Failure to Compensate: Nerve Injury Patients Use Their Injured Dominant Hand, Even When Their Non-Dominant is More Dexterous

Abstract

Objective:

To identify how individuals respond to unilateral upper extremity peripheral nerve injury via compensation (increased use of the non-dominant hand). We hypothesized that injury to the dominant hand would have a greater impact on hand usage (left vs. right choices). We also hypothesized that compensation would not depend on current (post-injury) non-dominant hand performance, because many patients undergo rehabilitation that is not designed to alter hand usage.

Design:

Observational survey, single-arm.

Settings:

Academic research institution and referral center.

Participants:

48 adults with unilateral upper extremity peripheral nerve injury. Another 14 declined participation. Referred sample, including all eligible patients from 16 months at one nerve injury clinic and one hand therapy clinic.

Interventions:

Not applicable.

Main Outcome Measures:

Hand usage (% of actions with each hand) via Block Building Task. Dexterity via Jebsen-Taylor Hand Function.

Results:

Participants preferred their dominant hand regardless of whether it was injured: hand usage (dominant/non-dominant) did not differ from typical adults, regardless of injured side (p > 0.07), even though most participants (77%) were more dexterous with their uninjured non-dominant hand (mean asymmetry index −0.16 ± 0.25). The Block Building Task was sensitive to hand dominance (p = 2.0×10⁻⁴), and moderately correlated with Motor Activity Log amount scores (r² = 0.33, p = 2.9×10⁻⁵). Compensation was associated only with dominant hand dexterity (p = 3.9×10⁻³), not on non-dominant hand dexterity, rehabilitation, or other patient/injury factors (p > 0.1).

Conclusions:

Peripheral nerve injury patients with dominant hand injury do not compensate with their unaffected non-dominant hand, even if it is more dexterous. For the subset of patients unlikely to recover function with the injured hand, they could benefit from rehabilitation that encourages compensation with the non-dominant hand.

Introduction

“Compensation,” use of the non-dominant hand (NDH) after impairment of the dominant hand (DH), could be critical for patients with chronic unilateral impairment of the DH following common clinical conditions including stroke, amputation, and nerve injury. However, few studies have focused on the specific consequences of unilateral impairment.¹ Impairment of the DH leads to greater impact on function and activities than impairment of the NDH.^{2, 3} While many activities are bimanual, use of the DH is critical for many activities.^4–6 Nevertheless, it remains unknown when and why patients compensate with the NDH after DH impairment.

Individuals with peripheral nerve injury (PNI) provide an ideal population to study unilateral impairment. Approximately 33–39% of individuals with upper extremity PNI (55–66,000 individuals/year in the USA) never achieve satisfactory motor recovery after surgical repair, regardless of any rehabilitation they receive.^7–13 Therefore, PNI patients have substantial unmet rehabilitation needs. Moreover, PNI research has broad relevance to many clinical conditions because PNI patients can provide a picture of “healthy brain” compensation, a critical baseline for understanding the motor consequences of stroke.

Few methods currently exist to measure compensation. To address this gap, we introduce a new construct and relevant assessments. The construct “hand usage” represents which hand an individual uses in an unconstrained task, and is a continuous measure that depends in part on handedness.^{14, 15} By definition, compensation reflects a change in hand usage. The current gold standard for hand usage measurement is the Motor Activity Log (MAL), which assesses affected hand usage after stroke.¹⁶ However, the MAL ignores the unaffected hand, and has the limited accuracy inherent to self-report surveys.^{17, 18} To directly measure hand usage in patients, we adapted a basic science measurement tool, the Block Building Task.^{14, 15}

Our overarching methodological goal was to quantify compensation after unilateral DH injury by measuring hand usage. With this quantification, we tested the hypothesis that patients with an affected DH would show greater atypicality in hand usage (i.e. compensation) than patients with an affected NDH. However, we found that patients maintained typical DH usage regardless of injury side. Because of this first finding, we tested a secondary hypothesis: that compensation did not depend on current NDH performance or rehabilitation, because many patients undergo rehabilitation that is not designed to alter hand usage.

