not undergo surgical treatment. Also with regard
to the other primary outcomes, the patients in the splint group who received
surgery showed statistically significant more improvement after 18 months
than the patients who did not have surgery.
Table 4 shows the frequency
of adverse effects reported by the patients in each group during the 18 months
of follow-up. Although many patients reported adverse effects, most of these
were relatively mild and of short duration. However, 1 patient in the surgery
group developed reflex sympathetic dystrophy.
The physiotherapists correctly indicated the treatment received 85%
(139/164) of the time after 3 months, 71% (114/161) after 6 months, and 65%
(101/156) after 12 months. Many patients inadvertently mentioned their treatment,
and guesses were often based on the course of symptoms during the follow-up
period and on the way patients reacted to the physical examinations.
This RCT compared the short-term and long-term efficacy of surgery and
splinting for relieving the symptoms of CTS. Although only 176 patients were
included, instead of the 190 originally planned, this did not reduce the power
of the trial because the differences between the groups were large enough
to be detected. Furthermore, although many outcomes were analyzed at different
follow-up times, the results are not likely to be due to chance, as they all
are in the same direction.
The results of the intention-to-treat analyses showed that surgery was
more effective than splinting after 3, 6, 12, and 18 months. These differences
became smaller after 12 months, probably because a large proportion of patients
in the splint group received additional surgery. The differences after 1 month
were mainly in favor of splinting, which might have resulted from patients
allocated to splinting having started their treatment almost immediately after
randomization, in contrast to patients in the surgery group. Randomization
was chosen as the reference point for the timing of the follow-up assessments,
and not the actual start of the treatment because including the time waiting
to undergo surgery reflects current clinical practice. However, as the median
time to undergo surgery was 35 days, taking the actual start of the treatment
as a reference point would result in a shift to the left (of 1 month) of the
line of the surgery group in Figure 2.
We believe the results of this trial are applicable to most patients
with clinically and electrophysiologically confirmed idiopathic CTS. However,
the least severe and the most severe cases were probably not included in this
trial. These patients (or physicians) typically have a strong preference for
splinting or surgery.
This trial was mainly based on subjective, patient-reported outcomes
because the perspective of the patient was considered to be the most important.
As blinding of the patients for the allocated treatment was not possible,
the results could be biased because patients tend to report greater effects
from their preferred method of treatment.16
However, subgroup analyses showed that treatment effects did not depend on
the patients' preference prior to randomization (data not shown). An attempt
was made to include a more objective evaluation of outcome by involving blinded
research physiotherapists who scored the overall severity of CTS complaints.
However, blinding of the research physiotherapists was not successful in most
Treatment success was defined a priori as "completely recovered" or
"much improved." When analyzing the data, it was found that most of the successes
in the surgery group were "completely recovered," while patients in the splint
group were only "much improved." Thus, if success was solely defined as "completely
recovered," the differences between the groups would have been even more pronounced.
The American Academy of Neurology recommends treatment of CTS with noninvasive
options (eg, wrist splints) first, and open carpal tunnel release only if
noninvasive treatment proves to be ineffective.17
This RCT showed that treatment of CTS with surgery results in better outcomes.
A splint might be used while a patient waits for surgery because the waiting
period for open carpal tunnel release in practice is often longer than in
this study, as efforts were made to make an appointment for the patients within
4 weeks after randomization. Patients not willing to undergo surgery could
also be offered a splint. Another recently conducted RCT found that patients
wearing a wrist splint showed more relief from symptoms than patients not
receiving any treatment.18