Background
Coordination of upper limb movements is often impaired after a stroke. This study aimed to kinematically evaluate in depth upper limb coordination in persons post-stroke using a common clinical pointing test, the Finger-to-Nose test (FNT) [
1]. The FNT is usually scored by the time to complete the task, while our goal was to assess the added value of a kinematic analysis of the test. According to the framework of the International Classification of Functioning, Disability, and Health [
2], coordination of simple and complex voluntary movements involves performing movements in an orderly combination (Body Functions), and performing coordinated actions such as carry, move and handle objects (Activities). The concept of coordination is complex and there is still a lack of consensus around a clear definition [
3]. Bernstein viewed coordination as the process of mastering the redundant degrees of freedom involved in a particular movement, and that motor redundancy was considered as a source of computational problems solved by a unique solution [
4]. In this paper, the operational term of coordination is referred to as the spatiotemporal relationship between component parts [
5].
In clinical practice, the FNT is an established test used to assess upper limb coordination [
6]. FNT is also included in the Fugl-Meyer Assessment for the upper extremity (FMA-UE) that evaluates upper limb impairments following stroke [
7]. Most commonly the person is seated and upon command moves the index fingertip back and forth between the ipsilateral knee and the tip of the nose five times, as fast and as accurately as possible. The quantitative clinical main outcome measure of the FNT is total time of performance, which is considered more reliable than the qualitative scoring of dysmetria and tremor on a ordinal rating scale [
1,
8], as it is performed in the FMA-UE [
7]. Compensatory movement strategies of the trunk and/or shoulder are however not taken into account. As persons post-stroke often use compensatory movement strategies to accomplish upper limb tasks, such movements are important to consider when evaluating upper limb recovery [
9]. Kinematic evaluation offers more detailed characteristics of motor performance than merely the total time to perform the task, since it provides comprehensive and informative data about joint motion sequencing and timing involved in coordinated movement. To the best of our knowledge, there are no studies that have used kinematics to quantify motor performance during the FNT, as used by Fugl-Meyer et al. [
10].
There are several kinematic variables that have been shown to be highly reliable and valid to characterize pointing and reaching movements in persons post-stroke [
11‐
14]. Motor performance, targeting aspects of movement quality, is usually assessed in terms of temporal (movement time, speed, smoothness) and spatial (joint angles, target errors, compensatory trunk movements) parameters, which may be obtained from joint and body segment kinematics and calculations of end-point positioning [
12,
15,
16]. Smoothness of the movement path is considered an important characteristic of well-coordinated movement [
17,
18], and a relevant variable that discriminates between persons post-stroke and non-disabled controls [
12,
17‐
20] as well as between persons with different levels of stroke severity [
12,
19,
20]. Previous studies in persons post-stroke have characterized motion deficits during pointing tasks to external targets, (e.g., away from the body) with [
20,
21] and without vision [
19,
22,
23]. Persons with stroke have prolonged movement times, smaller movement amplitudes, more variable upper limb movements, and disrupted elbow-shoulder coordination in the affected arm compared to the non-affected arm, as well as compared to non-disabled controls [
19,
20,
23]. In addition, persons with moderate-to-severe stroke use excessive trunk movements compared to persons with mild stroke when making forward pointing movements [
20]. Despite lower speed, pointing movements in persons post-stroke are less precise than those of non-disabled controls and decreases in movement accuracy correlate with level of stroke severity [
19,
20]. A few studies have investigated reaching-to-mouth tasks [
12,
24,
25], in which upper limb movements are similar to those used for the FNT, although the tasks have different visual conditions, accuracy, and time constraints. In some of these studies, compensatory movements such as increased shoulder abduction have been reported [
12,
25].
Despite increased research of upper limb movements in persons post-stroke during recent decades [
26], there are few studies of pointing movements to body-related targets, and sensorimotor control is not well understood. Detailed understanding of reaching movements undertaken in prevailing clinical tests are needed in order to fully comprehend their constructs and measurement properties [
27]. Therefore, it is important to investigate the construct validity of FNT as a test of coordination within persons with different functional levels post-stroke. In the present paper, we hypothesized that the kinematic variables would substantially contribute to reveal the inherent ability of the FNT to capture upper limb coordination and compensation.
Hence, we aimed to characterize the FNT performance in a group of persons post-stroke and a non-disabled control group with regard to kinematic variables of the hand (pointing time, peak speed, time to peak speed, number of movement units, path ratio, and pointing accuracy), elbow/shoulder joints (range of motion, interjoint coordination), and scapular/trunk movement. The second aim was to address two aspects of construct validity; i) to compare FNT performance between persons with varying levels of impairment post-stroke, and ii) to determine if the FNT is valid test of coordination by relating the clinical main outcome measure of the FNT, i.e., total movement time (TMT), to the kinematic variables of the pointing phase of the test.
