Diagnostic Test Accuracy of Provocative Maneuvers for the Diagnosis of Carpal Tunnel Syndrome: A Systematic Review and Meta-Analysis

Abstract

Objective

The purpose of this study was to summarize and evaluate the research on the accuracy of provocative maneuvers to diagnose carpal tunnel syndrome (CTS).

Methods

The MEDLINE, CINAHL, Cochrane, and Embase databases were searched, and studies that assessed the diagnostic accuracy of at least 1 provocative test for CTS were selected. Study characteristics and data about the diagnostic accuracy of the provocative tests for CTS were extracted. A random-effects meta-analysis of the sensitivity (Sn) and specificity (Sp) of the Phalen test and Tinel sign was conducted. The risk of bias (ROB) was rated using the QUADAS-2 tool.

Results

Thirty-one studies that assessed 12 provocative maneuvers were included. The Phalen test and the Tinel sign were the 2 most assessed tests (in 22 and 20 studies, respectively). The ROB was unclear or low in 20 studies, and at least 1 item was rated as having high ROB in 11 studies. Based on a meta-analysis of 7 studies (604 patients), the Phalen test had a pooled Sn of 0.57 (95% CI = 0.44–0.68; range = 0.12–0.92) and a pooled Sp of 0.67 (95% CI = 0.52–0.79; range = 0.30–0.95). For the Tinel sign (7 studies, 748 patients), the pooled Sn was 0.45 (95% CI = 0.34–0.57; range = 0.17–0.97) and the pooled Sp was 0.78 (95% CI = 0.60–0.89; range = 0.40–0.92). Other provocative maneuvers were less frequently studied and had conflicting diagnostic accuracies.

Conclusion

Meta-analyses are imprecise but suggest that the Phalen test has moderate Sn and Sp, whereas the Tinel test has low Sn and high Sp. Clinicians should combine provocative maneuvers with sensorimotor tests, hand diagrams, and diagnostic questionnaires to achieve better overall diagnostic accuracy rather than relying on individual clinical tests.

Impact

Evidence of unclear and high ROB do not support the use of any single provocative maneuver for the diagnosis of CTS. Clinicians should consider a combination of noninvasive clinical diagnostic tests as the first choice for the diagnosis of CTS.

Introduction

Carpal tunnel syndrome (CTS) is caused by the compression of the median nerve in the carpal canal and is the most prevalent compression neuropathy of the upper extremity (UE).¹ The prevalence of CTS has been estimated to be 6% in men and 9.2% in women.¹ Symptoms include pain, tingling, and numbness in the palmar surface of the radial 3.5 digits (the area innervated by the median nerve), as well as thenar muscle weakness and atrophy in more severe instances.² CTS is a significant contributor to days lost from work, with an average of $2 billion in annual costs of medical care.³

Diagnosing CTS is an essential step in the treatment of this condition. CTS can be treated conservatively if diagnosed early, whereas a late diagnosis might limit the treatment options for severe or persistent cases of carpal tunnel release surgery.⁴ Several diagnostic options are currently available, with the most common ones being electrodiagnosis or nerve conduction velocity testing. These tests are often costly, increase the wait time to receiving management, are painful, and maybe not be available at all settings. Clinicians can benefit from using clinical diagnostic tools, such as questionnaires, scales, hand diagrams, provocative tests, and sensorimotor tests, as outlined by the most recent guideline of the American Academy of Orthopedic Surgeons.⁵ Unlike nerve conduction velocity testing, clinical examination tests can be done in all clinical settings and yield immediate results.

A systematic review (SR) of the accuracy of clinical examination tests was done by a member of our research team in 2004 but is currently outdated.⁶ This SR is the last of a series assessing all categories of clinical examination tests for CTS diagnosis. We assessed the accuracy of the questionnaires, scales, hand symptom diagrams/maps,⁷ and sensorimotor tests⁸ in 2 previous SRs. We also assessed the accuracy of clinical diagnostic tests and questionnaires in screening for CTS among workers.⁹ In this SR, we focused on provocative maneuvers for CTS diagnosis, such as the Phalen test, Tinel sign, Durkan test, etc. Therefore, the purpose of the current SR was to identify, critically appraise, and synthesize the evidence on the diagnostic accuracy of provocative tests for diagnosing CTS in people with suspected CTS.

Methods

On December 20, 2018, we registered this SR with the International Prospective Register of Systematic Reviews (PROSPERO), with the registration number CRD42018109031. We followed the Diagnostic Test Accuracy extension of the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) and the Cochrane collaboration guidelines.¹⁰

Information Sources

In consultation with a health sciences librarian at McMaster University, we searched 4 electronic databases on January 20, 2020. We updated our search in February 2022. The electronic databases that we searched were MEDLINE (through Ovid from 1946), Embase, the Cochrane Database of Systematic Reviews, and CINAHL. We designed a search strategy to include studies evaluating the accuracy of any clinical examination tests for CTS (Suppl. Appendix A). We combined keywords and search terms for diagnostic accuracy with clinical diagnostic tests for CTS. The search terms regarding the names of the CTS clinical diagnostic tools were further reviewed for comprehensiveness by 2 authors (A.D. and J.M.), who are a physical therapist and a hand therapist. Finally, we hand-searched the reference lists of the included papers.

