Impact of Handedness on Disability After Unilateral Upper-Extremity Peripheral Nerve Disorder

Abstract

Background: Impairment of the dominant hand should lead to greater disability than impairment of the nondominant hand, but few studies have tested this directly, especially in the domain of upper-extremity peripheral nerve disorder. The aim of this study was to identify the association between hand dominance and standardized measures of disability and health status after upper-extremity peripheral nerve disorder. Methods: An existing database was reanalyzed to identify the relationship between affected-side (dominant vs nondominant) on individuals with unilateral upper-extremity peripheral nerve disorder (N = 400). Primary measure of disability was the Disabilities of the Arm, Shoulder, and Hand (DASH) questionnaire. Results: We found no differences in standardized measures of disability or health status between patients with affected dominant hand and patients with an affected nondominant hand. However, a post hoc exploratory analysis revealed that patients with an affected dominant hand reported substantially reduced ability to perform 2 activities in the DASH questionnaire: “write” and “turn a key.” Conclusions: Following unilateral upper-extremity peripheral nerve disorder, impairment of the dominant hand (compared with impairment of the nondominant hand) is associated with reduced ability to perform specific activities, but this reduced ability is not reflected in standardized measures of disability and health status. To adequately identify disability following unilateral impairment of the dominant hand with the DASH, individual items must be used instead of the total score. New or alternative measures are also recommended.

Introduction

Peripheral nerve disorders (PNDs) are a prevalent cause of disability, with approximately 500 000 patients per year seeking surgical treatment in the United States.¹ Upper-extremity peripheral nerve disorders (UE-PNDs) lead to decreased sensory and motor function in the arm, shoulder, and hand^2-7 and are followed by disability^8,9 and reduced quality of life.^10,11 Many individuals with UE-PNDs are unable to return to work; 31% of patients fail to return to work within a year,¹² compared with 5% of patients with nonneural traumatic injuries to the hand.¹³ Inability to participate in meaningful activities (including work) is associated with depression and reduced quality of life.^14,15 However, UE-PNDs are understudied compared with other upper-extremity diagnoses,¹⁶ and the connection between sensorimotor impairments and disability in these patients remains unknown.

According to the biopsychosocial model of disability, disability arises when an individual’s needs (arising from his or her biological and psychological characteristics) exceed the options presented by the social and physical environment.^17,18 Therefore, while impairment is neither necessary nor sufficient to cause disability, impairments often increase an individual’s needs and can contribute to disability. As a result, the nature of impairment may affect a patient’s disability because different impairments will have different effects on a patient’s needs and their ability to participate in activities. In the context of UE-PNDs, we would expect disability to depend on whether the dominant hand or nondominant hand was affected by the PND.

Approximately 80% of patients seen for UE-PND in 2 large studies had unilateral involvement.^8,19 One of these studies reported patient hand dominance and found dominant hand involvement in 67% of patients.⁸ It is widely known that the dominant hand has greater strength²⁰ and dexterity,²¹ as well as greater abilities at limb trajectory dynamics²² and tool use.²³ Therefore, we could expect individuals with dominant hand injuries to have greater impairment than individuals with nondominant hand injuries. As a result of their impairment, individuals with UE-PND of the dominant hand may be forced to use the less-capable nondominant hand, which will reduce these individuals’ capability to perform activities. (We use the familiar language of hand dominance, but include PND that affects the entire upper extremity, because arm/shoulder impairments also affect hand function, eg, by preventing the patient from positioning their hand.) Use of the dominant hand is critical for many activities, although other activities are performed with the nondominant hand or either hand.^24-26 This is especially true of the right-handed 90% of the population²⁷ because their behavior is more strongly lateralized than left-handers.²⁸ However, “affected-side” effects (differences between individuals with affected dominant hand and individuals with affected nondominant hand) have not been found in the few studies of patients with UE-PNDs that reported hand dominance, whether they were measuring clinical outcome²⁹ or disability, work disability, and health status.⁸ However, these studies had several limitations that may have resulted in their inability to detect affected-side effects.

To identify affected-side effects specific to the dominant hand, studies must separate out unilateral versus bilateral, hand dominance, and affected-side. In the current study, we further analyzed the data from Stonner et al⁸ to address the role of those 3 factors. Our aim was to identify the association between hand dominance and disability and health status in patients with UE-PND. We hypothesized that patients with unilateral UE-PND to the dominant hand should show reduced ability to perform activities (measured as greater disability and lower health status), compared with patients whose PND affected the nondominant hand—in other words, an affected-side effect on disability and health status. In addition, we hypothesized that the affected-side effect would interact with 3 factors: duration of disability (a proxy for the patient’s adaptation to their injured state), diagnosis category (a proxy for function in the affected limb), and hand dominance. We also performed an exploratory analysis to examine the effects of hand dominance on specific activities in the DASH.

