Abstract
Keywords
Arm and hand function have been found to decline with age (Carmeli, Patish, & Coleman 2003; Desrosiers, Hébert, Bravo, & Rochette, 1999) as the result of sensorimotor impairments such as decreased motor coordination (Verkerk, Schouten, & Oosterhuis, 1990), decreased manual dexterity (Desrosiers et al., 1999; Mathiowetz, Volland, Kashman, & Weber, 1985), and reduced grip strength (Desrosiers et al., 1999; Desrosiers, Bravo, Hébert, & Dutil, 1995; Jansen et al., 2008; Rantanen, Era, & Heikkinen, 1997). Older adults’ arm and hand function also is reduced because of impairments related to common diseases in this population, such as rheumatoid arthritis and osteoarthritis (Helmick et al., 2008); fractures; and neurological conditions, such as stroke (Rosamond et al., 2007). Although specific assessments of arm and hand function have been established (e.g., the Disabilities of the Arm, Shoulder, and Hand outcome questionnaire, Solway, Beaton, McConnell, & Bombardier, 2002, and the Box and Block Test, Mathiowetz et al., 1985), the extent to which healthy older adults use their hands in daily activities is unknown. Factors such as grip strength and dexterity may influence arm–hand use, as may hand dominance, age, gender, and previous vocation. Although hand strength and dexterity limitations are known to affect activities of daily living (ADLs; Flunn, Trombly-Latham, & Podolski, 2007), what remains unknown is whether the extent of hand use affects occupational performance.
Accelerometers, which measure the extent and intensity of acceleration (movement), are a relatively novel way to monitor arm–hand use. Traditionally, accelerometers have been used to monitor mobility and walking. The reliability and validity of accelerometers for the upper extremity have been established (de Niet, Bussmann, Ribbers, & Stam, 2007; Uswatte et al., 2000, 2006; Uswatte, Foo, et al., 2005; Vega-González & Granat, 2005), and accelerometers have been found to provide an objective way to assess participants’ real-world upper-extremity function outside the laboratory. This finding is especially valuable for people with stroke, who often experience “learned nonuse” of their weaker upper extremity. Learned nonuse is a well-known poststroke phenomenon that can be described as the disparity between a person’s motor ability to use the weaker hand and his or her actual use (Taub, 1980). Recently, accelerometers have been used to monitor both hands of two groups of people with stroke (N = 169) to measure patient compliance with restraint of the unaffected upper extremity and the arm use of the affected upper extremity before and after participating in constraint-induced movement therapy (Taub & Uswatte, 2003; Uswatte et al., 2006; Winstein et al., 2003). In addition, recovery and hand use may be linked; a recent study found that people with stroke who rarely used their paretic arm were found to have increased activation of secondary motor areas (e.g., contralesional motor regions) compared with an age-matched control group, possibly as a mechanism of compensation (Kokotilo, Eng, McKeown, & Boyd, 2010).
To date, occupational therapists have no objective measure of the extent to which clients use their affected upper extremity outside the clinical setting. The Motor Activity Log captures the amount and quality of arm–hand use at home but relies on self-report regarding 14 specific tasks rather than real-time objective measures (Uswatte, Taub, Morris, Vignolo, & McCulloch, 2005). Significant high correlations (rs = .81–.90) between the Motor Activity Log and the accelerometer reading of the unimpaired arm, impaired arm, and ratio of the use of both arms of people with stroke, however, have been reported (Uswatte et al., 2006).
Greater hand use in the dominant than in the nondominant hand has been found during ADLs in healthy young adults. Vega-González and Granat (2005) assessed hand use by means of an electrohydraulic activity sensor attached to the shoulders and wrists of healthy young adults performing their normal daily activities over the course of 8 hr. They reported that the participants’ dominant arm was 19% more active than the nondominant arm. De Niet et al. (2007) used a combined electrogoniometric–accelerometric system for 12 hr to monitor the upper extremity of 5 healthy participants and found greater activity for the dominant hand than for the nondominant hand.
By contrast, healthy older adults’ arm–hand use may be more bilateral. Kalisch, Wilimzig, Kleibel, Tegenthoff, and Dinse (2006) assessed arm–hand use in adults from three age groups (13 adults with a mean age of 25, 9 adults with a mean age of 50, and 14 adults with a mean age of 70) using accelerometers. Over several hours, adults from the oldest group used both hands with equal frequency; however, the younger adults used the dominant right hand more often than they did the nondominant left hand. Lang, Wagner, Edwards, and Dromerick (2007) and Kilbreath and Heard (2005) reported similar findings with healthy older adults.