Methods

Study Overview

This was a cross-sectional single-arm study involving a single laboratory visit. All participants gave informed consent, and all procedures were approved by the local Institutional Review Board. Data were stored and managed via the Research Electronic Data Capture (REDCap) system.¹⁹

Participants

Inclusion criteria included: ≥ 18 years old, English-speaking, unilateral upper extremity peripheral nerve injury (defined as mechanical origin rather than pathological or tumor), with a Quick Disabilities of the Arm, Shoulder, and Hand (Q-DASH)²⁰ score ≥ 18, measured at the start of the study session. This threshold was chosen to select individuals whose life is impacted by their impairment, at one Minimum Clinically Important Difference²¹ above zero. This threshold captures a wide range of PNI patients, because it also lies one standard deviation below the mean of upper extremity disorder patients.²²

Exclusion criteria included cognitive disorders, uncorrected visual impairment, chronic pain diagnoses, major mental health diagnosis (not including depression, anxiety, bipolar, or post-traumatic stress disorder), surgery within preceding two months, or motor function diagnosis affecting the arm contralateral to their PNI in preceding two years.

In order to examine effects of injury severity, injuries of all types and severity levels were recruited, within the above criteria.

48 individuals participated in the study, following the recruitment process shown in Figure 1. The recruitment sites were a nerve injury clinic [anonymized] and an outpatient hand therapy clinic [anonymized]. For excluded individuals, exclusion reasons were not recorded to avoid collecting data on individuals who did not consent to participation. Demographic information for our 48 participants is shown in Table 1. We found no significant differences between groups (affected DH vs. affected NDH).

Figure 1.

Recruitment flowchart.

Table 1:

Demographic and Historical Data

		Total n = 48		DH Affected n = 22		NDH Affected n = 26		Between Groups
		Mean	SD	Mean	SD	Mean	SD	t / χ2	p
Age		44.42	15.55	42.23	15.53	46.27	15.63	−0.895	0.375
Sex (Female)		28 (58%)		15 (68%)		13 (50%)		0.959	0.327
Race	Native Amer.	3		1		2		0.022	0.881
	White	37		18		19		0.139	0.709
	Black/AA	9		4		5		0.077	0.781
	Other	2		0		2		0.365	0.546
Education	< 9 years	2		0		2		0.365	0.546
	Some HS	10		4		6		0.004	0.953
	HS or equiv	16		9		7		0.514	0.473
	Some college	14		7		7		0.003	0.958
	College +	6		2		4		0.048	0.827
Dominant Hand = Right		41 (85%)		20 (91%)		21 (81%)		0.338	0.561
Affected Hand = Dominant		22 (46%)		22		0		–	–
Severity	Neurapraxia	8		4		4		0.017	0.897
	Axonotmesis	18		7		11		0.201	0.654
	Neurotmesis	22		11		11		0.059	0.809
PT prescriptions	None	23		12		11		0.309	0.578
	One	13		3		10		2.568	0.109
	> 1	12		7		5		0.448	0.503
OT prescriptions	None	23		8		15		1.402	0.236
	One	12		8		4		1.790	0.181
	> 1	13		6		7		0.089	0.765
# Surgeries		1.06	0.81	1.14	0.64	1.00	0.94	0.577	0.567
Months Since Last Surgery		5.56	8.79	7.26	11.62	3.65	3.12	1.242	0.223
Months Since Injury		18.56	26.60	17.27	12.55	19.65	34.57	−0.306	0.761
Affected Nerve	Ulnar	27		13		14		0.005	0.942
	Median	28		14		14		0.153	0.695
	Radial	22		7		15		2.256	0.133
	Cutaneous	4		3		1		1.495	0.221
	Other	8		4		4		0.067	0.796
Injury Level	Wrist	17		6		11		0.612	0.434
	Forearm	11		6		5		0.100	0.752
	Brachial Plex.	16		8		8		0.010	0.918
	Hand	5		2		3		0.039	0.843
	Upper Arm	5		3		2		0.039	0.843
	Elbow	7		4		3		0.057	0.811
Multiple Nerves Affected?		24		13		11		0.755	0.385
Completed follow-up?		19 (40%)		9 (41%)		10 (38%)		0.105	0.746
Months until follow-up		17.74	4.65	17.22	3.55	18.20	5.21	−0.447	0.661

