Evaluation of Hand Forces During a Joint-Protection Strategy for Women With Hand Osteoarthritis

Design

A 2 × 2 × 2 experimental cross-sectional design was used to investigate within-subject differences in peak grip force profiles, integral of the grip force (IGF; i.e., grip force across time, or the area under a force–time curve), and torque when participants attempted to open a sealed jar. Three variables were studied: (1) the approach to grasping the jar, (2) the hand used to twist the lid, and (3) the use of a nonskid material to twist the lid. This study was approved by the University of Minnesota’s institutional review board (No. 0908M71484), and all participants provided informed consent before taking part in the study.

Participants

Participants were recruited through orthopedic, women’s health, and hand therapy clinics as well as community-based centers serving older adults. Recruits were eligible if they were female, older than age 18 yr, and met one of the following two criteria: (1) had radiographically confirmed hand OA or (2) had a combination of self-reported physician diagnosis and symptoms (i.e., achiness and stiffness) of hand OA (Szoeke et al., 2008). Recruits were excluded if they had one or more of the following seven characteristics: (1) an upper-extremity (UE) movement disorder (e.g., Parkinson’s disease, stroke, head injury, intentional tremor), (2) UE amputation, (3) hand joint arthroplasty, (4) trauma within the past 6 mo that increased nonarthritic symptoms, (5) hand or wrist conditions (e.g., carpal tunnel syndrome), (6) hand deformities that prevent the grasping of instruments, and (7) strength testing contraindications due to medical comorbidities.

Instrumentation

Participants completed 16 trials at attempting to break the seal on a force-sensing jar apparatus through counterclockwise twists. This instrument (i.e., the jar apparatus; McGee, Nuckley, & Mathiowetz, 2011) has good construct and ecological validity and excellent intratester reliability for the outputs used in this study. The jar’s lid had six force-sensing resistors and a six-axis load cell, allowing the jar to detect grip forces (i.e., forces on the jar lid during a power grasp measured in newtons) and the torque (measured in newton meters [Nm]) about the lid’s axis of rotation. During each trial, the torque required to turn the jar device was set through use of a torque limiter to replicate the actual torque required to open a sealed jar with an 83-mm diameter lid of 4.24 Nm (Soroka, 2002). Once the torque threshold was exceeded, the jar lid would rotate freely for 90° to simulate breaking the seal.

The jar-opening simulations (see Figure 1) consisted of two trials of three factors: (1) the approach (supinated stabilizing hand vs. oblique stabilizing hand), (2) the lid-gripping hand (right vs. left), and (3) the use or nonuse of nonskid material. The jar-grasping approaches were chosen because they are common approaches (Bush, Bix, Bello, & Fair, 2013). Participants were randomly assigned to predetermined sequences of the eight conditions.

OPEN IN VIEWER

Figure 1.

Oblique and supinated approaches with and without use of nonskid material.

Standardized procedures (McGee et al., 2011) were used for each jar-turning simulation. In a standing position, participants maintained standardized glenohumeral and elbow joint positions as well as hand placements to control for distal kinetic variance that might result from nonstandardized posturing. Between trials, 30 s of rest were offered.

Measures

Demographic information (i.e., race, age, self-reported hand dominance; see Corey, Hurley, & Foundas, 2001), disease impact, and maximum hand strength were recorded at baseline. Self-perceived impact of hand OA on daily living was characterized via the Arthritis Impact Measurement Scales 2—Short Form (AIMS2–SF; Ren, Kazis, & Meenan, 1999). Maximal voluntary contraction (MVC) grip strength of the dominant and nondominant hands was measured by the Jamar dynamometer and the B&L Pinch Gauge in accordance with standardized positioning and instructions (Mathiowetz, Weber, Volland, & Kashman, 1984).

Across all trials, the primary outcomes—grip force and opening torque—were recorded for a maximum period of 6 s. Secondary outcomes were success in breaking the seal and pain as measured with the Numerical Rating Scale (NRS; Downie et al., 1978). The NRS, a 0–10 scale of pain intensity, is commonly used in clinical practice and has high responsiveness to pain intensity changes (Ferreira-Valente, Pais-Ribeiro, & Jensen, 2011). An NRS rating of 0 indicates no pain, whereas a 10 indicates extremely strong or maximal hand pain. The first author was the sole evaluator of the outcomes because of the specialized training required to manage the instrumentation.