One of the possible confounding factors in assessing arm–hand use is the influence of gender. Gender differences are present in manual abilities, and other factors, such as social and cultural background, may have an effect on arm–hand use, especially during instrumental activities of daily living (IADLs). Many IADL tasks that require arm–hand use are traditionally considered to fall into a woman’s role (Allen, Mor, Raveis, & Houts, 1993; Asberg & Sonn, 1989). Older men, although physically capable of doing IADL tasks such as meal preparation and laundry, may rely on their spouses or others to do these tasks for them (Asberg & Sonn, 1989; Koyano et al., 1988). Therefore, the influence of gender on arm–hand use, especially in older adults, is important to investigate.
ADLs, IADLs, education, work, play, leisure, and social participation all are areas of occupational performance (American Occupational Therapy Association, 2008). Occupational performance consists of an objective, observable component and a subjective component, both of which must be captured (McColl & Pollock, 2005). Accelerometers can enable an objective measure of the extent of occupational performance at home, where the occupational therapist is not present. This real-life measure of hand use in conjunction with findings from the clinical assessments will enhance understanding of the factors that relate to hand use. In the future, data from accelerometers may be able to demonstrate how diverse impairments can influence hand use and occupational performance. In addition, an accelerometer could be used as a clinical tool by enabling therapists to monitor the amount of clients’ hand use outside of therapy sessions and thereby have a positive effect on clients’ adherence to activity-based homework programs as recommended in task-based training programs to enhance motor skills in patients with mild to moderate hemiparesis after stroke (e.g., Bass-Haugen, Mathiowetz, & Flinn, 2007).
To date, no normative data exist on the extent of older adults’ arm–hand use during daily activities. Obtaining this important information will enable future comparisons of arm–hand usage of older adults, who commonly experience impairment of the upper extremities, such as hemiplegia, as a result of stroke. Additionally, hand strength and dexterity limitations have an impact on ADLs (Flunn et al., 2007), but whether the extent of hand use has an impact on occupational performance remains unknown. Studies have demonstrated that increased practice using the weaker upper extremity after stroke is beneficial for improving hand function (e.g., Barreca, Wolf, Fasoli, & Bohannon, 2003). Obtaining normative data can also guide treatment to increase arm–hand use and enhance occupational performance in clients.
Our study’s primary objective was to quantify hand use (measured by accelerometers for multiple days) of older men and women without a disability and determine the effect of hand dominance, gender, and day (Day 1–Day 5) on hand use. Our second objective was to use linear multiple regression to determine factors (e.g., age, dexterity, hand strength) that predict hand use. Understanding the factors that contribute to hand use will be useful for occupational therapists aiming to increase hand use and promote occupational performance.
Method
Population
Forty community-living older adults (20 men and 20 women) participated in the study. Inclusion criteria were as follows: Participants were (1) ages 65–80; (2) right handed, as assessed by the Edinburgh Handedness Inventory (Oldfield, 1971); and (3) fully capable of using both upper extremities. In addition, all participants were retired, so as to avoid any influence of vocation on arm–hand use. Exclusion criteria were (1) neurological or psychiatric condition; (2) impairments, such as peripheral neuropathies, osteoarthritis, or rheumatoid arthritis, that limited use of the hands; (3) upper-extremity fracture sustained within the past year; and (4) not independent in basic ADLs and IADLs.
The study was advertised at community and shopping centers. Forty-three people expressed a desire to participate in the study; however, 3 were excluded because they did not meet inclusion criteria (2 were left handed; 1 had rheumatoid arthritis). The study was approved by the local university ethics board, and all eligible participants gave written informed consent before participating. All were provided with an honorarium for their participation in the study.
Sample Size Justification
We calculated sample size using G*Power 3.0 software (Buchner, Erdfelder & Faul, 1997); the study was designed to provide sufficient power for three factors (hand [dominant or nondominant], day, and gender) using a three-way analysis of variance (ANOVA) with a moderate effect size of .30, an α of .05, and a power of .80.
