Hand kinematics in osteoarthritis patients while performing functional activities

Abstract Purpose To identify the impact of kinematic limitations on hand osteoarthritis patients’ ability to perform daily living activities. Methods An experiment was performed on 33 patients and 32 healthy subjects. Active ranges of motion (AROM) of 16 hand joint angles were measured, together with scores of different hand tests of dexterity (Box and Block, Nine Hole Peg, Kapandji) and function (Sollerman Hand Function Test, SHFT). Functional ranges of motion (FROM) were recorded during SHFT tasks. Results by task are also reported. Results Patients’ AROM is limited in flexion of thumb carpometacarpal and interphalangeal, and finger metacarpophalangeal (MCP) and proximal interphalangeal (PIP) joints, and in palmar arch. Patients scored worse in gross dexterity and opposition, but only Kapandji score was correlated with AROM limitations. Pain is mostly reported in patients with limited extension of finger MCP and PIP joints. Patients used significantly different FROM in almost all the joints, and needed more time to accomplish the SHFT tasks. Conclusions AROM measurements can be used as indicators for early diagnosis. Patients use specific strategies to accomplish each task, arising from AROM limitations; some tasks with very extreme postures. The tasks where precision or force are required for thumb are the most affected ones. Implications for rehabilitation Active range of motion is an indicator for early hand osteoarthritis diagnosis. Patients' functional ranges are reduced, and thumb opposition and gross dexterity are hindered. Rehabilitation should focus especially on tasks requiring precision and thumb strength. Rehabilitation should favor the improvement of task completion times.


Introduction
Hand osteoarthritis (HOA) is a prevalent disease involving hand joints that leads to pain and impairment of hand function [1,2]. HOA can be found to different degrees in 81% of the elderly population [3], women having a higher prevalence than men [4,5]. The most affected joints are the distal interphalangeal (DIP) joints of fingers and the carpometacarpal (CMC) joint of the thumb (rhizarthritis) [6], followed by the metacarpophalangeal (MCP) and proximal interphalangeal (PIP) joints of the fingers and interphalangeal (IP) joint of the thumb. Consequences of HOA are pain, stiffness, decrease in grip and pinch strength [7,8] and limited range of motion [9][10][11]. Disability depends on the joints affected and the level of compromise. Compromise of the thumb CMC and finger MCP joints is generally associated with hand disability [12,13] while compromise of the PIP and DIP joints commonly displays bone enlargement at the joints, producing rejection and a disability that is more perceived than real [14].
Despite its high prevalence, HOA is a silent degenerating disorder that is clinically disregarded in most cases, except in very severe situations as in the most severe and disabling form, the erosive HOA [3]. Applying adequate treatments in the early stages would benefit patients' quality of life and could prevent structural progression of the disease [15].
Disability assessment in HOA is commonly performed through subjective questionnaires about physical hand function, pain or satisfaction [16][17][18][19]. Therefore, the diagnosis and follow-up of the patients is very dependent on their willingness to recognize their functional limitations [20].
Recently, hand kinematics is being explored to understand hand motion in HOA patients. Leitkam and Bush identified active range of motion (AROM) loss in flexion and extension of all thumb joints and finger MCP joints, and in flexion of DIP joints [21]. Nonetheless, the relation between the loss of AROM and loss of functionality in terms of performing activities of daily living (ADL) is still unknown [22]. Some recent studies have compared the hand kinematics of healthy subjects and HOA patients while performing some ADL. Luker et al. [23] compared thumb kinematics of one HOA patient with advanced CMC arthritis with that of four healthy subjects while exerting 80% of maximal effort during pinching, grasping, and twisting. They found that the CMC joint was more extended and adducted in the patient. Tanashi et al. [24] measured the kinematics of thumb IP and MCP joints, together with the index and middle DIP and PIP joints, of 10 healthy participants and nine HOA patients while performing nine ADL. Range of motion used was lower in HOA patients, especially in the index DIP and middle DIP and PIP joints and during plug, key, and bottle tasks. Holland et al.'s study [10], with 20 healthy and 20 HOA patients and five ADL, reported differences in the angles between segments of thumb and index finger for the spray bottle squeeze task.
The studies of the kinematics of HOA patients during ADL seem promising but are still scarce and present some limitations such as the sample size, the hand joints analyzed, and the number and representativeness of the ADL considered. This work aims to address these limitations and delve deeper into the relationship between hand kinematics and impact on hand function of HOA patients. To do this, an instrumented glove was used to measure the AROM and the kinematics of the whole hand of healthy and HOA participants while performing the 20 activities of the Sollerman Hand Function Test (SHFT) [25]. Although wearing instrumented gloves reduces fine motor skills, it has been shown good enough for assessing tasks requiring medium and gross motor skills [26], and does not have the occultation problems of optical motion capture systems, which are critical during manipulation of objects in ADL performance. Different analyses were performed to check the impact of the changes in patients' AROM on the functional range of motion (FROM) during these activities, and on the scores of different widespread function tests and the pain perceived by the patients.
To the best of the authors' knowledge, this paper presents the most extensive study conducted to date in HOA patients, with the highest number of activities and joints studied, and relating AROM and kinematics in ADL to different scores of hand function in order to identify the impact of kinematic limitations on the patients' ability to perform tasks.