Means and SD for numerical data, counts for categorical data. Between-groups differences assessed by t-tests for numerical data and chi-square tests for categorical data. DH = dominant hand, NDH = non-dominant hand; AA = African American, HS = high school, PT = physical therapy, OT = occupational therapy, “Surgery” = for this injury. No participants identified as Asian American/Pacific Islander, or as Hispanic and Latino. Percentages given for key demographic items only.

This study was powered to detect an effect of Affected Side on hand usage (DH vs NDH). Our initial a priori power calculation determined that 80% power with two-sided t-tests at α = 0.05 would require 192 participants to detect an effect size of 0.5. The study was halted at n = 48 due to funding limitations, prior to any data analysis. After that point, we determined that our a priori calculation substantially underestimated the potential effect sizes (d > 1.0, see Results). Power analysis with an informed effect-size estimate of 0.9 led instead to a required sample size of 42.

Outcome Measures

Hand Usage

The Block Building Task^{4, 15} directly measured hand usage during precision reach-to-grasp movements in an unconstrained environment. Participants sat in front of a table with 40 Lego blocks (The Lego Group, Billund, Denmark): 4 copies each of 10 block types, in standardized locations. Participants were presented with a “model,” an abstract construction containing 10 blocks (one of each type). Participants built the model on a small Lego base plate, as shown in Figure 2. This was repeated for four models. Blocks were not replaced, so participants eventually picked up all 40 blocks.

Figure 2.

Block Building Task. Participants build 4 models (bottom center) from 40 Lego bricks, with no cues about how to accomplish this task.

Participants were instructed to build the model as quickly and accurately as possible; participants received no cues about how to use their hands. Performance was video-recorded, and each video was reviewed by two raters using BORIS event-logging software²³ to assess the fraction of grasps made with either hand.

To determine suitability of the Block Building Task, the effect of Affected Side (DH vs. NDH) on affected hand usage (% affected) was measured via a two-tailed between-groups t-test, α = 0.05.

Compensation

Compensation was measured as the difference in hand usage (fraction DH) between each patient and a typical adult baseline: 20 typical right-handed adults, age range 18–33; collected in 2013–2014 at University of Lethbridge, Alberta Canada)²⁴. Because typical adult data were only available for right-handed individuals, compensation analysis included only right-handed participants from the current study (n = 20 with Affected DH, n = 21 with Affected NH).

The presence of compensation was assessed via a one-way ANOVA on the effect of Group (Typical, Affected DH, Affected NDH) at α = 0.05. Post-hoc tests were performed using Tukey-Kramer’s HSD. To determine whether the groups differed in variance, a Bartlett’s test of equal variances was used (matlab “vartestn”) at α = 0.05. Post-hoc pairwise tests were performed using two-sample F-tests (matlab “vartest2”) at α = 0.0167 (i.e. 0.05 / 3 to Bonferroni correct for three pairs).

A secondary analysis was performed to identify possible drivers of compensation in patients with an affected right DH (n = 20). In this analysis, compensation was compared against five variables: (1) NDH dexterity, (2) DH dexterity, (3) asymmetry index, (4) # occupational therapy prescriptions, and (5) # physical therapy prescriptions.

