Predictors of the patient-centered outcomes of surgical carpal tunnel release – a prospective cohort study

Between August 1^st and December 31^st, 2013, participants were recruited at the neurosurgical outpatients’ clinic Dr. med. Dr. PH Michael A. Conzen and partners, Bielefeld. Patients were eligible if they had pathological nerve conduction difficulties in the median nerve and sufficient communication skills in the German language to fill in the questionnaires. Patients with neurological signs of widespread peripheral neuropathy, attendant arthrosis, tendon pathologies, or inflammatory diseases were excluded from the study. Patients underwent an open carpal tunnel release without tourniquet under local anesthesia by two neurosurgeons according to the standard procedures of the outpatient clinic. Because of the scarcity of such data from Germany, the study was primarily explorative. The targeted sample size was 100 patients to allow for estimation of prevalence at a precision of 10 %. Due to difficulties in recruitment, this number was not reached. However, we think that our study is of sufficient size to describe the main tendencies.

Patients answered questions about socio-demographic and psychological factors as well as medical history, well-being, quality of life, and physical stress due to work (via a modified version of the Latko scale [17]). Quality of life had been measured using the FLZ^M questionnaire [18], which expresses life satisfaction as a score ranging from −96 to +120. Patients assessed the severity of their symptoms at baseline (before surgery), at the three-month, and at the six-month follow-up appointment using the Boston Questionnaire (BQ) (slightly modified version, translated to German) [19]. At each visit, they also assessed the amount of their pain on a numeric rating scale of pain (NRS-P, 1 = “no pain at all”, 10 = “worst pain one could imagine” [20]) and their perceived strength in the affected hand on a numeric rating scale of perceived strength (NRS-PS, 1 = “no strength at all, paralyzed”, 10 = “full strength”). Every patient underwent a standardized clinical examination and nerve conduction measurement at all three visits. The clinical examination included a visual inspection for presence of muscular atrophy in the thenar area, functional muscle tests of the median-innervated hand muscles, and tests of sensibility using the WEST neurofilament test [21]. During the examination, the temperature of the hand was kept above 30 ° C. Median and ulnar nerves were examined bilaterally. The standard nerve conduction measurement used at the clinic included measurements of distal motor latencies (DML) to the M. abductor pollicis brevis and orthodromic or antidromic distal sensory latencies (DSL) (wrist – F2/F5) (ulnar and median). Nerve conduction velocity (NCV) results were graded according to the severity scale proposed by Bland [22] (Table 1). Severity was then re-categorized into three categories to divide the patients into equally sized groups: “normal/mild” (severity less than two), “moderate” (severity equals two), and “severe” (severity greater than two).

Socio-demographic characteristics and CTS-related variables were described as frequencies and mean or median values for the total sample as well as the severity levels of CTS. Analyses were focused on NCV and five patient-centered outcomes (BQ score of symptoms, BQ score of function, NRS-PS, WHO-Five well-being index, and subjective quality of life). The improvements of NCV were described as the percentage of patients who had pathological values at baseline and maintained similar pathological values six months after surgery. The association between NCV measured before surgery and six months later was further investigated using a spline model as implemented in R library “gam” (excluding those who had a complete block at any time point). The timeline of improvement for the NCV and the patient-centered outcomes between baseline, three, and six months after surgery was investigated by testing if there was a significant difference between the mean values at the different visits. Linear regression analysis was used to investigate if socio-demographic characteristics, the WHO-Five well-being index (as a measure of depressive symptoms), and CTS-related variables predict the success of the surgery with respect to NCV and patient-centered outcomes. A linear regression model was built for every combination of predictors and outcomes (value at six months after surgery). The change between the two time points is related to the baseline value, a phenomenon commonly known as regression to the mean. Therefore, we adjusted for the baseline value of the outcome in each model (analysis of covariance (ANCOVA) [23]). In addition to the regression coefficients and the 95 % confidence intervals (CI), the effect size was calculated as a measure of how much variation in the given outcome at six months could be explained by the given predictor. The calculation of effect sizes was based on partial eta squared [24]. A partial eta squared equal to or greater than 0.01 presents a small effect, equal to or greater than 0.06 presents a medium effect, and equal to or greater than 0.14 presents a strong effect [24]. Finally, a Cox regression model was used to assess characteristics associated with the duration of sick leave.

There were 71 cases of CTS surgery in the study sample (three patients had had surgery on both hands) (Table 2). More women had had surgery compared to men (63.4 % vs. 36.6 %, respectively; p = 0.03).

The distribution of sex and the median BMI differed among the three categories of severity (Table 2). There were no differences among these categories regarding either other socio-demographic variables (Table 2) or clinical characteristics associated with CTS at baseline (Table 3).

At baseline, 78.0 % of the patients had moderate to extremely severe scores of severity (based on NCV). Six months after surgery, this number had dropped to 32.2 %. The mean NCV had improved significantly post-surgery, and the association between the pre-surgical value and the value six months after surgery indicated a proportional improvement with relation to the baseline measurement (Fig. 1 – while a more complex model was allowed by the statistical procedure, the linear model was supported by the data).

Surgery generally improved the studied outcomes. Improvements in NCV, BQ scores, and WHO-Five well-being index occurred mainly in the first three months, with little progress observed thereafter. Improvements in perceived strength in the affected hand and subjective quality of life continued beyond three months (Fig. 2).

Most variables did not have a strong predictive effect on improvements of CTS in the univariate analyses (Table 4). The numbers in Table 4 are interpreted as follows: The betas (linear regression coefficients) represent the expected change in the outcome for a one-unit change in the predictor. In the case of categorical predictors, the betas represent the expected change in the outcome for switching from the reference category to the reported category. The partial η² represents the percentage of the variation in the outcome explained by the predictor.

Interestingly, the presence of muscular atrophy in the thenar area at baseline displayed medium size effects for multiple outcomes of CTS surgery: Patients suffering from muscular atrophy in the thenar area scored 0.35 (95 % CI [0.04, 0.65]) points higher in the BQ score of symptoms than patients without muscular atrophy, −12.98 [−23.1, −2.85] points lower in the WHO-Five well-being index, and −26.36 [−47.31, −5.42] points lower in the score of subjective quality of life. The effect size in the BQ score of function was also medium, but the regression coefficient was not statistically significant.

Similarly, there were medium to large effects of family history on DSL (−0.91 [−1.77, −0.05]), NRS-PS (1.49 [0.34, 2.63]), and the WHO-Five well-being index (−14.2 [−23.4, −5]). The WHO-Five well-being index itself as a measure of depressive symptoms at baseline played a rather marginal role with respect to surgery outcomes, with the exception of the functional BQ score. Age and sex had medium effects on NRS-PS.

The analysis of the predictors for sick leave was only possible for 42 cases providing this information. None of the univariable associations were significant, but there was some indication of variables which should become significant in a larger sample: type of work, severity of CTS, and whether the surgery affected the dominant hand (Table 5).

The strengths of the study lie in its prospective design with a standardized clinical and electrophysiological examination. The study reports experience from a single outpatient clinic performing CTS surgery. While we are not aware of specific differences, the results might not be generalizable to other settings. There was also a substantial drop out rate among participants between the baseline and the follow-up at three or six months. While this does not affect the observed effects on the individual level, the extrapolation of the findings to all CTS patients might be restricted.

The study was approved by the Regional Ethical Review Board, University of Münster. All patients provided written informed consent before entering the study.

The dataset supporting the conclusions of this article is available from the corresponding author.