Instruments
Accelerometers (Actical; Mini Mitter Co., Bend, OR) quantified the participants’ arm use (the amount and intensity of arm activity) using the mean total activity kilocounts per day over 7 consecutive days. The triaxial Actical accelerometer is a small (28 mm × 27 mm × 10 mm), lightweight (17 g), waterproof device that has a frequency range of 0.3–3 Hz. The unit is sensitive to 0.05–2.0 G-force and samples data at 32 Hz. Acceleration is detected in all three planes, although more sensitivity is present in the vertical plane. When worn on the hip, the accelerometer has a step-count function. The accelerometer record is rectified and integrated over the specified window (15 s) as activity counts; 1 kilocount is equal to 1,000 activity counts. Thus, higher activity counts occur with longer use (i.e., time), more movements (e.g., raking), and greater intensity of movement. The Actical accelerometer was found to be superior to two other commonly used accelerometers—Actigraph (ActiLife Co., Pensacola, FL) and RT3 (Stayhealthy, Inc., Monrovia, CA) for intrainstrument and interinstrument reliability (Esliger & Tremblay, 2006).
A limitation of accelerometers is that they cannot detect differences between a functional task (e.g., eating) and a nonfunctional task (e.g., moving the arm up and down). Also, sustained muscle functions that do not incorporate movement (e.g., holding an object in the hand without moving it) cannot be detected. The validity of accelerometers has been tested by correlating the accelerometer readings of 34 people with acute stroke with upper-extremity clinical assessments of function (rs = .40–.62, p < .01; Lang et al., 2007). Pretreatment to posttreatment changes in patients’ upper-extremity movement scores have been shown to strongly correlate with accelerometer readings (r = .91, p < .01; Uswatte, Taub, et al., 2005), and strong correlations have been established between accelerometer readings and observer ratings of arm activity in people with stroke (r = .93, p < .01; Uswatte, Taub, et al., 2000). Accelerometers’ discriminative validity has been established by comparing affected upper-extremity use of people with stroke with upper-extremity use of healthy individuals (p < .01; de Niet et al., 2007; Lang et al., 2007). Accelerometers were also found to be a sensitive instrument for revealing differences between affected and nonaffected arm use in people with stroke (p < .01; de Niet et al., 2007; Vega-González & Granat, 2005).
We used the Box and Block Test (Cromwell, 1965) as a test of manual dexterity, defined as the ability to make skillful, controlled arm–hand manipulation of larger objects (Mathiowetz & Bass-Haugen, 2007). The test requires participants to transfer as many blocks as they can, picking up one at a time, from one side of a box over a divider to the other side. The number of blocks transported from one side of the box to the other in 1 min is recorded. This instrument is a reliable and valid test for assessing dexterity in people >60 yr for the right (interclass correlation coefficient [ICC] = .97) and left (ICC = .96) hands (Desrosiers, Bravo, Hébert, Dutil, & Mercier, 1994) and has norms for age, gender, and hand dominance (Mathiowetz et al., 1985).
Grip strength was assessed with the Jamar Dynamometer (Sammons Preston Evaluation Equipment, Bolingbrook, IL). Each hand was assessed 3 times in a standardized seated posture (Fess, 1992) with the dynamometer handle in the second position. The mean of the three trials was recorded in kilograms. This test is reliable and valid (r > .08, p < .01) for assessing manual grip in healthy and hand-injured populations (Bohannon & Schaubert, 2005; Mathiowetz, Weber, Volland, & Kashman, 1984) and has norms for age, gender, and hand dominance (Jansen et al., 2008).
Procedure
Participants completed a demographic questionnaire that assessed their performance of IADL tasks (e.g., shopping, laundry, driving). The questionnaire was based on Lawton and Brody’s (1969) IADL questionnaire (see also Lawton, Moss, Fulcomer, & Kleban, 1982). Immediately afterward, the Box and Block Test and grip strength assessments were performed. The order of the two assessments was counterbalanced to eliminate possible fatigue.
Participants were provided instruction in the use of the accelerometers. All were given two accelerometers to be worn on each wrist. A third accelerometer was worn on a belt over the right anterior superior iliac spine to discriminate between arm activity that occurred while walking and activity that occurred during other tasks (Figure 1). Participants were requested to wear the accelerometers during all waking hours for 7 consecutive days beginning the next morning. Participants were reminded to disregard the testing equipment and to go about their normal daily activities. When the participants returned the accelerometers, they were asked whether they had worn them for all 7 days. In addition, they were also asked whether they had slept with the accelerometers on and whether any problems were encountered. This information was further verified with the data downloaded from the accelerometers.
Accelerometer configuration. A small accelerometer with a hook-and-loop watch strap was worn on each wrist. The right accelerometer was marked with the letter R, and the left accelerometer was marked with the letter L. A third accelerometer was worn on a belt over the right anterior superior iliac spine.