Discussion
To our knowledge, this is the first study to investigate the construct validity of the FNT, as performed in the FMA-UE, based on kinematic analyses. The results indicated that four temporal and three spatial of 16 investigated kinematic variables during performance of the FNT differed between controls and persons post-stroke. The TMT to perform the FNT was prolonged for the stroke group, which is consistent with prior studies of pointing tasks in persons post-stroke [
19,
22]. Nevertheless, two-thirds of the stroke group had similar movement times as half of the controls. This may be explained by the predominantly mild impairments in the present stroke group. The stroke group had also less straight and less accurate movements, indicated by increased Path ratio and end-point errors, in agreement with prior studies of pointing movements [
19,
20,
22]. Previous studies have reported high variability in accuracy during forward reaching post-stroke [
20,
30,
34]. One of these studies compared two reaching conditions and found slightly higher end-point errors in reaching upwards to a target compared to reaching forward to a target [
30]. Indeed, our stroke group also demonstrated high variability in end-point errors when pointing upwards to the nose. Although both groups in our study demonstrated similar variability in Path ratio and ROM in the shoulder and elbow joints, the stroke group showed higher variability in end-point errors compared to the control group. Interestingly, even though a subgroup of persons post-stroke had equal movement time as the controls, they still demonstrated reduced accuracy and compensatory movements. Thus, the 3D motion analysis captured deviations in pointing movements post-stroke which were not directly time-dependent.
The interjoint correlation values were high between shoulder and elbow joints of both the control group and the stroke group, but especially in the affected arm (IJC close to 1), representing a high coupling between these joints. In analogy with our findings, a study examining upward reaching towards a target close to the head (requiring shoulder and elbow flexion as in the FNT), found that persons post-stroke tended to produce concurrent flexion of both elbow and shoulder joints [
30]. In contrast to our results, another study showed lower cross-correlation between shoulder and elbow joints post-stroke [
23]. In that study, however, pointing towards an external target (requiring shoulder flexion and elbow extension) was investigated. Hence, a disrupted interjoint coordination during pointing movements post-stroke may be either an abnormally higher or lower coupling between shoulder and elbow joints depending on the task condition. Reaching upwards to a target is usually easier to perform compared to forward reaching post-stroke, which is probably related to the task requiring a sustained flexor synergy and a shorter lever arm [
30]. Sensorimotor control of egocentric movements (towards the body), as in the FNT, may differ from that of exocentric movements (away from the body) towards an object in extra-personal space. In both eyes-closed cases, the brain has to rely on a body-centered coordinate system instead of an eye-centered coordinate system to locate the target [
35]. However, the sensory feedback in the egocentric reaching task is obtained by proprioceptive information from the upper limb joints including mutual tactile information from the touch creating a sensory mapping that is based on signals from both the fingertip and the nose tip. The source of feedback is hence intrinsic, i.e. arises from the individual’s own sensory systems [
36].
A recent review concluded that trunk restraint is a beneficial method to limit compensatory movements during reaching post-stroke especially for those with moderate-to-severe impairments [
37]. In this study, however, there were no restrictions regarding head and trunk movements since this is not done in usual clinical practice. The stroke group used excessive scapular and trunk motions during the Pointing phase, as revealed by increased displacement of the acromion marker, while their head motion was similar to head motion of control subjects, as shown by equal displacement of the nose marker. During a forward reaching task, persons post-stroke have demonstrated compensatory
anterior sagittal movements such as trunk forward bending [
38] or scapular protraction [
37]. In our study of a point-to-nose task, the stroke group demonstrated a compensatory
posterior sagittal movement. Hence, excessive trunk/scapular movements during pointing or reaching may occur in either direction in stroke patients depending on the task. During forward reaching, it has been shown that trunk sagittal displacement occurs simultaneously with decreased shoulder flexion and decreased elbow extension post-stroke [
20]. In our study, however, there were few significant differences between the stroke group and the control group regarding ROM of shoulder and elbow joints. Unexpectedly, the stroke group did not display increased shoulder abduction during the FNT which is in contrast to results of earlier studies that have investigated hand-to mouth tasks in persons post-stroke [
12,
25]. Within the stroke group, persons with moderate impairments had more marked kinematic deviations from controls compared to persons with mild impairments. Prolonged Pointing time and decreased ROM in elbow flexion during the pointing phase were particularly pronounced for those who had moderate impairments. The latter may have been caused by a more flexed elbow at start position and/or use of excessive scapular and trunk movements during the pointing phase that resulted in less elbow flexion. However, as the mild and moderate impairments were unequally distributed, this should be interpreted with caution.