Subjects
Thirty-three HOA patients (sample P): all of them right-handed females (70.0 ± 10.0 years of age) were recruited by clinicians among hospital patients showing different disease stages and different levels of compromise. None of them underwent surgery.
Experiments on healthy subjects (sample H) were performed in two stages because of the pandemic situation, so that two different subsamples of healthy right-handed adult subjects were considered: H1 with 32 participants (17 females, 15 males; 37.6 ± 13.0 years of age), and H2 with 27 (14 females, 13 males; 38.7 ± 7.6 years of age), 12 subjects being present in both groups.

Materials
One right-hand instrumented glove (Cyberglove Systems LLC, San Jose, CA) was used to record (100 Hz) 16 hand joint angles (see Supplementary Appendix A, figure A1) in accordance with a previously validated protocol [27]. Flexion and abduction angles were considered positive.
Standard kits were used for the different tests: SHFT, Box and Block Test (BBT), and Nine Hole Peg Test (NHPT). The 20 tasks of the SHFT (see Supplementary Appendix A, figure A2) were considered to be representative of ADL.
A figure printed on a sheet of paper (see Supplementary Appendix A, figure A3) was shown to HOA patients for them to indicate the joints where they were experiencing pain.

Experiments
All participants signed a written informed consent document, approved by both the University and Hospital Ethics Committees.
First, P and H1 participants performed the dexterity tests (BBT and NHPT) and Kapandji test (KT) for thumb opposition with their bare right hand. Then, the instrumented glove was used to measure their AROMs through 12 static postures as in [22] (see Supplementary Appendix A, figure A4). P and H2 participants performed the SHFT while wearing the instrumented glove. Finally, patients were asked to report any difficulties they had when carrying out ADLs at home, and to indicate the joints where they were experiencing pain.

AROM, dexterity, and Kapandji tests
For each joint angle and participant, maximum and minimum angles were identified during the 12 static postures, and AROM was computed as the difference between them. Statistics for AROM and extreme joint angles were computed per sample. For each patient and joint angle, the percentile of the patient's AROM as if she belonged to the healthy sample was also obtained. Finally, a set of 48 analyses of variance (ANOVAs) were applied: the AROM and extreme values for each joint angle were the dependent variables, and the condition (patients -healthy participants) was the independent one.
Standard scores of the BBT, NHPT, and KT were obtained: the number of blocks passed in each trial, the total time spent (seconds) and the number of postures achieved, respectively. Scores on dexterity tests (BBT, NHPT) were normalized to make them non-dependent on the age and gender, according to Equation (1) and using normative data from [28] for BBT and from [29] for NHPT. Then, three ANOVAs were performed (scores as dependent variables, condition as the independent one).

Normalized value ¼ measured value
� ðmaximum normative value for any age of hose of the same gender as the subject= normative value for subject's gender and age: (1) Correlations were computed for the patient sample: between the AROM values and the scores from the dexterity tests (Pearson's correlations) and between the AROM values and KT scores (Spearman's correlations).
The relationship between the loss of AROM and the perception of pain was also investigated by cross-tabulating the subject's AROM percentile (as if she belonged to the healthy sample) and the pain reported. Three groups of percentiles (P group) were considered: lower than the 5th percentile (<p5), meaning decreased ranges with respect to healthy subjects; between the 5th and 95th percentiles (p5-p95), meaning normality; and higher than the 95th percentile (>p95), meaning high ranges. Joint angles were grouped into four sets (Angle group): thumb, flexion of finger MCPs, finger PIPs, and finger abduction.