Dexterity

Dexterity was measured via the Jebsen-Taylor Hand Function Test (JTHF).²⁵ Four of eight subtests were used (small object grasp, simulated feeding, stacking checkers, stacking heavy objects), because only these four measure unique constructs.^{26, 27} The JTHF and its subtests have excellent test-retest reliability (ICCs = 0.84–0.97), and good construct validity for evaluating hand function as measured by grip strength and DASH (r² = 0.11–0.25, p < 0.001).²⁸ Two experimenters recorded the time for each subtest, performance was video-recorded. Completion time (average across subtests, ceiling 90 sec) was calculated for each hand, and converted to “dexterity” as: 90 - completion time. For display purposes, dexterity values were divided by 90 to create a % score.

To compare dexterity across hands, we recalculated bilateral JTHF completion times as an “asymmetry index,” (NDH-DH)/(NDH+DH), such that positive asymmetry indicates greater dexterity for DH than NDH.

Dexterity and asymmetry were compared with NDH usage via correlation, using the nonparametric rank-based Kendall’s τ due to its robustness to the outliers in our data set. The two therapy variables were compared with NDH usage via a one-way ANOVA on the factor “Number of Prescriptions Completed.” All secondary analyses were performed at α = 0.01 (i.e. 0.05 / 5 to Bonferroni correct for five variables).

Secondary Measures

To provide preliminary validation of the Block Building Task, participants also completed the Motor Activity Log (MAL),¹⁶ a standard structured interview of hand usage at home, designed for stroke patients. Hand usage was surveyed via the MAL “amount scale,” the participant’s report of how much they used their affected upper extremity (0–5 scale), which is reliable (r = 0.82) and has been validated with accelerometry (r = 0.47, p < 0.01).¹⁶ Scores were averaged across 30 activities.

The Edinburgh Handedness Inventory⁵ was used to survey self-reported pre-injury hand preference.

Patient medical history was acquired from medical records. Therapy and pain information were collected from an interview with the patient. Pain was measured as injury-related pain during the last 7 days, on a 1–10 scale.²⁹

Data Analysis

Data analyses were performed in MATLAB 9.9.0 (Mathworks, Natick MA). Descriptive statistics are provided as mean ± standard deviation.

To identify patient factors that affected compensation in participants with DH injury, an established two-step approach was used.³⁰ First, factors were tested for univariate relationship (numerical variables via Kendall’s τ correlation, categorical variables via ANOVA), and factors approaching significance (a priori threshold: p < 0.2) were entered into a multiple linear regression (MATLAB “fitlm” function). 22 possible factors were tested (13 unique factors plus dummy variables): NDH dexterity, DH dexterity, asymmetry index, # OT prescriptions, # PT prescriptions, age, sex, months since surgery, pain, injury severity (neurapraxia, axonotmesis, neurotmesis), injury location (coded as 6 dummy variables: wrist, forearm, brachial plexus/thoracic outlet, hand, upper arm, elbow), injured nerve (5 dummy variables: ulnar, median, radial, cutaneous, other), and whether multiple nerves were affected (y/n). Due to the exploratory nature of the regression analysis, significance was determined at α = 0.05.

For “Affected Nerve,” to avoid rank-deficient models, the “Other” category encompassed injuries to the following nerves: thoracic (n=1 among participants with DH injury), spinal accessory (n=1), posterior interosseus (n=1), and anterior interosseus (n=1).

Results

Injury Side Affects Hand Usage

In our 48 unilateral upper extremity PNI patients, Affected Side (DH vs NDH) significantly affected their usage of the affected hand during the Block Building Task (t = 4.05, p = 1.95×10⁻⁴, effect size d = 1.02; Figure 3A). However, when measuring hand usage as “usage of the DH,” the two groups did not differ (t = −1.44, p = 0.156; Figure 3B). In other words, both groups tended to use their dominant hand.

Figure 3.

Participants with unilateral PNI use their DH regardless of injury. A Hand usage depends on affected side when usage is measured as “fraction affected,” B. but not when measured as “fraction dominant.” All participants (n = 48). C. Right-handed participants with PNI (n = 41), regardless of affected side, do not differ from typical right-handed adults (n = 20; Stone & Gonzalez 2015).