Statistical Methodology
The activity kilocounts from the accelerometers over the time period were downloaded to a computer. To reveal a more “functional” measure of hand use, we eliminated the activity kilocounts of arm swing while walking. Hand use was calculated as the total activity kilocounts of the hands minus the activity kilocounts of the hands performed simultaneously while walking (when at least five consecutive steps were taken; measured by the step-count function). This procedure provided us with functional arm–hand use for the waking hours of 5 consecutive days (the step count functions for 5 days only). Participants started wearing the accelerometers (Day 1) on different days (between Tuesdays and Saturdays).
We used descriptive statistics to characterize the two groups. We used t tests for independent samples to assess the group differences between men and women on age, years of education, and years since retirement and the Mann–Whitney U test (Portney & Watkins, 2009) to assess differences between the groups (men and women) for dichotomous variables. Descriptive statistics were used to generate the means and standard deviations (SDs) activity kilocounts per day for each hand.
To address our first objective of quantifying hand use and determining the effects of hand dominance, gender, and day on hand use, we performed a three-way ANOVA to assess the within-group factor of hand dominance (right dominant or left nondominant), between-groups factor of gender (men or women), and within-group factor of day (Days 1–5). For the second objective of quantifying determinants of hand use, we used two multiple linear regressions (mean daily activity kilocounts) for the right dominant and left nondominant hands after accounting for age and gender. To determine entry into the regression model, we used Pearson correlation coefficients to assess the relationships between hand use and dexterity and grip strength of the dominant and nondominant hands. Correlations ranging from .25 to .50 were considered fair; those ranging from .50 to .75 were considered moderate to good (Portney & Watkins, 2009). Variables with significant correlations (p < .05) were entered into the model.
Results
The sample’s demographic information is presented in Table 1. No significant differences were found between the groups for age, education, and time since retirement. Significantly more women than men reported they cooked (z = −2.4, p < .01) and washed and ironed clothes (z = −0.59, p < .002). Significantly more men performed home repair activities (z = −2.1, p < .03) and drove a car (z = −2.6, p < .009).
Participants’ Demographic Information
Note. SD = standard deviation.
Significant difference between men and women (p < .05).
All the participants wore the accelerometers for their waking hours during 7 days; some participants did not remove the accelerometers for the duration. No technical problems were encountered, and no one reported discomfort from the accelerometers. The mean activity kilocounts per day was 164.9 (SD = 76.9) for the men’s right hand and 193.6 (SD = 120.1) for their left hand. The mean activity kilocounts per day for the women was 224.3 (SD = 111.8) for the right hand and 177.7 (SD = 116.5) for the left hand. To determine the effect of hand dominance, gender, and day on hand use, we performed a three-way ANOVA. We found no main effect for hand, gender, or day but did find a significant Hand × Gender interaction effect (F[1, 190] = 24.4, p < .001; Figure 2). Women used the dominant hand 26% more than men. On average, women used the right hand 21% more than the left hand; interestingly, men used the right hand 15% less than the left hand.
Interaction between right dominant and left nondominant extent of hand–arm use (activity kilocounts) and gender.
We found a significant moderate correlation between increased use of the right hand (accelerometer activity kilocounts; r = .53, p < .001) and increased number of blocks transferred in the Box and Block Test. We found a fair correlation (r = .34, p < .001) for the left hand between those variables. We found a fair, significant correlation between hand use and grip strength only for the right hand (r = .33, p < .001). After adjusting for age and gender, the manual dexterity (Box and Block Test score) accounted for 18% of the total variance in right- hand use using linear regression (Table 2). The total variance (35%) in the final model (age, gender, and dexterity) was accounted for by right-hand use. Age, gender, dexterity, and grip strength were not found to significantly predict left-hand use.
Linear Regression Model Summary for Extent of Right Dominant Arm–Hand Use
Note. Grip strength was not a significant determinant (p = .552) and was removed from the model.
R 2 is the coefficient of determination and is the proportion of variability in a data set that is accounted for by the statistical model.
The standardized β coefficients represent the change in terms of standard deviations in the dependent variable that result from a change of 1 standard deviation in an independent variable.
Discussion
The dominant and nondominant arm–hand use of 40 older adults was quantified using triaxial accelerometers over 5 consecutive days. Normal arm-swing movements during walking were eliminated; thus, the activity kilocounts captured functional use of the hands.
Although both men and women were characterized as right handed in terms of ability and functional use, women demonstrated a significant preference for using the dominant hand, whereas men used their hands more bilaterally. For completing everyday nonvocational activities, women used their dominant hand 26% more than men used their dominant hand. On average, women used the right hand 21% more than the left hand. Surprisingly, men used the right hand 15% less than the left hand.
