Level-specific impairment of the 10-second grip-and-release test and grip strength in degenerative cervical myelopathy: analysis of 39 surgical cases

Abstract

Purpose

To characterise the relationship between the 10-second grip-and-release test (GRT) and grip strength—simple measures of upper limb function in degenerative cervical myelopathy (DCM)—with the responsible level.

Methods

We retrospectively evaluated patients with DCM who underwent posterior cervical decompression at our institution (2003–2020). Patients with ossification of the posterior longitudinal ligament, disc herniation, or prior cervical surgery were excluded. The responsible level was determined using neurological findings and kinematic magnetic resonance imaging (MRI). Patients were classified into C4/5, C5/6, C4/5 + 5/6, and C5/6 + 6/7 groups. Preoperative GRT and grip strength were compared between the groups using linear mixed models (LMMs) adjusted for age and sex, with patient ID as a random intercept to account for within-patient correlation of bilateral measurements.

Results

Thirty-nine patients (25 males, 14 females; mean age 59.7 ± 13.3 years) were analysed. In single-level disease, linear mixed models with patient ID as a random intercept were used to account for within-patient correlation of bilateral measurements. GRT was significantly lower in C5/6 than in C4/5 (p = 0.016), whereas grip strength showed no significant difference (p = 0.936). In two-level disease, neither GRT (p = 0.173) nor grip strength (p = 0.153) showed a statistically significant difference between C4/5 + 5/6 and C5/6 + 6/7 groups.

Conclusions

GRT was specifically and significantly impaired in C5/6 single-level disease, a finding that remained robust after accounting for within-patient correlation using linear mixed models. In two-level disease, similar trends were observed but did not reach statistical significance, warranting further investigation. These findings suggest that GRT may serve as a sensitive indicator of C5/6 involvement in DCM. The exploratory nature of this study necessitates validation in larger, prospective cohorts.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12891-026-09715-y.

Introduction

Degenerative cervical myelopathy (DCM), the most common cause of nontraumatic spinal cord dysfunction in adults, presents with a wide spectrum of symptoms, including hand clumsiness and gait disturbance [1–3]. Population-based and community-dwelling cohort studies have reported that cervical cord compression on magnetic resonance imaging (MRI) is prevalent and often accompanied by subclinical neurological symptoms, suggesting that a substantial proportion of affected individuals remain undiagnosed [4, 5]. If left untreated, DCM carries a high risk of deterioration, falls, and secondary spinal cord injury [6, 7], leading to irreversible motor deficits and significant impairment of activities of daily living and health-related quality of life [2, 8]. The socioeconomic burden is also considerable, with increased healthcare costs and loss of independence associated with mobility decline and long-term care needs [9]. Accurate identification of the responsible level is essential for surgical planning, particularly when selective posterior decompression is considered [10–12]. Even in cases where multilevel laminoplasty is performed, precise level diagnosis provides important information for understanding symptom–lesion correlations and predicting surgical outcomes. Dynamic factors, such as cervical extension, can also contribute to cord compression, and the value of kinematic MRI has been demonstrated in this context [13].

The 10-second grip-and-release test (GRT), originally proposed by Ono et al. as part of the “myelopathy hand” concept, has been widely used as a simple bedside measure to quantify upper limb dexterity impairment [14]. Recent motion analysis studies using inertial sensors have reported various abnormalities, such as reduced finger extension velocity and delayed switching, expanding their quantitative potential [15]. Grip strength is another simple, objective measure frequently used in orthopaedic and primary care settings. Its utility as a screening tool for DCM has been reported, and age- and sex-specific cut-off values with high sensitivity have been proposed [16]. Moreover, combining GRT and grip strength has been suggested to enhance diagnostic performance compared with either test alone [17].

However, how these two bedside tests differentially reflect impairments at the cervical level has not been fully elucidated, which is particularly problematic in multilevel diseases where overlapping symptoms often make level diagnoses difficult. Therefore, the purpose of this study was to compare preoperative GRT and grip strength across patients with responsible levels confined to single-level (C4/5, C5/6) or two-level (C4/5 + 5/6, C5/6 + 6/7) disease, adjusted for age and sex, to clarify level-specific functional patterns and assess their combined utility as adjunctive diagnostic indicators for responsible level diagnosis in DCM.