On Average, Patients Did Not Compensate With the Non-Dominant Hand

To measure compensation, we used a one-way ANOVA to determine the effect of Group (affected DH, affected NDH, typical adults) on DH usage in our 41 right-handed PNI participants. We found a significant main effect of Group (F(2,58) = 4.33, p = 0.017), but post-hoc analysis indicated that this effect arose from a difference between Affected DH vs. Affected NDH (p = 0.021): the frequency of DH use was significantly lower for patients with affected DH (59.2 ± 36.8%) than patients with affected NH (81.9 ± 22.0%). The post-hoc analysis with the typical group (previously published data²⁴) revealed that the typical group’s rate of DH usage (63.6 ± 7.2%) did not differ from either patient group, whether affected DH (p = 0.85) or affected NDH (p = 0.078), as shown in Figure 3C. Overall, our patients (regardless of affected side) did not show non-typical (i.e. compensatory) patterns of hand usage.

While our patient groups’ mean dominant hand usage was not atypical, the patients were more variable than the typical group: variance differed significantly between the three groups (χ² = 37.2, p = 8.2×10⁻⁹). Post-hoc tests indicated that this effect arose from low variance for the typical group compared to either Affected NDH (p = 8.2×10⁻⁶) or Affected DH (p = 1.6×10⁻⁹); the difference in variance between patient groups was not significant after correction for multiple comparisons (p = 0.028). Therefore, peripheral nerve injury led to increased inter-individual variability in hand usage, compared to typical adults. As a result, we performed a secondary analysis to identify why some patients compensated after DH nerve injury.

Compensation Depended on Dominant-Hand Impairment, Even When Non-Dominant Hand had Superior Performance

To explore our apparent lack of compensation, we tested whether compensation (NDH usage after DH injury) would depend on NDH dexterity, DH dexterity, asymmetry index, or # of prescriptions of physical or occupational therapy. Compensation correlated significantly with DH dexterity (τ = −0.460, p = 3.9×10⁻³) and asymmetry index (τ = −0.424, p = 7.9×10⁻³), but not NDH dexterity (τ = −0.198, p = 0.220), as shown in Figure 4A–B. Asymmetry index was redundant with DH dexterity: it correlated strongly with DH dexterity (τ = 0.791, p = 3.26×10⁻⁷) but not with NDH dexterity (τ = 0.148, p = 0.352). Despite the redundancy of the asymmetry index, it illustrates a key aspect of the data: most participants had greater dexterity for NDH than DH (17/22 participants, 77% of sample; mean asymmetry index −10.1 ± 26.3, Figure 4C).

Figure 4.

Compensation (NDH usage after DH injury) is correlated only with DH dexterity (n = 22 participants with DH injury). A. Compensation was negatively correlated with DH dexterity. Linear fit not shown due to two outlier participants with DH dexterity = 0. B. No relationship between compensation and NDH dexterity. C. Compensation negatively correlated with asymmetry index, i.e. when DH was less dexterous, more compensation occurred. Most participants showed asymmetry < 0, i.e. more dexterity for NDH than DH.

Together, these results show that, in patients with DH injury, compensation depended only on the severity of DH impairment. In other words, patients with less DH impairment were less likely to compensate, regardless of whether that (mild/moderate) impairment left their DH less capable than their unaffected NDH.

We found no effect of occupational therapy (τ = −0.216, p = 0.226) or physical therapy (τ = 0.088, p = 0.636) on compensation. We also found no effects of either kind of therapy on either hand’s dexterity (p > 0.5). Therefore, we found no sign that therapy influenced compensation patterns in our patients.