Materials and Methods

These analyses were performed on a preexisting database⁸ of 627 adult patients with confirmed UE-PND presenting to one hand surgeon for evaluation at an institutional referral center between December 2010 and October 2013. For the current study, participants were excluded if they had a bilateral PND (N = 156), if their bilateral/unilateral status was unknown (N = 2), if their hand dominance was unknown or self-described as “ambidextrous” (N = 12), or if they lacked a DASH score (N = 57), leaving the current study with 400 participants, all of whom had unilateral UE-PND, known handedness, and known DASH score. Table 1 shows diagnoses and demographics of patients in this sample. Sex and gender were not included in analyses because previous studies detected no effect of sex on outcome measures following UE-PND.⁸ This retrospective cross-sectional research was approved by the institutional review board of the Washington University School of Medicine. For access to the database, contact author Vicki Kaskutas. Our a priori analysis was preregistered via the Open Science Foundation at https://osf.io/psw4k/?view_only=96a77d9e5b5b45e8bb2c38b126077eb0.

Table 1.

Participant Diagnoses and Demographics.

Group	N	Demographics
Total	400
Median nerve disorder	57 (14%)	Age	48.6 ± 15.6 (18-85)
Ulnar nerve disorder	103 (26%)	Sex	171 females (43%)
Radial nerve disorder	36 (9%)
Proximal nerve injury	36 (9%)
Dual nerve compression	73 (18%)
Thoracic outlet syndrome	24 (6%)
Brachial plexus injury	71 (18%)

Three standard measures of disability and health status were used as outcome variables: (1) DASH, a 30-item measure of upper-extremity disability that assesses difficulty performing daily tasks, symptoms over the past week, and the impact of the injury on social activities, sleep, work, and self-esteem^30,31; (2) DASH work module (WDASH), a 4-item module that measures difficulty performing work-related tasks; and (3) the Short Form 8 (SF-8)³², a shortened version of the 36-Item Short Form Health Survey³³ that measures health status. The SF-8 is divided into a mental component score and a physical component score (PCS); only the PCS was used in this study. All 3 outcome variables were converted into a 0- to 100-point scale so that 100 represented poor status (ie, maximal disability or worst health status).

No other measures of disability or physical function were available in the database.

Of the 400 participants, all had DASH scores, 273 had WDASH scores, and 382 had PCS scores (ie, completed the SF-8). Each analysis only included participants with the relevant data, as visible in the degrees of freedom for each test. The different amounts of data did not affect analysis because multiple outcome measures were never combined into the same test.

The analyses below are described for a single outcome variable, but the process was repeated 3 times, once for each outcome variable, with appropriate Bonferroni correction (α = 0.017).

We followed a 3-step analysis procedure: univariate analysis on hypothesized variables, a stepwise multiple linear regression to detect factors associated with our outcome measure, and interaction analyses. Because of the lack of main effects, interaction effects were ultimately omitted, and thus are not described further.

Univariate analyses were performed via 1-way analysis of variance (ANOVA) of affected-side on outcome measure; for significant results, effect sizes were identified by Cohen’s d, as the difference in means divided by the pooled standard deviation.³⁴ Stepwise multiple linear regression was performed on the following candidate factors: (1) affected-side (dominant or nondominant); (2) factors expected to affect hypotheses: hand dominance (left or right), time since onset, and diagnosis category (7-level categorical converted into dummy variables); (3) number of prior surgeries; and (4) 2 to 8 confound variables, detailed below. Factors were entered into the model if the P value fell below .017 (.05/3 for Bonferroni correction) and were removed if the P value exceeded .033 (.10/3 for Bonferroni correction), except for affected-side, which was forced into the model.

Note that the database did not contain data on PND severity or injury type (eg, crush vs transection). The “diagnosis category” variable reflects our best available information about the nature of the injury.

We based our confounding factors on a list of items previously identified with this data set as contributing to each outcome measure.⁸ We used all factors from the list unless they were another outcome measure or if they had a specifically unilateral effect (e.g., right-hand grip strength). The resulting confounding factors for the DASH were depression, pain level, not working at present, difficulty sleeping, intimate relations affected, modified job demands, sudden-onset acute event, and stress at work. Confounds for the WDASH were intimate relations affected, others do household chores, reduced household chores, difficulty sleeping, and do household chores with pain. Confounds for the PCS were pain level and number of medications.