Methods

Study design and ethical considerations

In this single-centre, retrospective, observational study, we reviewed the cases of patients who underwent posterior cervical decompression surgery (segmental partial laminectomy) or cervical micro-endoscopic laminotomy [18] at our institution between January 2003 and December 2020. The study was approved by the Institutional Ethics Committee (approval number: REC2025-068) and conducted in accordance with the Declaration of Helsinki. As only existing clinical data were used, informed consent for participation was obtained using an opt-out procedure.

Participants and exclusion criteria

Eligible patients were those with DCM whose responsible level was confined to one or two intervertebral levels. Patients with ossification of the posterior longitudinal ligament, cervical disc herniation, or a history of cervical surgery were excluded. Additional exclusion criteria included concomitant cervical radiculopathy, cervical spondylotic amyotrophy (CSA), cerebral palsy (CP), history of wrist or elbow disorders, decompression involving three or more levels, levels of responsibility not corresponding to the study definition, and missing clinical data.

Determination of the responsible level

The responsible level was determined by integrating preoperative neurological findings, including those elicited by cervical extension loading tests and kinematic MRI, followed by a consensus among ≥ 2 experienced spine surgeons, as previously described [19].

Patients were classified into four groups: single-level disease (C4/5 and C5/6) and two-level disease (C4/5 + 5/6, C5/6 + 6/7). For descriptive purposes, decompression levels were also summarised, with each operated level counted individually; cases with multiple levels were allocated to the “multi-level” category.

Clinical assessments

Preoperative GRT and grip strength were evaluated in all patients. All measurements were performed by fully qualified orthopaedic surgeons on the day before or on the day of surgery. GRT was conducted according to the original method of Ono et al., in which patients repeatedly opened and closed their fingers as rapidly as possible for 10 s, and the number of movements was recorded [14]. Bilateral grip strength was measured using a Jamar dynamometer. Patients were tested in the standing position when possible or in the sitting position when standing was difficult. Both tests were performed bilaterally, and the right and left hands were analysed separately (i.e., bilateral values were not averaged, and neither the maximum value nor the dominant hand alone was used). To assess measurement reliability, a subset of patients underwent repeated testing. Inter-rater reliability was evaluated by comparing assessments performed by different examiners, and intra-rater reliability was examined in patients tested twice by the same examiner within a short interval (≤ 3 months).

Statistical analysis

Because both hands were measured in each patient, bilateral observations are not independent. To account for this within-patient correlation, between-group comparisons of GRT and grip strength were performed using linear mixed models (LMMs) with the lme4 and lmerTest packages in R, specifying patient ID as a random intercept. Each model included the responsible level as a fixed factor, with age and sex as covariates: Outcome ~ Responsible level + Age + Sex + (1 | Patient ID). Degrees of freedom and p-values were calculated using the Satterthwaite approximation. The intraclass correlation coefficient (ICC) was computed for each analysis to quantify the proportion of total variance attributable to between-patient differences. As a sensitivity analysis, we also performed ANCOVA using patient-level averages (mean of left and right hands) to confirm the robustness of the LMM results. For unadjusted analyses, Student’s t-test or Mann–Whitney U test was additionally applied, as appropriate, and effect sizes were calculated using Hedges’ g. Statistical significance was defined as a two-sided p-value < 0.05. All analyses were conducted using R software (version 4.5.1; R Foundation for Statistical Computing, Vienna, Austria).

Results

Patient flow and baseline characteristics

Among the 322 patients with DCM who underwent posterior decompression surgery during the study period, 62 were excluded because of patient background factors, including any history of cervical spine surgery (n = 18), concomitant cervical radiculopathy (n = 26), CSA (n = 11), CP (n = 5), or wrist or elbow disorders (n = 2). An additional 199 patients were excluded based on the surgical or responsible level criteria, including decompression involving three or more levels (n = 143) and responsible levels that did not meet the study definition (n = 56). Furthermore, 22 patients were excluded because of missing clinical data. Finally, 39 patients were included in the final analysis (Fig. 1). The cohort consisted of 26 males and 13 females, with a mean age of 50.9 ± 10.3 years. Demographic characteristics by group are summarised in Table 1. The overall distribution of decompression levels in the 260 eligible patients after background exclusion is shown in Fig. 2.

Fig. 1.

Patient flow diagram. Among 322 patients screened, 283 were excluded because of revision surgery, concomitant cervical radiculopathy, cervical spondylotic amyotrophy (CSA), cerebral palsy (CP), wrist or elbow disorders, decompression involving ≥ 3 levels, or unmatched responsible levels. An additional 22 patients were excluded owing to missing clinical data. Ultimately, 39 patients were included in the analysis. CSA, cervical spondylotic amyotrophy; CP, cerebral palsy; OPLL, ossification of the posterior longitudinal ligament

Table 1.

