This was a cross-sectional study approved by the local ethics committee (ethical approval number: HSEARS20160202006). The study objectives and assessment procedures were clearly explained to the participants, who provided written informed consent to participate before the beginning of the study. This study was conducted according to the guidelines of the Declaration of Helsinki.

Although no study has investigated the reliability of LCT in people with stroke, this test was shown to have excellent test–retest reliability [intra-class correlation (ICC) = 0.92] in people with knee OA (3). Assuming an ICC cut-off value of 0.75 for assessing test–retest reliability in stroke survivors, a sample size of at least 21 subjects would be required to achieve 80% power to detect an ICC of 0.9, with an ICC of 0.8 as the null hypothesis and a significance level of 0.05. The sample size was estimated using an online calculator (7).

Additionally, no study has investigated the correlations between LCT time and stroke-specific outcome measures in people with stroke. Assuming that LCT time is significantly and moderately strongly correlated (ρ = 0.5) with the selected stroke-specific outcome measures, a sample size of 21 subjects would be required to achieve 80% power and a significance level of 0.05. The sample size was estimated using G∗Power software, version 3.1.9.7 (Franz Faul, University of Kiel, Kiel, Germany).

The estimations indicated that a minimal sample size of 22 would be required to meet the study objectives. To increase the power of this study, 24 subjects with stroke were recruited for assessment of the psychometric properties of LCT.

Thirty subjects of stroke and 30 healthy older adults were recruited from the community via flyer. Six subjects with stroke failed to conduct the LCT. Thus, 24 subjects of stroke were included in our study (Figure 2). Participants were included if they (i) were aged 55 years or older; (ii) had a post-stroke duration>6 months; (iii) were medically stable; (iv) could walk 10 m without physical assistance from another person, with or without walking aids; and (v) had an Abbreviated Mental Test score ≥ 7 (8). Potential participants with any neurological or cardiovascular diseases that might influenced the assessment were excluded, such as: Parkinson’s disease, Alzheimer’s disease, Multiple Sclerosis and so on.

Twenty-four healthy older adults (3 men, 21 women) aged 55 years or older were recruited as controls.

To investigate the reliability of LCT in people with stroke, two LCT trials, each comprising one practice and 3 timed tests, were conducted 7 days apart to minimize the learning effect and natural recovery among the participants (Figure 3). Two raters, A and B, were trained at least 2 weeks prior to the start of study. LCT times were measured by both raters simultaneously, using digital stopwatches. The mean time of the 3 timed trials was recorded.

The healthy controls completed the LCT on day 1. The LCT cut-off score for people with stroke was determined.

Data analysis was conducted using SPSS software, version 26.0 (IBM Corporation, Armonk, NY, United States). The normality of the data and the homogeneity of variances were checked using the Kolmogorov–Smirnov Normality and Levene’s tests, respectively. Demographic data are summarized using mean difference (MD) and standard deviation (SD). The between-group differences of the baseline characteristic among the stroke patients and healthy control were analyzed using independent t-tests for parametric data and the chi-square and Mann–Whitney U-tests for non-parametric data (18).

A two-way mixed-effect model (ICC_3,1) was used for intra-rater and test–retest reliabilities, as the tests involved 2 independent raters, each of whom recorded 3 trials for each participant. ICC_3,2 was used for inter-rater reliability as the mean of 3 measurements recorded by a single rater on the same day was compared with the mean of the measurements recorded by the other rater (18). The strength of ICC determined using a general guideline wherein values of <0.5, 0.5–0.75, >0.75–0.9, and > 0.9 indicate poor, moderate, good and excellent reliability, respectively (19). Pearson correlation coefficients and Spearman’s rho is used to examine the correlation between LCT time and the other outcomes, as proper. The p-value, which indicates significance, was adjusted to 0.0125 (0.05/4) after Bonferroni adjustment for 4 outcome measures which commonly used to assess different domains of function in people with stroke, which included FMA, BBS, TUG and CIM. These outcome measures were chosen because FMA is a stroke-specific assessment related to lifting and carrying tasks (4, 9, 11), and because strong research evidence supports the excellent reliability of BBS, TUG and CIM, which are also related to walking tasks, in people with stroke (5, 6, 13, 16, 17, 20–22). The secondary outcome measures were ankle plantarflexor and dorsiflexor muscle strength.

The MDC at a 95% confidence level were calculated as follows (23):

SEM is the standard error of measurement of LCT time and was calculated as follows:

where S is the pooled standard deviation of LCT time, and r is the reliability coefficient.

The optimal LCT time cut-offs for differentiating between people with stroke and healthy controls were determined using receiver operating characteristics (ROC) curves. The cut-offs were calculated using Youden’s index to identify the trade-off point between the highest sensitivity and specificity. The area under the curve (AUC) was used to calculate discrimination accuracy.