Kinematics during ADL and functional scores
The kinematics during ADL for both samples were characterized by the statistics of the angles used by each subject during SHFT tasks, taking the tasks together: mean, 5th and 95th percentiles, and FROM (computed as the difference between the 95th and 5th percentiles) [22]. ANOVAs were performed with these statistical parameters as dependent variables, and the condition (patients -healthy subjects) as the independent one. These analyses were repeated per task and results are presented in Supplementary Appendix B.
Standard scoring was applied to the SHFT: each task was assigned a five-level score according to the type of grasp used, the level of difficulty observed and the time required to accomplish the task (4 when the accomplishment is as expected, 0 when the task is not performed) and a global score was computed as the sum of the individual scores. The total time required by each participant to complete the 20 tasks of the SHFT, in seconds, both per task and globally, was considered an alternative scoring, as proposed in [30]. As for BBT and NHPT, the SHFT time score was normalized to make it non-dependent on the age and gender, according to Equation (1) and using normative data from [30]. Two ANOVAs were performed with these two scores (standard score and normalized time) as dependent variables, and the condition (patients -healthy) as the independent one. These two ANOVAs were carried out for the global scores (20 tasks altogether) and per task. Table 1 shows descriptive statistics of AROM and extreme joint angles. Significant differences (p < 0.05) are marked with bold underlined numbers for the most extreme joint angles and for the highest AROM value of each pair.

AROM and dexterity and Kapandji tests
Mean (SD) values obtained for the normalized scores of the BBT and NHPT and the standard KT are (healthy subjects vs. patients): 85 (13) vs. 77 (17); 17 (2) vs. 19 (5); and 10 (0) vs. 8 (2), respectively. Only BBT and KT present significant differences between samples. Table 2 shows the correlation coefficients of the AROM values with the KT scores (Spearman's coefficient) and of the AROM values with the dexterity test scores (Pearson's coefficient), for the P sample. KT scores are significantly correlated with the AROM of most joints, while few significant correlations are found for dexterity tests. Table 3 shows the cross-tabulation between the "P group", corresponding to the percentile of patient's AROM, and "Pain" (reported/not reported) for each "Angle group". Chi-square computation showed significant differences for all joint angles together and for flexion of finger MCPs and PIPs groups. Although the thumb is the group with more reported pain (51%: 67 out of 132 cases), no significant difference in reported pain is found among patients with thumb AROM below the 5th percentile. A significantly higher number of patients reporting pain is observed amongst those with AROM below the 5th percentile of flexion of finger MCPs and finger PIPs.