Study Selection

Two reviewers (A.D. and J.Y.) independently screened the articles for inclusion in this SR in 2 consecutive phases. In the first phase, we reviewed the titles and abstracts according to a pre-determined set of eligibility criteria. We rated the agreement of the reviewers using κ statistics, which we calculated by STATA statistical analysis software, version 15. κ values <0.20 indicate poor agreement between the reviewers, 0.21 to 0.40 indicate fair agreement, 0.41 to 0.60 indicate moderate agreement, 0.61 to 0.80 indicate substantial agreement. Lastly, values of 0.81 to 1.00 indicate almost perfect agreement.¹¹ In the second phase of study selection, we retrieved the full texts of the articles and reviewed them for inclusion. If disagreements occurred between the 2 reviewers selecting the studies in the first or the second phases, a third research team member (J.M.) resolved the disagreement through discussion.

Eligibility Criteria

We did not exclude articles based on language, sex/gender of the participants, sample size, year of publication, or the type of reference standards. We excluded studies on conditions other than CTS and studies that did not address the diagnostic accuracy of provocative maneuvers. We included studies that met the following criteria.

Design: We included cohort, case–control, or cross-sectional studies with either prospective or retrospective data collections in a full-report format.

Participants: Studies with adult participants (18 years and older) with signs and symptoms of CTS. We excluded studies using participants who were healthy as “controls,” to avoid underestimating accuracy in clinical contexts.

Diagnostic test: Any provocative test for CTS as the diagnostic (or index) test.

Comparison: Studies using a comprehensive clinical examination by an appropriate hand specialist, electrodiagnosis, or a combination of the 2 as a reference standard were included.

Outcome: Studies assessing at least 1 diagnostic accuracy property for CTS diagnoses, such as sensitivity (Sn), specificity (Sp), positive likelihood ratio (+LR), negative likelihood ratio (−LR), and positive predictive values and negative predictive values (PPVs or NPVs).

Time: Any timeframe for the diagnosis of CTS using provocative tests.

Data Extraction

One research team member (E.B.) extracted data from included articles, and 2 authors (A.D. and J.Y.) checked the extracted information. We extracted data using a self-developed, pre-determined data extraction sheet adapted from a previous SR on the accuracy of scales, questionnaires, and hand symptom diagrams/maps for CTS diagnosis. J.M. was contacted if any uncertainty occurred in the data extraction process, and a consensus was established through discussion. We extracted the following information: authors, year of publication, country, conflicts of interest, study design, sample size, number of participants in the CTS and the control groups, characteristics of the participants (gender or sex, age, CTS duration, and severity), inclusion and exclusion criteria, the provocative maneuver, and its testing procedure, as well as its positive threshold criteria, reference standard, and any available information on the diagnostic accuracy properties.

Data Synthesis and Analysis

Where possible, we extracted Sn, Sp, PPV, NPV, +LR, and −LR of the provocative tests for CTS. If these variables were not reported directly in the articles, we calculated them based on the reported true positives and negatives and false positives and negatives. If enough information was reported, we created 2×2 contingency tables and calculated Sn, Sp, +LR, and −LR for each provocative test.

Sn is the ability of a test to identify people with truly positive results, and Sp is defined as the ability of a diagnostic test to truly identify people without a given condition (true negatives).¹² For example, with a Sn of 0.70, a diagnostic test can correctly identify 70% of the persons with a disease and misses 30% of them. With a Sp of 0.70, a diagnostic test correctly identifies 70% of persons who do NOT have a disease and falsely labels 30% of healthy people as having a disease. We calculated Sn as true positive/ (true positive + false negative) and Sp as true negative/ (true negative + false positive).¹²

Likelihood ratios are the only diagnostic properties that are not affected by the prevalence of the condition (pretest probability) and are valuable tools in the clinical decision-making process.¹³ We calculated + LR as Sn/ (1-Sp) and −LR as (1-Sn)/Sp. Positive likelihood ratios > 10 and −LRs < 0.10 indicate the highest diagnostic accuracies.¹⁴ Positive likelihood ratios < 5 and −LRs > 0.5 indicate poor diagnostic accuracy and the limited ability of a diagnostic test to be impactful in clinical decision-making.¹⁴

Lastly, the diagnostic odds ratio is a measure of the possibility of a diagnosis, combining both Sn and Sp into 1 value. The odds ratio can range from 1 to infinity, with 1 equal to 0 diagnostic odds and higher values corresponding to a higher possibility of a given diagnosis.¹⁵

Meta-Analyses

We ran a meta-analysis on studies when data were available (or could be calculated) on the true positive, false positive, true negative, and false negative values of the included participants. We used MetaDTA software (v2.01) to perform the meta-analysis (https://crsu.shinyapps.io/dta_ma/).^16,17 The MetaDTA software uses the R statistical software and runs a bivariate random-effects model, considering the possible correlations between Sn and Sp.¹⁶ For diagnostic test accuracy studies, bivariate models are increasingly recommended as the standard method of meta-analysis, as they are sophisticated enough to capture the complexity of the generally inversed Sn and Sp.¹⁸ Furthermore, we did not perform a heterogeneity analysis, since for meta-analysis of studies of diagnostic test accuracy, a high heterogeneity is always assumed as a rule of thumb. This is due to the variation in the tests’ methodologies and the reference standards and could be influenced by a threshold effect, which leads to generally high heterogeneity.^18,19 To tackle this, we performed a random-effects model, which is the recommended method when a high heterogeneity is assumed.¹⁸