No Effect of Patient Characteristics on Compensation

To identify patient characteristics that drove compensation after DH injury, we identified factors with a potential univariate effect (p < 0.2) on compensation, and then tested those factors in a multiple linear regression. Three factors met the criterion: DH dexterity, median nerve, and multiple nerves. Full univariate results are shown in Table 2. (Asymmetry index also met the criterion, but we omitted it due to redundancy with DH dexterity, as described in the previous section.) Using these factors, we found a significant linear regression (F(18,22) = 4.57, p = 0.015, adjusted r² = 0.338). Only DH Dexterity contributed significantly to the model (t = −2.76, p = 6.13×10⁻³), not median nerve (t = 1.26, p = 0.224) or multiple nerves (t = 0.119, p = 0.907). Overall, our univariate and multivariate analyses indicated that compensation had no relationship with participant or injury characteristics.

Table 2:

Univariate Relationships with Compensation

Variable	F/τ	p
DH dexterity	−0.460	3.9×10−3
NDH dexterity	−1.98	0.220
Asymmetry index	−0.424	7.9×10−3
# OT prescriptions	−0.216	0.226
# PT prescriptions	0.088	0.636
Age	0.086	0.607
Sex	−0.034	0.533
Months since surgery	0.120	0.887
Injury severity	0.945	0.406
Injured Nerve – Ulnar	0.068	0.798
Injured Nerve – Median	2.173	0.160
Injured Nerve – Radial	0.288	0.599
Injured Nerve – Cutaneous	0.669	0.425
Injured Nerve – Other	0.235	0.634
Injury location – Wrist	0.717	0.793
Injury location – Forearm	0.317	0.582
Injury location – Plexus	0.018	0.895
Injury location – Hand	0.386	0.544
Injury location – Upper arm	0.381	0.546
Injury location – Elbow	0.029	0.866
Multiple nerves? (y/n)	2.36	0.141

Univariate relationship between each variable and compensation (NDH usage after DH injury). Numerical variables assessed via nonparametric correlation (Kendall’s τ), categorical variables (italicized) via ANOVA. “Plexus” = brachial plexus or thoracic outlet.

Preliminary Validation of the Block Building Task to Measure Hand Usage

To validate our Block Building Task, we compared it with the MAL. Affected hand usage correlated moderately between the two measures (Pearson r² = 0.325, p = 2.90×10⁻⁵; Figure 5).

Figure 5.

Moderate correlation between Block Building and Motor Activity Log measures of affected hand usage.

The Block Building Task provided a rapid measurement of hand usage. Participants completed the Block Building Task in 277 ± 126 seconds. 35/48 participants completed it in <5 minutes, and 46/48 completed it in <7 minutes.

Discussion

We quantified hand usage and non-dominant hand (NDH) compensation in individuals with unilateral upper extremity peripheral nerve injury (PNI). We found that participants used their dominant hand (DH) at a typical rate, regardless of which hand was injured – and regardless of which hand was more dexterous. In other words, patients did not compensate, even when compensation might benefit them. This suggests that some patients with chronic DH impairment could benefit from therapeutic approaches that encourage them to use their more-dexterous non-dominant hand (NDH).

Participants Favored the Dominant Hand Regardless of Injury Side and Dexterity

Patients with unilateral peripheral nerve injury primarily used their dominant hand, regardless of whether it was the affected hand, or whether it was currently (post-injury and -therapy) their more dexterous hand. This lack of compensation is unlikely to arise from inadequate time for patients to adapt, given the substantial time since injury (18.5 ± 26.6 months) and lack of correlation between compensation and time since surgery (p = 0.887). To our knowledge, this reflects the first quantitative measurement of hand usage and compensation (or lack thereof) for precision reach-to-grasp in patients with PNI: the Motor Activity Log (MAL) has not previously been used in the PNI population, and the Block Building task has not been used in any clinical population. A few studies have quantified compensation via accelerometry in stroke patients, who rarely return to use of the affected hand despite intensive therapy.^{31, 32} This common result, despite differences in clinical condition and methodology, imply that lack of compensation after unilateral injury may occur, at least in part, because hand preferences are difficult to change.³³

On average, our participants’ hand usage patterns followed the known rates for typical adults, though PNI patients showed greater variability than typical adults. This elevated variance seemed to arise from the small minority of participants who were entirely unable to use their affected hand. This matches our finding that hand usage was driven only by DH dexterity: our patients continued using their dominant hand if at all possible.