The Gender × Hand Dominance interaction effect has not been previously explored or analyzed and may explain why some studies revealed greater use of the dominant hand, and other studies found equal use of both hands. In this study, we identified women’s (Allen et al., 1993 Asberg & Sonn, 1989; van Heuvelen, Kempen, Brouwer, & de Greef, 2000) and men’s (Jenson, Suls, & Lemos, 2003) traditional roles. Cooking, washing, and ironing clothes, which are frequently done daily by women, possibly influenced the women’s arm–hand use, causing the asymmetry with greater dominant right-hand use. By contrast, yardwork repairs and carrying, which were performed more often by men (but may be carried out less frequently), may include more bimanual activity that requires using the dominant hand and the nondominant hand simultaneously. More so, women may tend to perform dexterous activities that more frequently incorporate the use of the dominant hand; men tend to perform strength-related activities that more often incorporate the nondominant hand.
The manual dexterity of the right hand accounted for 18% of the total variance of right arm–hand use after controlling for age and gender. This one-time performance on the Box and Block Test predicted the extent of arm–hand use in the real-world setting over 5 days. Although manual dexterity has been found to be a strong predictor of functional independence and disability in basic ADLs and IADLs (Ostwald, Snowdon, Rysavy, Keenan, & Kane 1989; van Heuvelen et al., 2000; Williams, Hadler, & Earp, 1982), our study extends those findings to dominant arm–hand use. Because our study was cross-sectional, we were not able to determine causation. It is possible that older adults who frequently use their hands develop better dexterity, or alternatively, that those who have better dexterity use their hands more frequently. The task of transferring blocks is similar to many daily activities such as packing and unpacking the dishwasher or dealing cards, both of which are usually performed with the dominant hand. Interestingly, nondominant arm–hand use could not be predicted by dexterity and grip strength.
Although grip strength has been found to be related to several important variables, including mortality (Sasaki, Kasagi, Yamada, & Fujita, 2007) and frailty (Syddall, Cooper, Martin, Briggs, & Aihie-Sayer, 2003), it was not found to be a determinant of hand use. One can assume that this pattern exists because maximum force is typically not required in everyday activities.
Although differences in hand use have been found between young and older adults (Kalisch et al., 2006), we did not find age to be predictive of hand use. Our study sample included retired older adults ranging from age 65 (retirement age) to age 78, which is just within the average Canadian life span (Statistics Canada, 2007). Despite this 13-yr age range, we found no relationship between age and arm–hand use. Thus, differences in hand use among young and older adults may result from the groups’ distinct activities (e.g., vocation), although in our cohort, all participants were retired. Future studies should include young adults to assess differences in hand use across age groups.
Conclusion
In this study, we demonstrated that by using accelerometers, one can quantify the extent of arm–hand use during older adults’ daily activities. The accelerometers add a new dimension to hand function; the devices can potentially enable occupational therapists to obtain information about the extent of clients’ hand use outside of clinical settings. This valuable information, along with data from traditional clinical assessments, can enhance the understanding of how different impairments may affect older adults’ arm–hand use. Such information can also be used to assess intervention effects.
This study’s findings also determined that both hands may be used to a similar extent for daily function. Awareness of these findings might lead clinicians to focus interventions on both hands equally, not predominantly on the dominant hand. When possible, clients should be encouraged to use both hands for daily function and not rely on one hand using compensatory techniques.
Accelerometers would be beneficial to use with diverse upper-extremity conditions common in older adults to measure arm–hand use. To date, several studies aiming to quantify the hand use of people with stroke have been completed (de Niet et al., 2007; Uswatte et al., 2000, 2006; Uswatte, Foo, et al., 2005, Uswatte, Taub et al., 2005). To establish characteristic arm–hand use for special populations, additional studies with common upper-extremity conditions that affect older adults are warranted. Finally, longitudinal studies that focus on changes in hand use resulting from normal aging or recovery from injury should be pursued.
Footnotes
Acknowledgments
We thank the participants for their contributions to the study and acknowledge the support of BC Medical Services Foundation (BCM08–0098 to Janice J. Eng and Debbie Rand), postdoctoral funding (to Debbie Rand) from the Heart and Stroke Foundation of Canada, Canadian Stroke Network, Canadian Institutes of Health Research/Rx&D Collaborative Research Program with AstraZeneca Canada Inc., and career scientist awards (to Janice J. Eng) from the Canadian Institutes of Health Research (MSH–63617) and the Michael Smith Foundation for Health Research.