Statistical Analysis

MATLAB 155 (Version 7.9; MathWorks, Natick, MA) was used to compute grip force at peak torque of the jar lid, the IGF, and peak torque acting on the jar’s lid. NRS change scores (postbaseline) were calculated with IBM SPSS Statistics (Version 22; IBM Corp., Armonk, NY). Descriptive statistics and within-subjects comparisons (Student’s t, Cochran’s Q, and Wilcoxon signed-rank test) for baseline characteristics, hand force, torque, success, and pain were performed with SPSS. We used SAS (Proc Mixed, Version 9.4; SAS Corp., Cary, NC) to estimate the effect of and interactions among the three experimental factors on pain, peak grip force, IGF, and peak torque using a mixed-effects linear model with hands as the repeated factor and participants as the random effect. Post hoc within-subject comparisons of significant interactions were performed with Student’s t, using Bonferroni correction for multiple comparisons. Cohen’s d effect sizes for significant main and interaction effects were calculated and later interpreted (Cohen, 1988).

Participants

Thirty-one women with hand OA participated. One woman was unable to complete eight trials with her left hand because of increased arthritic symptomatology during forearm pronation. The sample’s AIMS2–SF average total health score of 10.62 (standard deviation [SD] = 5.28) indicated that perceived health was mildly affected by the condition. Participants also reported mild disease impact on the Physical Functioning, UE Functioning, Social Interactions, and Affect subscales of the AIMS2–SF. See Supplemental Table 1 (available online at http://otjournal.net; navigate to this article, and click on “Supplemental”) for details on the sample’s characteristics.

On average, participants had OA in two digits in both the right and the left hand (mean [M] = 2.1, SD = 1.5, and M = 2.0, SD = 1.5 digits, respectively). Ninety percent had radiographically confirmed or self-reported thumb OA; the percentages for other digits were as follows: 45% index, 42% long, 23% ring, and 29% small. These percentages vary from reported distribution patterns in which index, middle, and small fingers are most often radiographically confirmed (Addimanda et al., 2012). This finding may be due to the large portion of the sample who were seeking occupational therapy for first carpometacarpal (CMC) OA at orthopedic clinics, which, when compared with interphalangeal hand arthritis, yields significantly worse hand function and symptomatology (p < .01; Bijsterbosch et al., 2010). The participants’ left hand had significantly more small finger proximal interphalangeal joints affected by OA than did their right (z = 2.00, p = .05). There were no other significant differences in the right–left distribution of joint involvement, the total affected joint counts, or maximum grip or pinch strength between right and left hands in our sample.

Force Outcomes

Success

Success rates ranged from 18.3% to 85.2%. The right-hand oblique approach without nonskid material had the least success; the left-hand supinated approach with nonskid material had the highest (see Table 1). Three independent effects of hand, nonskid material, and approach on success were discovered: (1) A left-hand turn was more successful (95% confidence interval [CI] [40%, 68%]) than the right (95% CI [27%, 54%], p = .01), (2) the use of nonskid material resulted in greater success (95% CI [42%, 67%]) than nonuse (95% CI [26%, 54%], p = .001), and (3) the supinated approach had greater success (95% CI [53%, 79%]) than the oblique approach (95% CI [17%, 41%]). These effects were respectively small, moderate, and strong (Cohen’s ds = 0.39, 0.42, and 1.09; sees Tables 2 and 3 for details).

Pain

The average pain intensity score ranged from 2.21 (SD = 2.2) to 3.42 (SD = 2.5), with the highest pain intensity after a left oblique approach with nonskid material and the least pain intensity after a left supinated approach with nonskid material. The largest change in pain (M = 2.74, SD = 2.6) was when a right-hand oblique approach with nonskid material was used, whereas the smallest change in pain (M = 1.3, SD = 2.5) occurred when a left-hand supinated grasp with nonskid material was used (see Table 1).

Pain NRS scores were significantly higher, with more change, when an oblique approach was used than when a supinated approach was used, Fs(1, 411) = 25.98 and 23.77, ps < .0001, respectively. There were significantly larger pain increases when the right hand as opposed to the left was used, F(1, 411) = 3.88, p = .05. A nonskid material + oblique approach yielded significantly higher pain intensity ratings than any other combination (p < .05). Post hoc comparisons identified pain intensities to be different in all instances except for between the supination approaches. Although different, the relative effects of these combinations on pain intensity were small, with the largest effects on pain intensity noted when we compared the supinated + nonskid approach with that of the oblique + nonskid approach, and the smallest effects noted when we compared the two oblique approaches (see Tables 2 and 3).

Effect of Nonskid Materials

The results of this study reveal that the use of nonskid material alone enhances hand torque acting on a jar lid. Across both hands and all approaches, nonskid material enhanced torque by roughly 6.1% and had a small significant effect on success. The effect of nonskid material, however, did not independently influence torque at peak torque or IGF but likely contributed to the development of downward compressive force that may have enhanced the frictional interplay of the hand and lid and necessitated less grip force. This assumption requires further exploration.

Effects of Approach

The supinated approach, independent of hand and nonskid material, produced the greatest torque. Although the pain-stabilizing effect of the supinated approach was small, the mean pain change of participants who used a supinated approach did not exceed the clinically significant change of 2 units on the NRS (Farrar, Young, LaMoreaux, Werth, & Poole, 2001), whereas the pain change of all oblique approaches did. Here, exceeding the clinically significant change is undesirable because it indicates that such an increase in pain would be of functional importance to the client.