No variables of the Return phase were entered in the regression model since we focused on the egocentric part of the FNT. Within the stroke group, TMT was highly correlated with smoothness (NMU of the Pointing phase
r
s
= 0.71), a feature that is assumed for well-coordinated movement [
17,
18]. A high correlation between TMT and NMU has also been reported in persons post-stroke during a drinking task, where it was suggested that the TMT could be used as an indirect measure of movement smoothness [
12]. However, a prior study concluded that slower movement speed does not entirely explain the increased temporal segmentation of endpoint movement evident in persons with stroke [
20]. Our stroke group showed a moderate correlation between TPS% and TMT (
r
s
= 0.56). The persons post-stroke had lower amplitudes of Peak speed, prolonged deceleration phases, and left-shifted velocity profiles when moving their finger towards the nose compared to controls. The velocity profiles in our study groups are comparable to those velocity profiles seen in similar groups during a glass-to-mouth task [
12], where controls had only one movement unit while persons post-stroke had multiple movement units and lower peak speeds. Other indirect measures of coordination are movement energy efficiency and accuracy [
39], which were represented in this study by Path ratio and Total variability. Those spatial variables were, however, only weakly correlated with TMT. Notably, TMT did not correlate with IJC, which may seem contradictory for a coordination test such as the FNT.
The strongest associations, according to the multiple regression model, were found between the time to perform the FNT (TMT) and NMU, TPS% and Total variability. Total variability alone was weakly associated with TMT. This indicated that faster performance of FNT in a person post-stroke did not necessarily reflect more precise finger-to-nose contact. It may be that movement accuracy was sacrificed for increased speed, in which case a less accurate movement might risk being misinterpreted as an improvement of coordination [
40].
Although the movement time measured by a stopwatch may be considered as an easy and ‘objective’ measure of the FNT, there are limitations. First, a person with mild stroke may have altered upper limb movements albeit the movement time is comparable with healthy subjects. Second, a reduction in time to perform the FNT between two evaluations may be a ‘false’ improvement as accuracy may be sacrificed in favor of faster speed. As the instructions of the FNT encompass two difficulties; 1) touch nose without vison and 2) as fast as possible, this dual task command may lead to different movement times depending on task priority (cf. Fitts’ law [
41]). Third, the time-monitored FNT corresponds to movement speed and is mainly associated with movement smoothness (temporal end-point variables) but not to joint motions and compensatory movements (spatial variables). For a coordination test, both temporal and spatial aspects are of importance. Fortunately, more recently developed scales aim to assess quality of movement during goal-directed tasks in persons with stroke [
42‐
44]. Such scales are good alternatives or supplements when assessing multi-joint coordination in the stroke-affected arm.
Limitations
The results of this study cannot be extended to a stroke population with severe impairments. Although our stroke sample does not fully represent the broad range of post-stroke hemiparesis, they are an important subpopulation as they may suffer from subtle deficits that are not clearly identified. Additionally, the sample size restricted the number of variables entered in the regression model and the power of the sub-group analysis. At the same time, the sample size (>30 in each group) is considered relatively large for a kinematic study of goal-oriented arm movements in persons post-stroke [
26]. This study was not designed to compare data between stroke groups with different severity, instead we aimed to investigate relationships between movement assessment variables and severity of stroke symptoms. The results indicate that almost half of the selected kinematic variables are indeed sensitive to stroke severity, and we suggest that this should be evaluated in a larger stroke population. Because coordination of reaching is complex, multiple variables were included in the analysis to represent different movement characteristics. To counteract the problem of multiple comparisons, a Bonferroni correction was employed. Although this correction may be a simple and effective way of avoiding Type I errors (detecting a difference that is not truly present), it is very conservative and therefore strongly increases the risk of Type II errors (failing to detect a difference that is present) [
45]. Finally, the results of this study are based on kinematic outcomes of the pointing phase and do not take into account the movement performance throughout the whole test.
Acknowledgements
All the participants are gratefully acknowledged. We also thank Gunilla Elmgren Frykberg and Monica Edström for assistance in data collection, Jonas Selling for technical assistance, and Lina Schelin for statistical advice. Thanks also to Mercado Medic for the loan of a chair.