A sensitivity analysis was performed using G*Power 3.1.9.2³⁵ to identify the size of effect our analysis could detect. With an α of 0.017, power of 0.9, and groups of size 174 (nondominant hand affected) and 226 (dominant hand affected), the univariate analysis (a 2-tailed test for a difference in means) had sufficient sensitivity to detect an effect of d = 0.371, an effect size midway between small (0.2) and medium (0.5).³⁴

All other statistical analyses were performed using MATLAB (version 9.1.0; MathWorks Inc., Natick, MA).

Secondary Survey: Exploratory Study of DASH Questions

Based on the results of the post hoc analysis, a separate survey was performed to test the authors’ opinions about which DASH activities were unilateral versus bilateral and how much each activity would be affected by using the left versus right hand (ie, how hand-specific it was). This was a rapid post hoc survey, so no power analysis was performed. No demographic information was collected on the 19 adult participants, all of whom were associated directly or indirectly with the Program in Occupational Therapy at the Washington University School of Medicine.

See Supplemental Materials for detailed survey methods. In brief, the participants were asked 2 questions about each of the 21 activities in the DASH survey: whether the activity was unilateral versus bilateral, and how much each activity would be affected by using the left versus right hand (ie, how hand-specific it was). Each survey question was assessed on a 5-point Likert scale. For each activity in the DASH, we calculated whether the participants rated it as more unilateral or hand-specific than the other 20 activities.

Results

Effect of Affected-Side on Primary Outcome Measures

Overall and individual-group mean ± standard deviation outcome scores are shown in Table 2. We found higher disability for “dominant hand affected” (46.2 ± 24.4) than for “non-dominant hand affected” (41.3 ± 23.0), but this difference was not statistically significant. Our univariate analyses found no effect of affected-side on any of our outcome measures: a nonsignificant trend for DASH (F_1,398 = 4.196, P = .041) and no effects for WDASH (F_1,271 = 1.85, P = .175) or PCS (F_1,386 = 1.10, P = .295).

Table 2.

Outcome Measures (Mean ± Standard Deviation).

Group	N	DASH	WDASH	PCS
Total	400	44.1 ± 23.9	53.6 ± 32.1	43.3 ± 20.4
Dominant hand affected	226	46.2 ± 24.4	56.0 ± 32.0	42.3 ± 20.7
Nondominant hand affected	174	41.3 ± 23.0	50.6 ± 32.2	44.5 ± 19.9
RH dominant	344	44.4 ± 23.7	55.1 ± 31.6	43.3 ± 20.3
LH dominant	56	42.2 ± 24.9	44.8 ± 34.4	43.0 ± 21.3
Bilaterala	156a	41.1 ± 22.4	46.7 ± 29.1	45.3 ± 20.1

Note. No significant differences between groups. DASH = Disabilities of the Arm, Shoulder, and Hand; WDASH = DASH work module; PCS = physical component score; RH = right hand; LH = left hand.

^a“Bilateral” = the 156 participants with bilateral upper-extremity peripheral nerve disorder, excluded from all other analyses in this article, shown here for comparison.

To confirm that our sample of patients with unilateral UE-PND showed similar levels of disability and health status compared with other (ie, bilaterally impaired) patients with UE-PND, we compared outcome variable means against the mean of the 156 patients with bilateral UE-PND who were excluded from all other analyses. We found no statistically significant differences between patients with unilateral and bilateral UE-PND in any outcome measure (P > .05).

Stepwise multiple linear regression revealed that our hypothesized variables (affected-side, hand dominance, time since onset, diagnosis category) had no effect on outcome measures. As shown in Table 3, only confound variables were included in the regression models for each outcome measure. Affected-side is included in Table 2 because it was forced into the model, but its contribution was not statistically significant for any outcome variable: neither DASH (β = 4.127, P = .0963), WDASH (β = 4.075, P = .219), nor PCS (β = −0.377, P = .849).

Table 3.

Stepwise Multiple Regression Results for Each Outcome Variable, Showing All Variables Included in Model.

Stepwise multiple linear regression factors
Outcome variable	Factor	Hypothesized or confound?	β coefficient	P value
DASH	Affected-side	Hypothesized	4.127	.0963a
	Pain level	Confound	0.190	5.09 × 10−5
	Not working	Confound	25.056	3.22 × 10−8
	Intimate relations	Confound	15.495	2.21 × 10−8
	Stress at work	Confound	0.142	.00035
WDASH	Affected-side	Hypothesized	3.675	.275a
	Intimate relations	Confound	16.721	1.58 × 10−5
	Others do chores	Confound	52.040	8.58 × 10−12
	Reduced chores	Confound	37.717	8.30 × 10−9
	Do chores with pain	Confound	17.389	.0036
PCS	Affected-side	Hypothesized	−1.526	.460a
	Pain level	Confound	0.313	3.57 × 10−15
	No. of medications	Confound	1.488	2.08 × 10−7

Note. Affected-side (^a) was forced into all models, but was never a statistically significant factor (α = 0.017). DASH = Disabilities of the Arm, Shoulder, and Hand; WDASH = DASH work module; PCS = physical component score.