Effects of Turning Hand

Although the sample was 90.3% right handed and the right hand’s maximum strength was slightly larger than the left’s, the left hand appeared to be better at generating torque. Because the arthritis distribution is similar across right and left joints, except for the small finger proximal interphalangeal, these results cannot be attributed to a higher right-hand joint count. An argument also cannot be made that the left hand generated less force across all factors because it was weaker; this hypothesis is not supported by the sample’s baseline hand-strength characteristics.

Participants’ baseline pain intensity had no effect on grip force at peak torque (p < .32). The lid-turning hand did, however, influence the pain experienced. Participants who twisted the jar lid with the left hand reported a smaller increase in pain than did those twisting with the right (MD = 0.31, SD = 0.16, p = .05). This difference may be explained by the left hand’s reduced grip force profile and inferred changes in the hand’s joint reactive forces, in particular the thumb, which constitutes 41% of the grip force profile during jar twisting (Shim, Huang, Hooke, Latsh, & Zatsiorsky, 2007), and the high prevalence of participants with first CMC OA. The left hand’s reduced grip force profiles are supported by the significantly greater safety margins (i.e., ratio of actual grip forces to those needed to prevent slippage) of healthy-handed participants when opening a simulated lid compared with closing it with the right hand (Shim et al., 2007).

The findings of Shim et al. (2007) support ours, given that the right hand’s closing turn is analogous to (i.e., mirrors) the left hand’s opening turn. These differences are likely explained by Shim et al.’s additional conclusions that, when opening with the right hand, the thumb’s contributions to torque are inefficient relative to the grip force it exerts and, as a result, higher grip forces are needed to compensate. Although not explored by Shim et al., we speculate that this is explained by differences in frictional interplay between the thenar eminence and jar lid when opening (i.e., the thenar eminence and all of its surface area are perhaps more intimately connected to the lid when opening with the left hand than with the right).

Combined Effects of Hand, Approach, and Nonskid Material

The left-hand supinated approach with nonskid material resulted in an 85.5% success rate, the lowest pain intensity, and the least increased pain. This result may be due to the positive effects of nonskid material on torque, the reduced left-hand grip force requirements at peak torque, and the IGF-reducing effects of the supinated approach + nonskid material. We also speculate that, for participants with first CMC arthritis, the girth of the jar base stressed the stabilizing hand’s thumb into maximal palmar abduction when using the oblique approach, which resulted in increased pain and difficulty in providing the counterforce needed to rotate the lid independent of the base.

The first CMC joint compressive force in the left, supinated, nonskid approach was likely between 367.6 and 527.9 Ns, whereas it was projected to be between 638.1 and 916.5 Ns in a right, oblique, nonskid approach. This estimate is supported by the findings of Shim et al. (2007) and the cadaveric work of Cooney and Chao (1977), in which for every 1 N of lateral pinch, the same pinch used in this experiment, between 9.3 and 13.4 N are acting on the first CMC joint. These findings are particularly striking given the high percentage of participants with first CMC OA and the projected between-approach differences in first CMC loads of up to about 400 Ns.

Although least painful, most efficient, and most successful, the left supinated grasp with nonskid material approach still required about 30.6 lb (136.2 N) of force, which was 60.4% of the participants’ left-hand MVC grip strength. Because frequent use of forces exceeding 15%–20% of MVC may be linked to the development of UE musculoskeletal disorders (Tanaka et al., 1994), nonskid materials may not suffice as a stand-alone JP approach. Instead, women with hand OA should also use a preferred approach, other JP strategies, or assistive jar opening devices.

Given that opening a jar is often part of meal preparation and is of importance to women with hand OA (Kjeken et al., 2005), these findings call into question the validity of the assumption that grip strength of 20 lb is enough to engage in daily activities (Terrono, Nalebuff, & Phillips, 2011). On average, our participants possessed MVC grip strengths 30 lb greater than this benchmark. In addition, participants exerted grips between 30.6 lb (136.2 N) and 38.2 lb (169.9 N) when attempting to break the seal. Although both MVC grip strength and grip forces during the task exceeded the recommended 20-lb threshold, many participants were unable to break the seal. Given the large forces required, there is a need for manufacturers to consider alternative jar design to mitigate risks.

This study had multiple limitations. Participants had mild symptoms with low effects on function per the AIMS2–SF Total Health Scale and were predominately right handed. Radiographic staging was not available to determine arthritis severity, but radiographic changes in hand arthritis do not often correspond with clinical symptomatology (Bijsterbosch et al., 2011). Although we evaluated the turning hand’s force and torque, we did not explore the effect of the stabilizing hand’s kinetics or wrist joint kinematics during the task. Participants did not rate pain in the stabilizing hand, only in the actively gripping hand.