Kinematics during ADL and functional scores
As expected, all the healthy subjects accomplished all the tasks completely. For patients, only 0.5% of the 660 tasks performed (33 patients � 20 tasks) were unaccomplished: task 5 (two patients) and task 12 (one patient), and 2.6% of tasks were partially accomplished: tasks 3 (three patients), 7 (nine patients) and   6, 10, 15, and 18 (one patient). The patients reported some difficulties to carry out ADL at home: 72% reported problems when it came to handling very small objects and 52% for picking up coins. Additionally, 67% reported difficulties for handling heavy objects, and 58% had some discomfort when preparing or cutting food. Moreover, 82% of patients reported difficulties opening jars. Table 4 shows descriptive statistics of joint angles used during ADL. Significant differences (p < 0.05 in ANOVAs) are marked with bold underlined numbers for the highest mean and FROM, and most extreme 5th and 95th percentiles of each pair. Patients' thumb has a more abducted and less flexed CMC joint (even when extended), with lower FROM for both flexion/extension and abduction/adduction, a lower IP FROM and a slightly more flexed MCP joint. Patients also present lower FROM in flexion of finger MCP joints, PIP joints and palmar arch, and less extreme flexion/ extension values in general during ADL performance. Table B1 (Supplementary Appendix B) shows the same results for each SHFT task. Specific strategies arising from AROM limitations can be observed. For example, patients cannot reach the extreme flexed postures used by healthy subjects in the CMC joint in tasks 1, 7, 8, 11, and 16 (Figure 1), and consequently they use more extreme flexed postures of the thumb MCP joint. The FROM of the index finger MCP joint is significantly lower in precision tasks such as 1 and 3 (pick up coins), 7 (pick up nuts) or 11 (do up buttons). More strongly affected tasks can be identified; for example, patients use less FROM in task 12 in all joints except for flexion of the thumb MCP joint and abduction between the ring and little fingers. Tasks involving very extreme postures for HOA patients Table 3. Cross-tabulation "P group"�"Pain"�"Angle group". % within pain   Pain   Total   Pain   Total   Pain   Total  No  Yes  No  Yes  No  Yes   Angle group  Thumb  P group  <p5  17  17  34  50  50  100  26  25  26  p5-p95  48  49  97  50  51  100  74  73  74  >p95  0  1  1  0  100  100  0  2  1  Total  65  67  132  49  51  100  100  100  P groups: lower than 5th percentile (<p5) meaning decreased ranges, between 5th and 95th percentiles (p5-p95) meaning normality, and higher than 95th percentile (>p95) meaning high ranges. Angle groups: thumb, flexion of finger MCPs, finger PIPs, and finger abduction. Angle groups with significant differences are marked with bold underlined numbers.  can also be identified. For example, the thumb CMC joint of patients reaches extension values close to 30 � in tasks 4 (lift wooden cubes) and 10 (unscrew lid of jars). HOA patients performed significantly worse (p < 0.05) in both SHFT scores (standard score: 76 ± 2 vs. 68 ± 10; total normalized time: 248 s ± 44 s vs. 295 s ± 107 s). Percentage differences in both scores per task can be observed in Figure 1. Scores per task are lower in patients in all tasks, and are significantly lower (p < 0.05) in most of them. Differences in standard scores range from 2 to 28, with tasks 7, 5, and 13 showing the highest differences. Normalized times per task are mostly higher in patients, and are significantly higher (p < 0.05) in half of the tasks. Differences in normalized time per task have a wider range of variation (from À 43% to 88%), with tasks 14, 13, and 6 showing the highest differences. Note that task 12 presents a significant lower mean normalized time in patients.