The stability of hand usage following unilateral PNI, regardless of any factor except DH dexterity, is a key finding for clinicians. Among patients with greater NDH dexterity (77% in our sample), many would benefit from compensatory therapy that teaches them to use their more-dexterous hand. Importantly, this is not true for all patients with greater NDH dexterity: for individuals who may recover DH function, they should be encouraged to use the impaired DH to avoid learned disuse.^34–36 However, learned disuse is not a problem for individuals whose injury is so persistent that they will never regain adequate function of the affected hand. While these irreversible injuries represent a minority of PNI patients (33–39%), that minority may encompass 55–65,000 people per year in the USA, who could benefit from rehabilitation that would encourage compensation with the NDH.^7–10 For these patients, preliminary studies indicate that targeted rehabilitation may allow them to improve performance and compensation with the NDH.^37–39

The Block Building Task Quantifies Actual Hand Usage

The Block Building Task provides a rapid, feasible method to directly quantify hand usage. Further validation is required in larger sample sizes, but Block Building Task results correlate with the self-report gold standard MAL. Further validation studies should continue to use the MAL, because traditional handedness measures have shown no correlation with hand usage.⁴⁰ Rapid objective measurement of hand usage will allow clinicians to avoid the inherent unreliability of self-report surveys.^{17, 18}

The moderate correlation between Block Building Task and MAL may arise from the weak relationship between self-report measures of hand preference and actual hand usage.⁴⁰ Moreover, the MAL does not ask about the unaffected hand; as a result, it cannot distinguish between compensation and decreased activity participation (both of which would result in lower usage of the affected hand). However, the MAL measures a wide range of activities,¹⁶ while Block Building evaluates only precision reach-to-grasp. In general, clinicians should be aware of the tradeoff between self-reported hand preference (which can rapidly cover a wide range of activities) and actual hand usage (more accurate), and select the appropriate tool for their needs.

Study Limitations

This was a cross-sectional study. We found no effect of therapy, but this effect should be confirmed in a longitudinal study that can evaluate the same participants pre/post therapy.

This was a single-site study. Results may differ in other sites that emphasize different rehabilitation interventions. The therapists in our outpatient hand rehabilitation clinic reported not using approaches or practice models that emphasized compensation. Such approaches exist,^{39, 41, 42} though evidence for their efficacy remains limited.^{37, 43}

This study was not designed to assess the effect of therapy. As a result, our conclusion about the ineffectiveness of these patients’ therapy should be taken with caution. Future studies should collect more detailed data on the nature of each patient’s current and past therapy.

This study was not designed with an a priori power analysis for its secondary hypothesis (compensation). However, Type II errors are more likely to support our main finding than contradict it: we found a non-significant trend only toward non-compensatory changes, i.e. atypical hand usage for patients with affected NDH rather than DH (Figure 3C). Nevertheless, trend analysis cannot replace a power analysis, and our findings should be replicated with a study powered to detect changes in compensation.

Conclusions

The Block Building Task demonstrates that individuals with unilateral peripheral nerve injury prefer to use their dominant hand even if it is injured -- and more importantly, even if that injury has left their dominant hand less dexterous than their non-dominant.

For many patients with unilateral impairment of the upper extremity, rehabilitation should involve restoring function of the affected hand. However, for the substantial minority whose dominant hand will never regain adequate function, compensation with the non-dominant hand may be key to regaining the functions of daily life. We found that many patients do not receive rehabilitation that helps them use their non-dominant hand more, which indicates a major avenue for improvement and expansion of rehabilitation after unilateral impairment.

Abbreviations:

DH

Dominant Hand

NDH

Non-Dominant Hand

PNI

Peripheral Nerve Injury

OT

Occupational Therapy

PT

Physical Therapy

MAL

Motor Activity Log

JTHF

Jebsen-Taylor Hand Function test