We estimated the pooled Sn, Sp, +LR, −LR, diagnostic odds ratio, and their associated 95% CIs for the Phalen test and the Tinel sign. We did not do a meta-analysis of the other provocative tests for CTS when fewer than 4 studies were available to avoid any bias introduced by the correlation between Sn and Sp and to produce reliable estimates of the diagnostic accuracies.¹⁸

For a graphical display of the findings of the meta-analysis of diagnostic test accuracy studies, 2 methods are recommended by the Cochrane Handbook²⁰ and an extension of PRISMA.¹⁰ We incorporated both methods, which were: (1) Forest plots of the individual study results and meta-analysis pooled results for Sp and Sn separately and (2) Summary Receiver Operating Characteristic (SROC) plots of the global summary of the Phalen test and the Tinel sign. For the SROC plots, we used the quality assessment-enhanced method suggested by Patel et al.¹⁷ This method provides better visual feedback of the QUADAS-2 risk of bias (ROB) appraisal incorporated markers of study ROB into the SROC plots. Each data point on the SROC plot demonstrates the relationship between true positive and false positive rates in a single study.

Assessment of ROB and Applicability Concerns

Two authors (A.D. and J.Y.) independently rated the ROB and the applicability concerns of the included studies. If any disagreements occurred, a third author (the most senior 1, J.M.) was reached, and a consensus was achieved with discussion. We used the Quality Assessment of Diagnostic Accuracy Studies (QUADAS-2) tool to assess ROB and applicability concerns.²¹ QUADAS-2 tool rates the ROB of the studies in 4 domains: (1) Patient selection: Prior testing, presentation, the intended use of index test and setting; (2) index test: How it was conducted and interpreted using a pre-specified threshold; (3) reference standard: Its likeliness to correctly identify the patients, and if the results were interpreted without the knowledge of the index test; (4) flow and timing: If there was an appropriate interval between index and reference test, and if the patients all received the same reference standards and were included in the analysis.²¹ Applicability concerns are rated in the same domains, except for flow and timing.²¹

For each domain, the signaling questions can be rated as low, unclear, or high ROB or applicability concerns.²¹ To generate an overall rating of ROB and applicability concerns, we rated the overall ROB as low if a study had low ROB in all 4 domains. Similarly, if a study had low concerns regarding applicability in all 3 domains, we rated that study as having “low applicability concerns.” On the contrary, studies with high or unclear ratings in any domain (even 1) were rated as having a high ROB or applicability concerns.²¹

Role of the Funding Source

The funders played no role in this study’s design, conduct, and reporting.

Results

Study Selection

We identified 5555 citations by electronic database search. When the duplicates were removed, we screened 4052 titles and abstracts, of which 161 articles were eligible for full-text screening. At last, 31 articles met the inclusion criteria of this SR as presented in the PRISMA flow chart (Suppl. Appendix B). The κ agreement of the reviewers in the screening of titles and abstracts was 0.70 (95% CI = 0.66–0.74; SE = 0.02).

Characteristics of Studies

A thorough description of the study characteristics, including authors, year, study design, country, number of participants and hands with CTS, diagnostic tool, and reference standards, is presented in Table 1. Most of the included studies (n = 28) had prospective cross-sectional study designs, with only 3 having retrospective cross-sectional study designs.^22–24 The studies were conducted in Brazil, Canada, Columbia, Egypt, France, India, Israel, Italy, Serbia, Spain, Turkey, the UK, and the USA. All studies were published in English, except for the study by Amo et al, which was published in Spanish,²⁵ and the study by Yilmaz et al, which was published in Turkish.²⁶ We translated these studies using Google translate and invited native speakers to verify the extracted information.

Table 1.