This is the first study to determine that LCT time has excellent intra-rater, inter-rater and test–retest reliabilities and is significantly correlated with other stroke-specific outcome measures. It is also the first study to determine the optimal cut-off LCT time for differentiating performance of LCT between people with stroke and healthy controls.

As noted, LCT exhibited excellent intra-rater, inter-rater and test–retest reliabilities. A study also indicated that LCT has good to excellent test–retest reliability in people with knee OA and hip (ICC = 0.92 at 14 days, ICC = 0.77 at 3 months) (2). This excellent reliability may be attributable to the standardization of the testing procedure and a high level of consistency in various rater and participant aspects. Our raters were well-trained in performing the measurement and strictly followed the formulated test procedures. Clear instructions were delivered to the participants, whose comprehension was ensured. The 7-day test–retest span partly eliminated the practice effect and minimized the cultivation effect on the participants’ performance (26).

People with stroke had a nearly two-fold significant longer mean LCT time than did healthy controls. LCT is an assessment of advanced functional activity and assesses complex and coordinated upper and lower limb motor control, as well as sensory and cognitive function. LCT comprises sequential tasks involving walking, bending down, picking up objects and turning (2). Previous studies (27, 28) found that motor sequence learning, which is suggested to be dominated by the working memory (29), was impaired after stroke. Impairments in motor control function and cognitive function following stroke may explain why people with stroke needed a longer time to complete the LCT than healthy controls (30).

In fact, our findings were consistent with those of previous studies on the performance of individual components of LCT (2). People with stroke were found to walk at a slower speed (stroke = 0.88 m/s, healthy = 1.49 m/s) and a shorter distance (stroke = 261.5 m, healthy = 660.1 m) than those of healthy controls on level ground (31), and had difficulties in picking up objects, requiring 50% more time to complete trunk–scapular forward and backward bending than healthy controls (32). Moreover, in one study, people with stroke required 3 more seconds to turn than healthy controls (33). Poor post-stroke muscle strength resulting from reductions in the firing rates and number of fast-twitch fibers; poor motor control, coordination and balance; and spasticity may explain why people with stroke took longer than healthy controls to perform each movement in LCT (2, 34). In addition, sensory deficits can also lead to an increase in LCT time (35). Sensorimotor integration is affected in people with stroke, which may disrupt coordination of the gaze and posture while turning. Such deficits may have led to a large discrepancy in LCT times between the groups in our study.

Apart from motor control, the sequential tasks in the LCT place a cognitive demand on the individual. Studies have demonstrated that reductions in gait and balance in people with stroke during dual tasks were due to cognitive–motor interference (36). Some other studies (27, 28) also supported that motor sequence learning was impaired after stroke, which can affect their ability to perform motor skill with complex sequence of movement. During the LCT, the participants in our study were required to remember to pick up an object from the lower shelf, walk in a fixed sequence while carrying it, and return it to the higher shelf. It is possible that the participants with stroke may have taken longer to complete the test, in part, due to a higher cognitive demand related to the task.

The MDC of the LCT time was 2.32 s. In this study, the actual mean difference in LCT time between stroke survivors and healthy controls, 16.00 s, was much greater than the MDC and thus indicates an actual difference in functional ability rather than a measurement error.

LCT was originally designed to examine the functional limitations of people with knee OA, as people with severe OA experience greater dysfunction when lifting and carrying objects than people with a less severe condition (37). In a comparison with the findings from previous studies, people with stroke in our study were found to require a much longer mean time to complete LCT (29.70s) than people with knee OA (11.0 s). This difference may be attributed to greater dysfunction in people with stroke. In people with knee OA, walking ability is only impeded by musculoskeletal problems in the lower limbs. In contrast, people with stroke are also restricted by a spectrum of stroke-specific deficits, including upper and lower limbs, trunk control and cognition (2, 34, 35).

LCT time was not found to be significantly correlated with FMA total score. FMA was originally designed to assess motor impairment in people with stroke. This assessment of the motor control, reflex and coordination of the upper and lower extremities is performed in static positions, such as lying, sitting or standing (9). In this study, the mean FMA-UE and FMA-LE score was 39.00 (16.65) and 24.88 (6.52), respectively, which indicated moderate to upper limb (24) and lower limb motor control (25) of the affected side. However, LCT is an advanced functional task that places both physical and cognitive demands on the participant. For example, during LCT, participants perform dynamic transitional movements, such as trunk bending and walking, and are required to follow multi-step commands, as mentioned previously (2), and these requirements pose challenges to participants’ memory and cognitive skills. Moreover, there are no limitations on participants’ use of their upper extremities to pick up the 4.5-kg object during the LCT, which means that they can use their unaffected upper extremity. Rather than Solely focusing on the motor performance of the affected side as assessed by FMA, the LCT more likely to consider how people with stroke can modify and adjust their performance of multiple ADL task with the motor function of both the affected and unaffected side. These may explain the lack of correlation between LCT time and FMA total score.