Baseline demographic and clinical characteristics of the study cohort

Characteristic	Overall (N = 39)	C4/5 (N = 10)	C5/6 (N = 5)	C4/5 + 5/6 (N = 12)	C5/6 + 6/7 (N = 12)
Age (years)	59.7 (13.3)	62.4 (18.5)	54.6 (7.5)	59.2 (12.9)	60.1 (11.3)
Sex, Male n (%)	25 (64%)	6 (60%)	4 (80%)	7 (58%)	8 (67%)
JOA score, mean (SD)	10.4 (4.7)	12.4 (1.5)	10.4 (6.2)	8.3 (6.4)	11.1 (2.2)
Grip Strength, mean (kg)	22.2 (11.1)	22.9 (13.5)	29.2 (7.2)	17.9 (12.3)	23.0 (8.0)
GRT count, mean (times)	18.6 (8.0)	21.6 (10.0)	19.8 (5.9)	15.5 (7.9)	18.7 (6.4)

Data are presented as mean ± standard deviation (SD) or number (percentage)

GRT = 10-second grip-and-release test

Fig. 2.

Overall distribution of decompressed levels in 260 patients with degenerative cervical myelopathy who were eligible after background exclusions. Each decompressed intervertebral level was counted individually. Patients with ≥ 3 decompressed levels or non-target responsible levels were subsequently excluded from the present analysis, resulting in the final cohort of 39 patients

10-second GRT

In single-level disease, the LMM adjusted for age and sex showed that GRT was significantly lower in the C5/6 group than in the C4/5 group (coefficient = −5.99; SE = 2.09; p = 0.016; ICC = 0.635) (Table 2).

Table 2.

Linear mixed model estimates for age- and sex-adjusted grip-and-release test (GRT) and grip strength across responsible levels

Contrast	Outcome	Coefficient	SE	p-value (LMM)
Single-Level (C5/6) - (C4/5)	GRT (times)	−5.99	2.09	0.016
	Grip Strength (kg)	0.29	3.48	0.936
Two-Levels (C5/6 + 6/7) - (C4/5 + 5/6)	GRT (times)	3.44	2.43	0.173
	Grip Strength (kg)	4.90	3.30	0.153

Data represent coefficients from linear mixed models with standard errors (SE) and p-values

ICC Intraclass correlation coefficient

In two-level disease, GRT tended to be lower in the C4/5 + 5/6 group than in the C5/6 + 6/7 group, but this difference did not reach statistical significance (coefficient = 3.44; SE = 2.43; p = 0.173; ICC = 0.877) (Table 2). Adjusted means (± SE) for each group are listed in Supplementary Tables 1, and distributions are illustrated in Fig. 3a, b. Unadjusted comparisons are shown in Supplementary Tables 2 and did not alter these conclusions.

Fig. 3.

Dot and box plots of preoperative GRT residuals. Dot plots with mean bars of grip-and-release test (GRT) residuals, adjusted for age and sex by ordinary least-squares regression. a Single-level disease (C4/5 vs. C5/6). b Two-level disease (C4/5+5/6 vs. C5/6+6/7). The p-values were derived from linear mixed models accounting for within-patient correlation

Grip strength

In single-level disease, no significant difference in grip strength was detected between the C4/5 and C5/6 groups by LMM (coefficient = 0.29; SE = 3.48; p = 0.936; ICC = 0.803).

In two-level disease, grip strength tended to be lower in the C4/5 + 5/6 group than in the C5/6 + 6/7 group, but this difference did not reach statistical significance (coefficient = 4.90; SE = 3.30; p = 0.153; ICC = 0.774). Adjusted means (± SE) are listed in Supplementary Tables 1, and distributions are depicted in Fig. 4a, b. Unadjusted analyses are shown in Supplementary Table 2.

Fig. 4.

Dot and box plots of preoperative grip strength residuals. Dot plots with mean bars of grip strength residuals, adjusted for age and sex by ordinary least-squares regression. a Single-level disease (C4/5 vs. C5/6). b Two-level disease (C4/5+5/6 vs. C5/6+6/7). The p-values were derived from linear mixed models accounting for within-patient correlation

We additionally explored the association between baseline JOA score and functional measures. JOA score showed positive correlations with the average GRT count (Spearman’s ρ = 0.445, p = 0.014, n = 30) and with the average grip strength (Spearman’s ρ = 0.488, p = 0.005, n = 31) (Supplementary Fig. S1).