Study Characteristics^a

Author (Year), Study Design	Country	N-Individuals	N-Hands	N-CTS	Provocative Test	Reference Standard	ROB
Amo et al25 (1998), Prospective cross-sectional	Spain	57	100	NR	Tinel, Phalen	NCS	Unclear
Bilkis et al 27 (2012), Prospective cross-sectional	USA	37	66	46	Modified Phalen, Phalen	NCS	Unclear
Blok et al49 (2013), Prospective cross-sectional	USA	41	NR	31	Scratch-collapse test	NCS	Unclear
Buch-Jaeger & Foucher45 (1994), Prospective cross-sectional	France	112	172	68	Phalen, Tinel, Gilliatt	NCS	Unclear
Dale et al29 (2011), Prospective cross-sectional	USA	1108	2216	35	Tinel, Phalen	Modified Katz Hand Diagram; NCS; The Consensus Criteria CTS Case Definition	High
Denham et al 30 (2015), Prospective cross-sectional	USA	18	30	30	Phalen, ULTT II	Diagnosis of CTS by physical exam, EMG, NCS	High
Franzblau et al 31 (1993), Prospective cross-sectional	USA	130	260		Tinel, Phalen	EDX; EDX+ surveillance symptom definitions for CTS; Physical examination + surveillance symptom definitions for CTS	High
de Jesus Filho et al24 (2014), Retrospective cross-sectional	Brazil	56	70	70	Tinel, Durkan, Phalen	EMG; Ultrasound	High
Hansen et al32 (2004), Prospective cross-sectional	USA	142	NR	95	Flick, Tinel, Phalen	NCS	High
Heller et al33 (1986), Prospective cross-sectional	Israel	60	80	58	Phalen, Tinel	EMG	Unclear
Karl et al53 (2001), Prospective cross-sectional	USA	96	96	51	Lumbrical Provocation test	EDX; Katz hand diagrams	Unclear
Katz et al23 (1990), Retrospective cross-sectional	USA	110	220	44	Tinel, Phalen	NCS	High
Kaul et al52 (2000), Prospective cross-sectional	USA	112	NR	58	Tethered median nerve stress test	NCS	High
Kaul et al47 (2001), Prospective cross-sectional	USA	269	269	58	CCT, PPT	EMG	High
Kuhlman et al34 (1997), Prospective cross-sectional	USA	143	228	142	Phalen, Hoffman-Tinel	NCS	High
Lajoie et al22 (2005), Retrospective cross-sectional	India	81	162	NR	Tinel, Phalen	Latent Class Analysis of provocative test and NCS	Unclear
MacDermid et al35 (1997), Prospective cross-sectional	Canada	42	84	36 hands	Tinel, Phalen, Rev-Phalen, Tethered median nerve stress test	EMG; Clinical examination	Low
Makanji et al50 (2013), Prospective Cohort	USA	88	NR	65	Scratch-collapse test	EMG & NCS	Unclear
Martic46 (2015), Prospective cross-sectional	Serbia	181	NR	60	History, clinical diagnosis, reflex activity, impaired sensation; Combination of muscle strength and trophic changes, reflex activity, impaired sensation, Bickeles, Tinel, and Flick	EMG	Unclear
Naranjo et al36 (2007), Prospective cross-sectional	Spain	68	105	80 hands	Tinel, Phalen	US; NCS; Clinical history	Unclear
O’Gradaigh & Merry37 (2000), Prospective cross-sectional	UK	105	NR	75	Phalen, Tinel	NCS	Unclear
Raudino38 (2000), Prospective cross-sectional	Italy	83	140	140	Tinel, Phalen, Median nerve Stress test	EMG	Unclear
Sawaya & Sakr39 (2009), Prospective cross-sectional	Lebanon	27	54	27 hands	Phalen	EMG & NCS	Unclear
Simon et al51 (2017), Prospective cross-sectional	USA	40	40	32	Scratch-collapse test	EMG	Unclear

(Continued)

Supplementary Appendix C summarizes the included studies’ conflicts of interest (if declared). Participants’ characteristics, including their age, sex/gender, duration of CTS symptoms, inclusion and exclusion criteria, are presented in Supplementary Appendix D.

We identified the following provocative maneuvers for CTS diagnosis from the included clinical studies, in order of frequency: Phalen test (22 studies, 2906 patients), Tinel sign (20 studies, 3005 patients), Durkan test or carpal compression test (CCT, 5 studies, 609 patients), scratch collapse test (SCT, 3 studies, 167 patients), tethered median nerve stress test (3 studies, 237 patients), upper limb tension tests I & II (ULTT, 2 studies, 136 patients), Flick sign (1 study, 142 patients), Gilliatt test (1 study, 112 patients), lumbrical provocation test (1 study, 96 patients), modified Phalen test (MPT, 1 study, 37 patients), pressure provocative test (1 study, 269 patients), and reverse Phalen test (1 study, 42 patients). Table 2 describes the methodology and positive test criteria for these provocative maneuvers.

Table 2.