Because we found no main effects of affected-side, we did not perform interaction analyses. However, because the DASH showed a nonsignificant trend, in addition to being our most complex outcome measure (30 questions, vs 4 each for WDASH and PCS), we analyzed the DASH survey in more detail via a post hoc exploratory analysis.

Exploratory Analysis: Affected-Side Influence on Individual DASH Activities

We performed a post hoc exploratory analysis to identify whether any individual DASH survey questions showed an “affected-side” effect. Thirty 1-way ANOVAs were performed: each one tested the effect of affected-side on 1 DASH question (Bonferroni-corrected to α = 0.0017).

We found significant effects of affected-side on 2 questions of the DASH: “write” (F_1,412 = 103.236; P = 8.60 ×10⁻²², Cohen’s d = 0.93) and “turn a key” (F_1,416 = 29.748, P = 8.46 × 10⁻⁸, d = 0.54). We also found nonsignificant trends (P above our threshold but below .05) for 5 questions: “push open a heavy door” (F_1,419 = 5.680, P = .018), “carry a shopping bag or briefcase” (F_1,416 = 5.372, P = .021), “wash or blow dry your hair” (F_1,411 = 5.355, P = .021), “recreational activities with force/impact” (F_1,403 = 3.969, P = .047), and “recreational activities with free movement” (F_1,401 = 7.840, P = .005). In all of these cases, mean disability scores were higher for the “dominant hand affected” group than the “nondominant hand affected” group.

Figure 1 compares “dominant hand affected” versus “nondominant hand affected” participants for the 21 activity questions in the DASH. The 9 nonactivity questions (omitted from figure for brevity) showed no trends or significant effects.

Figure 1.

Exploratory analysis of the 21 Disabilities of the Arm, Shoulder, and Hand activities, showing significant affected-side effect for “write” and “key” only. Mean ± standard deviation. * = statistically significant (α = 0.0017); † = nonsignificant trend (P between 0.0017 and 0.05).

Note. DH = dominant hand; NH = nondomimant hand.

Secondary Survey: Nature of DASH Activities

To test our intuitions about what distinguished “write” and “turn a key” from other DASH questions, we performed a secondary survey to assess which DASH activities were unilateral versus bilateral and how much each activity would be affected by using the left versus right hand (ie, how hand-specific it was).

The survey participants rated 5 of the DASH activities as significantly more unilateral than the rest: “write” (z = −3.678, P = 1.17 × 10⁻⁴, d = 1.18), “turn a key” (z = −3.902, P = 4.76 ×10⁻⁵, d = 1.66), “place an object overhead” (z = −3.267, P = 5.4 × 10⁻⁴, d = 0.75), “carry a bag or briefcase” (z = −3.865, P = 5.6 × 10⁻⁵, d = 1.44), and “change a light bulb overhead” (z = −3.492, P = 2.4 × 10⁻⁴, d = 0.93), as shown in Figure 2. The 2 activities identified in our exploratory analysis, “write” and “turn a key,” were, respectively, rated as the third and first most unilateral activities in the DASH.

Figure 2.

The 21 Disabilities of the Arm, Shoulder, and Hand activities, rated for unilaterality (“would most people do this with 1 hand or 2?”) in post hoc survey. Mean ± standard deviation. Highlighted bars indicate the activities identified in exploratory analysis (Figure 1). Significance tests (Wilcoxon signed rank) indicate answer greater than all other questions, α < 0.0017; all others P > 0.05.

For our second survey question (hand specificity), “write” was significantly more hand-specific than most activities (z = −3.902, P = 4.78 × 10⁻⁵, d = 1.21; the first most hand-specific activity), but “turn a key” was not (z = −0.765, P = .22, d = 0.10; the eighth most hand-specific activity). The other hand-specific activities were “prepare a meal” (z = 2.987, P = 7.06 × 10⁻⁴, d = 0.51), “use a knife to cut food” (z = −3.192, P = 7.06 × 10⁻⁴, d = 0.59), “recreational activities with force/impact” (z = −3.827, P = 6.49 × 10⁻⁵, d = 1.09), and “recreational activities with free movement” (z = −3.790, P = 7.54 × 10⁻⁵, d = 0.99).

