Well-designed studies that have focused on the diagnosis of CTS are uncommon. The deficiencies in the design of studies involving clinical and complementary tests for CTS are associated with a super-estimation of the performance of these diagnostic tests and the lack of a universally accepted reference standard for diagnosing CTS [
6,
8‐
10]. Most CTS diagnostic accuracy studies were unlikely to report results that are applicable in clinical practice [
1,
27]. The current study design followed an algorithm based on evaluating an actual clinical practice routine for diagnosing CTS: from the consecutive and random eligibility of patients based on a high probability of a clinical diagnosis of CTS, confirmed (or not) by US and NCS—performed such that the examiners were unaware of the clinical condition of the patients and the results of the concurrent complementary examination treatment follow-up and its clinical outcomes. To ensure good diagnostic accuracy it is important that the estimates provided by a diagnostic test reflect its good performance in clinical practice [
1]. The remission of paresthesia (reference standard) after surgical treatment was used in this study to evaluate the effectiveness and performance of the diagnosis of CTS by the US and NCS [
3]. The classical open approach is considered a definitive surgical treatment for CTS with good results in 75 to 90% of patients [
23]. In this study, 90.4% of the 115 operated patients obtained remission of paresthesia with four months pos operative, while in another study 78.9% of the 113 operated patients obtained remission of paresthesia with ten years pos operative by same surgical technique. [
28]. As for the 9.6% of operated patients who did not obtain remission of paresthesia with surgical treatment in our study, it can be attributed to proximal compressions of the median nerve, double cervical compression syndrome, cervical radiculopathy or even to an advanced degree of CTS [
22]. The evaluation of the patients who did not obtain remission of paresthesia after 4 months of the surgical treatment was based on clinical criteria, using the parameters to the CTS-6 and secondly BCTQ [
1,
5,
24]. Though the false-negative patients had normal (absent) results for STC by US and NCS in this study, they were submitted to surgical treatment, supported by the results obtained by the clinical diagnosis for CTS in the initial clinical evaluation. Patients with normal results (absent) for CTS by the US showed better performance in the remission of paresthesia after surgical treatment, as shown in the Table
1. To obtain good quality of evidence in this accuracy study, the degree of CTS impairment provided by US and NCS was discarded and reduced to a simple dichotomous (present / absent, abnormal / normal) diagnosis to make these tests useful in clinical practice of routine [
8,
9,
11]. The parameters and cut-off points considered in this study for the diagnosis of CTS by the US and NCS had a direct impact on the primary and secondary outcomes, producing the four possible types of results of a diagnostic test, showing balanced results for the values of accuracy, especially between sensitivity and specificity. The positive results obtained by the kappa index in the evaluation of NCS and US concordance in relation to the reference standard in this study expressed the reliability of these tests for the diagnosis of CTS, with NCS showing a better performance. Mondelli et al. evaluated the diagnostic usefulness of US and NCS in patients with a clinical diagnosis of CTS and obtained positive kappa values with good concordance [
19]. The confirmation of the clinical diagnosis of CTS in 90.4% of patients, with a total of 84.3% for NCS and 78.3% for the US—using remission of paresthesia as the reference standard—validated the CTS-6 clinical diagnostic probability instrument used during the initial clinical evaluation. Fowler et al. compared US and NCS using CTS-6 as a reference standard and obtained results relative to confirmation of the clinical diagnosis of CTS in 65% of patients, relative to US diagnosis in 61%, and relative to NCS diagnosis in 65% [
29].
The sensitivity, specificity, and positive predictive value of the US and NCS obtained in the current study were similar, showing that both diagnostic methods were effective for diagnosing CTS, with good sensitivity. Well-designed diagnostic accuracy studies with STARD-compliant methodology obtained similar results when comparing US and NCS [
15,
29,
30]. Because of the low negative predictive values, however, neither US nor NCS could adequately rule out the clinical suspicion of CTS in this study when compared with other studies with different reference standards [
3,
29‐
32].
The main limitation of this study was that we did not consider the degree of severity of the initial clinical and complementary diagnosis of CTS. The two-point discrimination test and atrophy of the thenar musculature used in the CTS-6 model are directly linked to the severity of CTS [
34]. A value of > 13 mm2 for the median nerve area found during US would correspond to a moderate degree of impairment [
35,
36]. Absence of a sensory response and abnormal distal motor latency during NCS would correspond to a severe degree of impairment [
3,
19].