Description of Provocative Maneuvers for Carpal Tunnel Syndrome Diagnosis

Diagnostic Test	Method	Positive Result Threshold
Carpal Compression test47,48 or Durkan test24,40,43	“Applying moderate pressure with both thumbs over the transverse carpal ligament. The patient’s wrist was in the neutral position with the forearm supinated.”47 “The subject’s hand was comfortably positioned on a table with the palm facing up. The footplate of the gauge was placed on the volar surface of the wrist, directly over the transverse carpal ligament, and 12–15 psi of pressure was applied for a 30-s period.”48 “Durkan test was performed using a calibrated piston (Gorge Medical, Hood River, OR) to apply 9 to 10 N/cm2 to the palm over the transverse carpal ligament.”40	If median-distribution paresthesia was induced within 30 s47 Reproduction of CTS symptoms within 30 s48
Flick sign32	“By asking the patient what they do with their hands and wrists when symptoms are most severe.”32	If the patient demonstrates a flicking motion of the hands and wrists (like shaking down a thermometer)
Gilliatt test28	“Was performed by inflating a blood pressure cuff around the arm beyond systemic pressure for 1 minute.”28	If this caused paresthesia in the median nerve territory
Lumbrical Provocation Test53	“Each patient was asked to make a fist for 1 minute”53	If median-inclusive paresthesia was induced or exacerbated
Modified Phalen test27	“The MPT places the hands as in Phalen test while performing sensory testing with a Semmes-Weinstein 2.83-unit monofilament applied perpendicular to the skin surface until it bends. The monofilament was applied to the palmar surface and lateral side of each finger’s distal phalanx 3 times.”27	If the participant did not register the touch in any 1 or more fingers in a median nerve distribution
Phalen test22–43	“The testing position for the Phalen test required participants to sit with a relaxed posture, with shoulders flexed to approximately 70° and elbows flexed to approximately 90°. Participants were then asked to actively assume a position of extreme but not forced wrist flexion for one minute.”30 “Phalen test was performed as originally described. Subjects were asked to hold their forearms upright, resting their elbows on the examination table and letting their wrists drop into flexion with gravity assistance”40	The onset or increase in numbness or paresthesia in the median nerve distribution within 60 s of holding the wrists in a position of fixed flexion40
Pressure Provocative test47	“A pressure cuff 1 inch wide was applied to the patient’s wrist. The patient’s arm was placed on a table in the supinated forearm position, wrist in neutral. The cuff was inflated to 50 mmHg. Direct pressure was then applied on the cuff over the median nerve to bring the sphygmomanometer reading to 150 mmHg. This pressure was sustained for 30 s.”47	If median-distribution paresthesia was induced within 30 s and symptom relief when the pressure was removed
Reverse Phalen test35	“Reverse Phalen was performed by placing the palms together and maintaining a position of full passive extension.”35	If CTS symptoms reproduced after holding the position for 60 s
Scratch-Collapse test49–51	“The examiner applies bilateral shoulder adduction / internal rotation to the forearms and the patient tries to resist the examiner’s force; Next, the examiner ‘scratches or swipes with the fingertips over the area of the transverse carpal ligament, where the median nerve is potentially compressed; and Immediately afterwards Step 1 is repeated.”49	If momentary loss of the patient’s external resistance tone is observed, in other words the patient’s arm collapses against the resistance
Tinel sign22–26,28,29,31–33,35–38,  40–42,46 or Hoffman-Tinel34	“Tinel test was performed by a reflex hammer to percuss over the palm from the proximal palmar crease to the distal wrist.”35	If percussion over the median nerve at the wrist produced paresthesia in the median nerve distribution40
Tethered median nerve stress test35,38,62	“The median nerve was placed on stretch by passively hyperextending the supinated wrist and distal interphalangeal joint of the index finger.”38	If proximal volar forearm pain was produced38,62
Upper Limb Tension Test I & IIa30,54	ULTT IIa: “The subjects were asked to lay supine and relaxed. The therapist first depressed the shoulder girdle, while simultaneously abducting the humerus to approximately 10°, fully extending the elbow, laterally rotating the humerus, and extending the wrist, fingers, and thumb. The position was held for one minute, and participants were then asked to describe any sensory changes in the tested limb.”30 ULTT I: “Participants lay supine without a pillow, arms along the body, and legs straight. Tests were carried out slowly, and participants were instructed to indicate the point at which it was too uncomfortable to continue with the movement (point of pain tolerance). Angle measurements were then taken at this point. The total time for each test was always under 1 minute, but there was no standardization of the movement time. Once the test was ended, the location and nature of the sensory response was marked on a body chart. The starting position for ULNT1 was 90 abduction and 90 external rotations of the shoulder, 90 elbow flexion, forearms supination, maximum extension of wrist and fingers, and abduction of the thumb. One of the physiotherapist’s hands was placed on the scapula to prevent elevation, the other hand-maintained finger abduction. The elbow was then slowly extended until the point of pain tolerance, and the elbow angle was measured.”54	If the participant reported pain, numbness, or tingling along the median nerve distribution.30 “Was graded according to Wainner criteria, and symptoms were located as proposed by Lohkamp and Small.”54

Risk of Bias and Applicability Concerns

Overall, the ROB and applicability concerns of the included studies were low or unclear in 20 of the studies. Eleven studies were at high ROB or applicability concerns in at least 1 domain. The “flow and timing” domain of the ROB assessment was unclear in 18 studies. Furthermore, the “reference standard” domain was rated as having a ROB in 23 studies, as studies did not consistently report the reference standards test procedures.

For applicability concerns, the domain with the most “applicability concerns” rating (n = 7) was the patient selection domain. Most of the studies (n = 19) had no concerns regarding applicability in the “patient selection” and “reference standard” domains. The ROB and applicability concerns are presented in Figure 1.

Figure 1.

Risk of bias and applicability concerns

Table 1.

Continued

Author (Year), Study Design	Country	N-Individuals	N-Hands	N-CTS	Provocative Test	Reference Standard	ROB
Szabo et al40 (1999), Prospective cross-sectional	USA	CTS = 87, other diagnoses of UE = 90	100	50 (87 hands)	Phalen, Tinel, Durkan	NCS	Unclear
Trillos et al54 (2016), Prospective cross-sectional	Colombia	118	230	186	ULTT I	NCS	Low
Wainner et al48 (2000), Prospective cross-sectional	USA	52	55	25	CCT with Durkan gauge	NCS	High
Walters & Rice41 (2002), Prospective cross-sectional	USA	55	77	NR	Phalen, Tinel	NCS	High
Yilmaz et al26 (2002), Prospective cross-sectional	Turkey	188	346	NR	Phalen, Tinel	EMG	Unclear
Zaher et al42 (2012), Prospective cross-sectional	Egypt	52	52	52	Phalen, Tinel	EDX; MRI; US	Unclear
Zhang et al43 (2020), Prospective cross-sectional	USA	55	85	76	Phalen, Tinel, Durkan	EDX	Low

^a CCT = carpal compression test; EDS = electrodiagnosis; EMG = electromyography; HSD = hand symptom diagram; MPT = modified Phalen test; MRI = magnetic resonance imaging; NCS = nerve conduction studies; PPT = pressure provocative test; ULLT = upper limb tension test; US = ultrasound.