Overall, “write” and “turn a key” were among the most strongly lateralized among the 21 DASH activities, and “write” was the activity most dependent on use of the dominant hand.

Discussion

We compared physical disability and health status between UE-PND patients with an affected dominant hand and patients with an affected nondominant hand. Contrary to our hypotheses, we found no effect of which side was affected by their UE-PND. However, this lack of affected-side effect seems to arise because dominant hand impairment is followed by reduced participation in specific unilateral activities, which are not captured by standard measures of disability.

The DASH is one of the most widely used standardized outcome measures of upper-extremity overall disability and (via its WDASH module) work disability; as of April 2018, the original publication³¹ has been cited over 3500 times and is regularly used to assess patients with upper limb injury.³⁶ The SF-8 is a widely used survey of health status, with over 900 citations of its manual.³² However, neither of these measures were able to detect an effect of affected-side on outcome in our database of UE-PND patients.

This lack of affected-side effect could occur if disability is not strongly driven by impairment of only 1 hand, ie, a patient with unilateral impairment is unlikely to experience disability. However, our data are inconsistent with this hypothesis because our patients with unilateral UE-PND showed similar levels of disability and health status compared with patients with bilateral UE-PND. Moreover, our average DASH score of 44 ± 24 was consistent with the established mean of 44 ± 22 for all UE-PND patients³⁷ and well above the healthy adult mean of 10 ± 15.³⁸ Therefore, our patients showed substantial disability despite the unilateral nature of their UE-PND.

It is also possible that the limited size of our database concealed a between-group difference. This is supported by the trends toward effects of affected-side on DASH score, although of course trends cannot prove what would happen with a larger sample size. Regardless, this does not affect our core conclusion: These 3 standard measures of disability and health status are relatively insensitive to the unilateral nature of deficits following PND to the dominant hand.

Adaptation provides another possible explanation for our lack of affected-side effects. Our participants could have fully adapted to the dominant hand–specific aspects of their impairment, whether from nondominant hand compensation, adaptive equipment, therapy, and so on. However, our regression analysis found no relationship between disability/health and time since onset, which indicates that at least on a group level, adaptation did not drive individuals’ current level of disability/health. Other adaptation-related variables (eg, time since surgery) could potentially play a role, but those data were not included in our database.

Our exploratory analysis indicated clear associations between affected-side and 2 DASH questions: “write” and “turn a key.” These appear to be activities that uniquely require unimanual precision movement (compared with other DASH activities), as confirmed by our post hoc survey. The lateralized nature of these 2 activities is likely a major reason why these particular activities were impeded more heavily after dominant hand impairment.

The DASH was not designed to detect unilateral disability. Its original design included a focus on 2 concepts (symptoms and functional status), 3 dimensions (physical, social, and psychological), and 18 components (eg, daily activities, self-care, socializing).³¹ Under this framework, conceptual item-reduction techniques considered our 2 unilateral activities (“write” and “turn a key”) as conceptually redundant with other physical activity questions.³⁹ Moreover, the scores of our 2 unilateral activities correlated poorly with overall score,³⁹ which indicates that they capture a characteristic that is not otherwise well represented in DASH total score.

One important caveat is that our data did not include the severity of injury or physical impairment. It is possible that one group (dominant hand vs nondominant hand impaired) was more severely injured than the other. We cannot rule this out, although we have no reason to expect a systematic difference between patients with dominant hand and nondominant hand impairment.

Conclusions

For some specific activities, unilateral UE-PND patients with dominant hand impairment have greater disability than patients with nondominant hand impairment. However, because unilateral impairment can spare many other activities, standard measures of disability and health status do not capture these specific consequences of dominant hand impairment. To better characterize disability in patients with unilateral impairment, the rehabilitation research community must use and develop additional methods to identify and quantify disability in unimanual activities.

We recommend that clinicians and therapists recognize that the DASH score does not assess the consequences of unilateral impairment. Disability and health status after unilateral impairment can instead be assessed via individual items of the DASH questionnaire, or potentially the Michigan Health Questionnaire.⁴⁰ Ideally, we recommend development of new activity surveys designed specifically around unilateral deficits and for future studies of patients with unilateral impairment to report the effects of hand dominance. In addition, future studies could investigate the connection (and potential causal links) between unilateral impairment and disability, because this knowledge could connect neurophysiological and behavioral measurements (ie, of impairments) with patient-centered measures (ie, disability and health status) and thereby allow the development of impairment-level interventions that may reduce patient disability.