Reliability analysis

Among the 39 patients included in the final analysis, inter-rater reliability data were available for 25 patients, whereas intra-rater reliability data were available for 3 patients. Inter-rater reliability showed excellent agreement for both grip strength (intraclass correlation coefficient [ICC] = 0.888, 95% CI 0.760–0.949) and GRT (ICC = 0.805, 95% CI 0.544–0.923) (Supplementary Table 3). Because the intra-rater assessment was limited to only 3 patients, the precision of the reliability estimate is limited; therefore, the intra-rater results are presented descriptively and interpreted with caution.

Discussion

In this study, we investigated whether preoperative GRT and grip strength exhibit distinct, level-specific patterns of decline in patients with DCM. Using linear mixed models to appropriately account for within-patient correlation of bilateral measurements, we confirmed that GRT was specifically and significantly impaired in C5/6 single-level disease (p = 0.016), whereas grip strength showed no significant difference between C4/5 and C5/6 groups. In two-level disease, although similar trends were observed, the differences in both GRT (p = 0.173) and grip strength (p = 0.153) did not reach statistical significance. These findings suggest that GRT is particularly sensitive to C5/6 involvement in DCM, potentially reflecting corticospinal tract dysfunction at this level.

GRT, initially described by Ono et al., is widely recognised as a bedside measure of hand dexterity [14]. Recent motion analyses have reported that impaired finger extension velocity and delayed switching are characteristic abnormalities in DCM [15], implicating dysfunction of the C7-innervated finger extensors and possible long tract involvement [20]. Therefore, the marked decline in GRT observed in the C5/6 group may reflect frequent C7 involvement, leading to impaired coordination of finger movements. In addition, loss of the flexion tenodesis effect, where wrist flexion typically facilitates finger extension, may further aggravate this decline in C7 lesions. Although physiologically plausible, this remains a supplementary mechanism, as direct evidence is still lacking.

In two-level disease, although a similar trend was observed with GRT slightly lower in the C4/5 + 5/6 group than in the C5/6 + 6/7 group, the difference did not reach statistical significance after accounting for within-patient correlation (p = 0.173). This may partly reflect insufficient statistical power given the moderate sample size. Alternatively, as both patterns involve C5/6, the trend may be better explained by wrist kinematic factors—particularly reduced wrist extension and loss of the tenodesis effect—in more rostral lesions, rather than by segmental differences alone.

In contrast, grip strength reflects maximal grasp force, which relies primarily on the counterforce produced by C6-innervated wrist extensors [21–23]. Impairment of the wrist extension torque due to C6 dysfunction substantially reduces grip efficiency. C7 further contributes by stabilising the wrist in neutral position to slight extension through the action of both wrist flexors and extensors [19]. When C6 and C7 dysfunction coexist, as in C4/5 + 5/6 lesions, the resulting imbalance between the agonist and antagonist muscles may lead to flexion collapse during grasping, thereby producing more pronounced weakness. Although the difference in grip strength between C4/5 + 5/6 and C5/6 + 6/7 groups did not reach statistical significance after correcting for within-patient correlation (p = 0.153), the observed trend is consistent with this neuroanatomical interpretation and with previous reports of tenodesis-related grip phenomena in cervical spinal cord injury [24].

Although the present study mainly focused on the C6–C7 segments, contribution from the C8-innervated finger flexors cannot be excluded, as these muscles are also involved in generating maximal grip force. Our cohort did not include patients with C7/T1 as a responsible level; however, subclinical C8 involvement through indirect compression effects cannot be entirely excluded, particularly in multi-level disease, and this may have influenced grip strength measurements.

The principal finding of this study was that GRT is specifically and significantly impaired in C5/6 single-level disease, a result that remained robust after accounting for within-patient correlation using LMMs and was confirmed by sensitivity analysis using patient-level averages. In two-level disease, although similar trends were observed for both GRT and grip strength, the differences did not reach statistical significance, which may reflect limited statistical power. These two-level findings should therefore be interpreted as exploratory and hypothesis-generating rather than confirmatory. Although GRT decline in C5/6 lesions may indicate greater C7 involvement compared with C4/5, it should be interpreted as a relative tendency rather than a definitive marker for level diagnosis. Further studies with larger cohorts, including other levels such as C3/4 or C6/7, are needed to validate these findings and determine level-specific cut-off values that could assist in estimating the responsible level in individual cases.