Diagnostic Accuracy of the Phalen Test and Variations for CTS Diagnosis

The diagnostic accuracy of the Phalen test was assessed in 22 studies^22–43 with 2906 patients (Suppl. Appendix E). The methodology was consistent across the included studies and performed according to the original method described by Phalen, which is explained in Table 2.⁴⁴ The Sn ranged from 0.12²⁹ to 0.92,²² and the Sp ranged from 0.30³⁶ to 0.95.²⁵ The meta-analysis of 7 studies,^{27,32–34,41,43,45} (604 patients), revealed pooled Sn of 0.57 (95% CI = 0.44–0.68), pooled Sp of 0.67 (95% CI = 0.52–0.79), pooled +LR of 1.71 (95% CI = 1.18–2.50), pooled −LR of 0.65 (95% CI = 0.50–0.84), and pooled diagnostic odds ratio of 2.65 (95% CI = 1.56–4.80). The SROC and forest plots of the Phalen test are presented in Figure 2. Only 1 study assessed the diagnostic accuracy of the Reverse Phalen test, which demonstrated good Sn (0.69) and excellent Sp (0.96).³⁵

Figure 2.

Phalen test meta-analyses summary receiver operating characteristic (SROC) and forest plots

Diagnostic Accuracy of the Tinel Sign for CTS Diagnosis

The Tinel sign was assessed in 20 studies with 3005 patients.^{22–26,28,29,31–33,35–38,40–42,46} One study referred to this test as “Hoffman-Tinel”,³⁴ but the methodology was consistent across all the studies, as summarized in Table 2. The Sn ranged from 0.17²⁹ to 0.97,²² and the Sp ranged from 0.40³⁶ to 0.92.³⁵ The meta-analysis of 7 studies,^{32–34,41,43,45,46} (748 patients), indicated pooled Sn of 0.45 (95% CI = 0.34–0.57), pooled Sp of 0.78 (95% CI = 0.60–0.89), pooled +LR of 2.10 (95% CI = 1.16–3.83), pooled −LR of 0.67 (95% CI = 0.56–0.76), and pooled diagnostic odds ratio of 3.01 (1.40–6.47). Supplementary Appendix F summarizes the diagnostic accuracies for the Tinel sign, and Figure 3 depicts the SROC and the forest plots of Sn and Sp of this test.

Figure 3.

Tinel sign meta-analyses summary receiver operating characteristic (SROC) and forest plots

Diagnostic Accuracy of the Durkan Test for CTS Diagnosis

The diagnostic accuracy of the Durkan test (also known as the CCT or McMurtry sign or median nerve compression test) was assessed in 5 studies. Two of the 5 studies called this CCT^47,48 and had the same test methodology as the Durkan test, where the pressure on the median nerve was applied at the transverse carpal ligament, either by the therapist’s thumbs, a gauge, or a calibrated piston.^24,40,43 In all cases, the test was positive if symptoms were provoked in the median innervated territory within 30 seconds.^24,40,43 This is further described in Table 2. For the Durkan test, Sn ranged from 0.36⁴⁸ to 0.86,²⁴ Sp ranged from 0.33⁴³ to 0.62,⁴⁷ +LR ranged from 0.83⁴⁸ to 1.62,⁴⁰ and −LR ranged from 0.24⁴⁰ to 1.12.⁴⁸ Diagnostic accuracy of the Durkan test is summarized in Table 3.

Table 3.

Diagnostic Accuracy of Other Provocative Tests for CTS Diagnosis^a

Study  [Author (Year)]	Sensitivity	Specificity	PPV	NPV	+LR	−LR	ROB
Durkan test
de Jesus Filho et al24 (2014)	0.86	NR	NR	NR	NR	NR	High
Kaul et al47 (2001) (CCT)	0.53	0.62	0.67	0.47	1.39b	0.76b	High
Szabo et al40 (1999)	0.89	0.45	0.22	0.96	1.62b	0.24b	Unclear
Wainner et al48 (2000) (CCT)	0.36	0.57	0.84	1.12	0.83b	1.12b	High
Zhang et al43 (2020)	0.78	0.33	0.90	0.15	1.16b	0.66b	Low
Flick test
Hansen et al32 (2004)	0.37	0.74	0.74	0.37	1.42b	0.85b	High
Gilliatt test
Buch-Jaeger & Foucher45 (1994)	0.52	0.36	0.53	0.52	0.81b	1.33b	Unclear
Lumbrical Provocation test
Karl et al53 (2001)	0.37	0.71	0.59	0.50	1.27b	0.88b	Unclear
Pressure Provocative test
Kaul et al47 (2001)	0.55	0.68	0.70	0.53	1.71b	0.66b	High
Scratch Collapse test
Blok et al49 (2013)	0.32	NR	NR	NR	NR	NR	Unclear
Makanji et al50 (2013)	0.31	0.61	0.71	0.25	0.79b	1.13b	Unclear
Simon et al51 (2017)	0.24	0.62	0.75	0.18	0.75b	1.22b	Unclear
Tethered Median Nerve Stress test
Kaul et al52 (2000)	0.50	0.59	0.62	0.47	1.21b	0.85b	High
MacDermid et al35 (1997)	0.52	0.92	NR	NR	6.50b	0.52b	Low
Raudino38 (2000)	0.43	NR	NR	NR	NR	NR	Unclear
Upper Limb Tension test
Denham et al30 (2015)	0.73	0.50	0.98	0.06	1.46b	0.54b	High
Trillos et al54 (2016)	0.93	0.07	0.87	0.12	1.00	1.05	Low

^a CCT = carpal compression test; +LR = positive likelihood ratio; −LR: = negative likelihood ratio; NPV = negative predictive value; NR = not reported; PPV = positive predictive value.