Discussion
Several limitations in the joint mobility of HOA patients have been identified. A significant reduction in AROM has been observed in flexion/extension of those joints more affected by  HOA [1,2], namely, thumb CMC, finger MCP, thumb IP, and finger PIP joints (except for the little finger). These AROM limitations are in accordance with previous studies [21,31]. The AROM of the palmar arch, the measurement of which is a novelty, is also significantly reduced. The effect of the disease on the mobility of these joints is clear, as the reduction is present in all patients, regardless of the level of compromise. This makes the AROM measurement on these joints a possible indicator for the early diagnosis of HOA, avoiding the need of using radiation imaging diagnose techniques (such as X-ray).
The effect of these limitations on patients' ability to perform tasks has been studied first by analyzing the scores of widespread clinical tests. Patients have difficulties in performing thumb opposition, with significantly worse KT scores. Highly significant positive correlations have been found between KT scores and AROM for flexion/extension of all joints, so that, as expected, the higher the AROM reduction in these joints is, the greater the difficulties for thumb opposition will be. Worse dexterity scores have also been found, but differences are significant only for BBT. This confirms a loss of gross dexterity that is not accompanied by significant correlations with AROM, except for flexion/extension of little MCP and PIP joints, which seems spurious. The worse scores found for the NHPT were non-significant, although most patients (72%) reported problems when handling very small objects. The higher cognitive requirements of NHPT, highly dependent on age, might partially explain this fact. The normalization applied considers age, which was high for the sample of patients, and this might have attenuated the effect of loss in fine dexterity. Furthermore, elder subjects with incipient HOA might be present within the sample considered in the experiment where NHPT data were collected [29], given the inclusion criteria used.
The relationship between the limitations in joint mobility and pain has also been investigated by cross-tabulating the percentile of patient's AROM and reported pain. Indeed, reported pain is significantly related to the AROM of finger MCP and PIP joints but not to thumb AROM. These data should be taken with caution, as some patients said that they were taking medicines for pain from different sources. In any case, the relationship between AROM percentiles and reported pain reinforces the idea of using AROM as an indicator for early diagnosis.
The implications of joint mobility limitations on the performance of daily living activities have been addressed by analyzing the hand kinematics during the 20 SHFT tasks, as representative of ADL. The FROM of HOA patients in the thumb MCP joint is not significantly altered, in accordance with [24]. The reduced flexion/ extension AROM of patients in the thumb CMC joint becomes a significantly reduced FROM, not only in flexion/extension but also in abduction/adduction. The thumb CMC joint is used in a significantly more extended posture, as observed in [32], showing more extreme extended postures during ADL performance. This is probably a consequence of laxity of the anterior oblique ligament, which is one of the main contributors to osteoarthritis of the CMC joint [33]. The instability caused by this laxity seems to prevent the use of more adducted and flexed postures, thereby causing the FROM reductions observed. The AROM reduction observed for flexion/extension of MCP and PIP joints finally affects hand function, becoming a reduced FROM (not significantly for the index finger MCP joint globally, although significant differences appear in precision tasks). Interestingly, the AROM reduction of the palmar arch, not studied in previous work, is also transformed into a reduction in FROM, therefore having an effect on hand function.
Despite the mobility limitations observed in HOA patients, they have been able to fully accomplish the SHFT tasks requested (96.9%). However, significant differences between samples were found for the SHFT standard score and the total normalized time, thus revealing difficulties when performing the tasks. Patients obtained significantly lower scores in all tasks, but especially in those where precision is required for their thumb (e.g., pick up nuts) or those requiring force (e.g., lift an iron). The normalized time score identified significant differences in fewer tasks, probably because the standard SHFT scores have not been corrected for differences in the age of samples, because there are no normative data available to do so, and no correction is considered in current clinical practice. Furthermore, the normalized time score has also been shown to be more sensitive to HOA patients' difficulties than the standard score. In general, patients required significantly more time to accomplish the tasks, especially in those where the thumb is essential for a stable grasp, such as in writing with a pen, screwing with a screwdriver and cutting with a knife and fork. These results are in accordance with the information given by the patients, who reported difficulties in activities where precision is required, such as handling very small objects, where the thumb plays a key role, or when force is required, such as handling heavy objects, opening a jar or cutting food. Surprisingly, patients required significantly less time to put a Tubigrip stocking on the other hand. Further work is needed in order to determine whether this reduction in time is due to a higher motivation or to another cause.
Deeper insight into the implications of joint mobility limitations on the performance of daily living activities may be achieved by analyzing data per task. Supplementary Appendix B provides table B.1, with statistics of the functional angles during SHFT per task, that is, mean posture, FROM and 5th and 95th percentiles (p5 and p95, respectively) per joint and significant differences are highlighted. Specific strategies used by patients to accomplish each task, arising from AROM limitations, can be observed and tasks involving very extreme postures for HOA patients can also be identified. This information might be used to design and indicate proper assistive devices to reduce the functional requirements in those tasks with higher demands. It could serve as an aid to teach patients to change the way they carry out the activities in order to overcome progressive degeneration. This paper presents some limitations, such as the effect of using a glove during the recordings, although previous studies have shown that this effect is small in tasks requiring medium and gross motor skills. Another limitation is that the DIP joints, which are commonly affected, are not included. Patients included in this study reported a certain aesthetic rejection to the deformation of these joints, but they did not consider them disabling, unlike what happens when thumb joints are involved. Furthermore, force, which is commonly reported to decrease with pain as a consequence of HOA, has not been recorded, although recent studies [34] claim that in the initial stages HOA does not affect the extrinsic muscles of the hand. They also stress that early detection of the illness is of utmost importance to be able to apply adequate rehabilitation.

Conclusions
The AROM limitations in the joints commonly affected by HOA, together with the relationship between AROM percentiles and reported pain, support the idea of using AROM measurements as indicators for early diagnosis.
Despite the reduced FROM, the difficulties in thumb opposition and the loss of gross dexterity, patients could in general fully accomplish the SHFT tasks requested. Some difficulties were observed though, especially when precision of the thumb or force is required. These difficulties have an impact in the time spent on performing the tasks.
The data provided broaden the information available for the design of assistive devices for HOA patients. Patients use specific strategies to accomplish each task, as a result of AROM limitations and some tasks are performed by patients using very extreme postures.