Strengths of the study

The principal strength of this study was the precise identification of the responsible level in patients with degenerative cervical myelopathy. Surgical levels were determined by integrating neurological findings, including those from cervical extension loading tests and kinematic MRI, with consensus among experienced spine surgeons. This approach enabled a precise evaluation of the relationship between the responsible level and preoperative GRT and grip strength, while restricting the analysis to one- or two-level lesions to minimise confounding. Clinically, the use of two simple bedside tests that require no special equipment provides objective and practical support for level diagnosis, particularly in patients with multilevel compression, where decision-making is often challenging. These tests may improve the accuracy of surgical planning by supplementing surgeons’ experience with quantitative information. Furthermore, the accumulation of such quantitative data may provide a foundation for future diagnostic innovations, as demonstrated by recent studies showing the feasibility of machine learning–based approaches for DCM screening using smartphone-based GRT analysis [25] and for predicting postoperative outcomes with high accuracy [26, 27].

Limitations

This study has several limitations that warrant consideration. The primary limitation was the relatively small sample size, particularly the C5/6 single-level subgroup (n = 5 patients); therefore, our findings should be considered exploratory and hypothesis-generating rather than confirmatory. While the GRT finding in single-level disease was statistically robust (p = 0.016 by LMM), the study was likely underpowered to detect moderate effect sizes in two-level disease subgroups, which may explain the non-significant results. This exploratory nature also precluded a separate analysis of isolated C6/7 lesions. However, to the best of our knowledge, this study is the first to report suggestive evidence for differential level-specific patterns between GRT and grip strength in DCM. This highlights the need for future large-scale, prospective, multicentre studies to validate these findings and investigate patterns at rarer lesion levels, for which our current work provides a foundation.

In addition, because the right and left hands were measured in each patient, bilateral observations are not independent. To address this, we used linear mixed models with patient ID as a random intercept to account for within-patient correlation, and the sensitivity analysis using patient-level averages confirmed the robustness of our findings. Furthermore, our cohort excluded patients with disc herniation, as these patients typically underwent anterior approaches at our institution. While this restriction ensured a homogeneous pathophysiology (degenerative spondylotic compression), it limits generalisability to the broader DCM population including those with disc herniation-related myelopathy. Although the right and left hands were analysed separately, the use of LMMs in the present analysis appropriately accounted for this clustering.

Furthermore, our cohort was intentionally restricted to one- or two-level surgical cases to ensure analytical clarity; thus, caution is warranted when extrapolating these findings to patients with more extensive multilevel diseases. This cross-sectional analysis focused on a single preoperative assessment. Future longitudinal studies are necessary to elucidate the temporal relationship between functional measures and postoperative recovery.

Finally, although our statistical models were adjusted for age and sex, other clinical factors that may influence grip strength or manual dexterity—such as hand dominance, occupational hand use, comorbidities (e.g., peripheral neuropathy and diabetes), sarcopenia, and time from symptom onset—were not systematically available in this retrospective cohort; therefore, residual confounding may remain. Since this was a single-centre study in a Japanese cohort, the generalisability of these findings to other populations remains to be determined, further emphasising the importance of subsequent multicentre investigations.

Conclusion

We compared preoperative GRT and grip strength according to rigorously defined responsible levels in patients with degenerative cervical myelopathy using linear mixed models to account for within-patient correlation. GRT was specifically and significantly impaired in C5/6 single-level disease, a finding that remained robust across multiple analytical approaches. In two-level disease, similar trends were observed but did not reach statistical significance, warranting validation in larger cohorts. These exploratory findings suggest that GRT may serve as a sensitive indicator of C5/6 involvement in DCM and provide a foundation for future confirmatory studies.

Authors’ contributions

H.K. designed the study, performed the statistical analysis, and drafted the manuscript. H.K., K.W., K.O., T.N., K.K., Y.K., K.K., and M.N. collected the data. H.K., K.W., K.O., T.N., K.K., M.H., Y.K., and T.K. contributed to data interpretation and critical revision. All authors read and approved the final manuscript.

Funding

This study received no funding.

Data availability

Data sharing is not applicable to this article as no datasets were generated or analysed beyond existing clinical records.

Declarations

Ethics approval and consent to participate

This study was approved by the Ethics Committee of Fukushima Medical University (approval number: REC2025-068) and conducted in accordance with the Declaration of Helsinki. Because only existing clinical data were used, informed consent for participation was obtained in the form of an opt-out on the institutional website.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.