^b Values calculated by the authors of this study.

One study combined Durkan test with the Phalen test and described it as the Phdurkan test which “was performed by manual compression of the carpal tunnel with the examiner’s index and middle fingers and flexion of the wrist with the examiner’s thumb and recorded as positive if associated with reproduction of distal paresthesia in the median nerve distribution.”⁴³ Phdurkan test resulted in a high Sn of 0.84, but a low Sp of 0.11.⁴³

Diagnostic Accuracy of Other Provocative Maneuvers for CTS Diagnosis

The diagnostic accuracy of other provocative maneuvers for CTS diagnosis, including Flick sign, SCT, tethered median nerve stress test, ULTT I and II, Gilliatt test, lumbrical provocation test, and pressure provocative test, was assessed in a fewer number of studies (n ≤ 3). The diagnostic accuracy properties of these tests are summarized in Table 3. In summary, SCT had consistently low Sn (0.32, 0.31, and 0.24) and Sp (0.61, 0.62) in 3 studies (167 patients) with unclear ROB.^49–51 The tethered median nerve stress test consistently demonstrated low Sn (0.50, 0.52. 0.43)^35,38,52 and high Sp of 0.92 in 1 study with low ROB.³⁵ Other provocative maneuvers were assessed in fewer than 2 clinical studies and were at high ROB.^{28,30,32,47,53}

Discussion

This SR summarizes 31 studies with highly variable qualities and results suggesting low-to-moderate accuracy for 12 provocative tests when used individually to diagnose CTS. Meta-analyses of 8 studies that addressed either the Phalen or Tinel test had wide CIs but suggested that the Phalen test has moderate Sn and Sp, whereas the Tinel test has low Sn and high Sp.

We found different diagnostic accuracy results for the provocative maneuvers compared with sensorimotor tests. In our previous SR of sensorimotor tests, we had half the number of studies included compared with this SR (16 sensorimotor vs 31 provocative maneuvers studies), and most of the sensorimotor studies had high ROB; therefore, making it impossible to make a firm conclusion about the diagnostic accuracy of sensorimotor tests for CTS diagnosis.⁸ In our SR of sensorimotor tests, we concluded that “among the included studies, none of the sensory or motor tests had consistently high diagnostic accuracy properties reported by high-quality evidence. Confirming the value of a single sensory or motor test for CTS diagnosis is pending future robust research.”⁸ This was different from the results of this study, where higher Sn and Sp values were reported for the Phalen test and Tinel sign.

We believe the differences observed in the diagnostic accuracy values of the provocative maneuvers compared with sensorimotor tests could be due to (1) different testing procedures and (2) they are testing different things, the provocation of the nerve and the other the effects of nerve injury. With provocative maneuvers, some sort of pressure (either with positioning, eg, the Phalen test, or with applying external pressure, eg, CCT) is applied in the carpal tunnel, which decreases the space of the tunnel.⁵⁵ This decreased space in the carpal tunnel increases the pressure on the median nerve, hence the provocation of CTS symptoms.^35,43 Whereas with sensorimotor testing for CTS, we test the skin area or muscles innervated by the median nerve and measure the loss in sensation, strength, or dexterity.^31,35 In addition, sensory and strength measures may be good for examining sensory and strength loss, but these could result from a number of conditions, not just CTS. Therefore, they may be less valuable for confirming a CTS diagnosis. These differences in testing procedure further suggest the importance of including provocative maneuvers and sensorimotor tests in clinical practice to confirm the diagnosis of CTS.

Comparing our findings with the scales, diagnostic questionnaires, and hand symptom diagrams for CTS diagnosis from our previous SR, we note that scales, questionnaires, and hand symptom diagrams could play a critical role in diagnosing CTS.⁷ For example, the Kamath and Stothard Questionnaire (KSQ) had consistently high diagnostic accuracy values in several studies.^2,56,57 These high accuracy values were further improved when combined with provocative maneuvers.⁷ For instance, Bland et al assessed the accuracy of the KSQ combined with the CTS-6 scale (a combination of sensorimotor tests, the Phalen test and Tinel sign), which led to an excellent Sn of 0.95.⁵⁸ All of these confirm our conclusion that a combined testing approach, including history taking, sensorimotor testing, subjective assessment, and provocative maneuvers are the most definitive approaches for clinical diagnosis of CTS.