No significant correlation was shown between the ankle muscle strength on either the affected or unaffected side and LCT time. A significant association between lower limb muscle strength, especially the ankle dorsiflexor, and walking speed in the swing phase has been reported. The ankle dorsiflexors are important during the swing phase (38), as they allow for sufficient foot clearance. Indeed, LCT is a sequential advanced functional assessment that places both physical and cognitive demands on the participant. The insignificant correlation between ankle muscle strength and LCT time may indicate that the cognitive function may play a more important role than the ankle muscle strength in the LCT completion time. In addition, the LCT is a multiple task required the coordination of both the affected and unaffected limbs, which is far more complicate than the pure muscle contraction in the unilateral muscle testing. It is reasonable that there was no significant correlation found between LCT times and muscle strength. Further complementary study should include the assessment tool of cognitive function, so that to determine the role of cognitive function in the LCT completion time.

LCT time was found to have a significant negative correlation with BBS score. BBS includes tasks such as turning (items 10, 11) and picking up objects from the floor (item 9), which are highly related to several skill components in LCT (e.g., bending to pick up a 4.5-kg weight and turning). These similarities may explain the observed correlation between LCT time and BBS score.

No significant correlation was observed between LCT time and CIM score in this study. There are several reasons which may contribute to this insignificant relationship between LCT time and CIM score. First, the measurement domains between LCT and CIM is different. The CIM is a subjective measurement of community integration that considers the self-reported subjective feelings of participants (2) while the LCT is an objective measurement of sequential motor tasks in a laboratory setting (15), which may not reflect the subjective feelings of participants in real-life situations. Second, previous studies (39, 40) have proven that motor function alone is insufficient in predicting the community reintegration of people with stroke. Other factors, including poststroke depression, personal perceptions, and stroke recovery, are also important predictors of community reintegration in this population (r = 0.499–0.743, p < 0.01)(39, 40). Third, 3-factor structure had been identified in CIM for use in stroke survivors living in Hong Kong by Principal Component Analysis, including “relationship and engagement,” “sense of knowing” and “independent living” (17). The LCT is an objective measurement of sequential motor tasks which relates on “independent living,” but not on “relationship and engagement,” “sense of knowing.”

This study has several limitations. First, only LCT time was measured in this study. Although the LCT completion time is easy to obtain and easy-to-administrate, it may not be valid to assess the quality of movement. The quality of the participants’ movement and how they used compensatory strategies to execute the tasks were not examined; therefore, future studies might further investigate the qualitative aspects of LCT. Second, the measure of intra-rater reliability was constrained to 10 min after the first trial. Accordingly, this measure could only reflect excellent test–retest reliability over a short timespan across trials, and therefore, the short-term learning effect may have affected the results. Future studies may consider using a longer test–retest time span. Third, the cut-off score may not be generalizable due to the significant difference in sex ratio between the groups. Studies (41, 42) have reported significant differences in muscle strength and motor performance between male and female participants, and such gender-relate differences may also contribute to the significant difference of the motor function between the stroke and health groups, and eventually lead to the significant difference of the LCT performance. Precaution should be taken before interpreting the result. To enlarge the subject pool to ensure the equivalent gender-ratio between the 2 groups in the future study could help to address this limitation. Fourth, a sample size of 22 was estimated for this study in accordance with previous studies involving people with knee osteoarthritis. The conclusion drawn from our study may only applicable to the subject who meet our inclusion criteria. We may not have sufficient power to confirm whether there was a potential correlation in some of the outcomes. Fifth, the LCT is a multidimensional outcome, which encompasses tasks related to walking ability, upper and lower limb function, strength, balance, and cognitive function. This confluence of diverse factors implies a violation of unidimensionality of the outcome measures, thereby prompting the variation of the latent variable being investigated by LCT completion time in people with stroke. By elucidating the relationship between LCT performance and various clinical measures in the future study, it becomes conceivable to ascertain the specific functional abilities and limitations captured by LCT. Finally, people with stroke were recruited from the community for this study. These participants generally had better walking capability and functional ability, which may have led to selection bias. In summary, future studies aiming to investigate the concurrent and predictive validity of LCT should consider using larger sample sizes and including a more diverse population of people with stroke in terms of walking ability, which will enable generalization of the results to all people with stroke.