We did not include studies that had participants that were asymptomatic (healthy) as their comparators. This resulted in excluding 53 articles in the overall 3 SRs. While acknowledging the amount of information not considered in this SR by making this decision, we believe that by doing so, we provide a synthesis of studies that are the most similar to a clinical setting: A clinician would not see clients that were asymptomatic in daily practice. Furthermore, it has been proven statistically that including these participants who are healthy would inflate the diagnostic accuracy values and shift them toward better outcomes,¹² which could introduce a measurement bias. When assessing any diagnostic tool in a sample of healthy people, the test would be 100% specific, meaning that nobody will be diagnosed as “sick”.⁵⁹

Our results align with the clinical recommendations made by the most recent Academy of Hand and Upper Extremity Physical Therapy and the Academy of Orthopedic Physical Therapy of the American Physical Therapy Association clinical practice guidelines.⁶⁰ Based on this guideline, “in those with suspected CTS, clinicians should use the Katz hand diagram, Phalen test, Tinel sign, and carpal compression test to determine the likelihood of CTS and interpret examination results in the context of all clinical exam findings.”⁶⁰ Furthermore, although not the focus of the current study, but this guideline also suggested that the Phalen test can not only be used as a diagnostic method as suggested in this current study, but it can also be used as a responsiveness outcome measure to track change over time.⁶⁰

Comparing our study to Núñez de Arenas-Arroyo et al,⁶¹ they reported higher Sn and Sp for the Phalen test, higher Sn and similar Sp for the Tinel sign. We believe these differences are due to the several methodological differences between the 2 SRs. First, Núñez de Arenas-Arroyo et al⁶¹ limited their reference standard test to electrodiagnosis, whereas we included studies with any type of reference standards used. Furthermore, as explained before, we did not include studies with asymptomatic participants as their comparators, as this has been statistically proven to artificially inflate the diagnostic accuracies.⁵⁹ We suspect that the higher reported diagnostic accuracies could be attributed to the differences in the patients’ characteristics. Our study differed from Núñez de Arenas-Arroyo et al⁶¹ since they addressed the most common provocative maneuvers and not all the tests. Therefore, we believe the current SR and the other 2 published SRs on assessing the accuracy of the questionnaires, scales, hand symptom diagrams/maps,⁷ and sensorimotor tests⁸ are the most comprehensive updates of the literature and should be considered when developing the future clinical practice guidelines.

Study Limitations

The largest limitation in our work was related to deficiencies in the evidence, which included limited studies for many of the 12 identified tests. Even though we developed our search strategy in consultation with a health science librarian, we might have missed studies. Furthermore, a potential publication bias exists because we only included published articles, and we did not search the gray literature.

Although we reported promising results for some provocative maneuvers, one must be cautious when interpreting our results because some studies had unclear or high ROB and applicability concerns. To tackle this, we recommend future studies comply rigorously with the well-established recommendation guidelines and design and implement low ROB and high-quality studies. Furthermore, given the COVID-19 pandemic and the increased emphasis on the online diagnosis and management of CTS, we recommend future studies to assess the validity, reliability, and diagnostic accuracy of these clinical tests in online settings to produce clinically relevant evidence.

Clinical Implications

Accurate diagnosis of persons with suspected CTS symptoms seems warranted to improve treatment outcomes and individual patient care. Consequently, the efficiency of interventions for CTS management can be maximized by making a correct and timely diagnosis. To achieve this, clinicians must adhere to high-quality evidence for implementing clinical diagnostic tests in clinical settings. The provocative maneuvers studied in this review did not yield consistently high diagnostic accuracies; therefore, we do not recommend any single test for diagnosing CTS. Combining multiple tests, including history taking, sensorimotor testing, the Phalen test, and the Tinel sign, might optimize diagnostic accuracy.⁷ Studies examining different tests combinations are needed to verify the ideal combination of tests, balancing accuracy, and brevity.

Conclusion

This study synthesized 31 studies on the diagnostic accuracy of the provocative maneuvers in the differential diagnosis of CTS. Studies with moderate ROB and highly variable results suggest low-to-moderate accuracy for 12 provocative tests when used individually to diagnose carpal tunnel syndrome. Meta-analyses of 8 studies suggest that the Phalen test has moderate Sn and Sp, whereas the Tinel test has low Sn and high Sp.

Author Contributions

Concept/idea/research design: A. Dabbagh, J.C. MacDermid, J. Yong, T.L. Packham

Writing: A. Dabbagh, J.C. MacDermid, T.L. Packham, R. Grewal

Data collection: A. Dabbagh, J. Yong, R. Grewal, E.C. Boutsikari

Data analysis: A. Dabbagh, J. Yong, E.C. Boutsikari

Project management: A. Dabbagh, J.C. MacDermid

Fund procurement: J.C. MacDermid

Providing participants: R. Grewal

Providing facilities/equipment: J.C. MacDermid

Consultation (including review of manuscript before submitting): J.C. MacDermid, J. Yong, R. Grewal

Funding

J. MacDermid was supported by a grant from the Canada Research Chair in Musculoskeletal Health Outcomes and Knowledge Translation and the Dr James Roth Chair in Musculoskeletal Measurement and Knowledge Translation. Her work is supported by a foundation grant from the Canadian Institutes of Health Research (#167284). C. Ziebart and R. Furtado were supported by the CIHR Doctoral Award. A. Dabbagh was supported by the Ontario Graduate Scholarship.

Systematic Review Registration

This review was registered in PROSPERO (CRD42018109031).

Data Availability Statement

Any required links or identifiers for data are present in the manuscript.

Disclosures

The authors completed the ICMJE Form for Disclosure of Potential Conflicts of Interest and reported no conflicts of interest.