Aging-related decrements during specific phases of the dual-task Timed Up-and-Go test

Background It is unclear how young and older adults modulate dual-task mobility under changing postural challenges. Aim To examine age-related changes in dual-task processing during specific phases of dual-task Timed Up-and-Go (TUGdual-task). Method Healthy young and older adults performed the Timed Up-and-Go (TUG) with the following dual-task conditions: (1) serial-three subtractions, (2) carrying cup of water, (3) combined subtraction and carrying water, and (4) dialing cell phone. The primary outcome was the dual-task cost on performance of TUG (percent change from single- to dual-task) based on duration and peak trunk velocity of each phase: (a) straight-walk, (b) sit-to-stand, (c) turn, (d) turn-to-sit. Mixed-design univariate analysis of variance was performed for each type of task. Results Older adults had more pronounced mobility decrements than young adults during straight-ahead walking and turns when the secondary task engaged both cognitive and manual modalities. Simple cognitive or manual tasks during TUGdual-task did not differentiate young from older participants. Subtraction performance during simple and complex cognitive conditions differed by phase of the TUG. Manual task performance of carrying water did not vary by phase or age. Discussion Our findings suggest that dual-task processing is dynamic across phases of TUGdual-task. Aging-related dual-task decrements are demonstrated during straight-ahead walking and turning, particularly when the secondary task is more complex. Conclusion Older adults are susceptible to reduced dual-task mobility during straight-ahead walking and turning particularly when attentional loading was increased.


Introduction
Gait in older adults is compromised during dual-task conditions [1,2], such that speed and stride length are reduced, and stride time and its variability are increased [3,4]. Dualtask-related gait decrements can lead to instability and increased fall risk [5]. Falls are the leading cause of accidental deaths among older adults [6], thus understanding mobility and fall risk may reduce this burden. Cognitive functions, particularly attention, are necessary during gait; hence, gait decrements during dual-tasks can be explained by limited capacity of attention processing, or due to competition for cognitive resources [7,8].
Research on dual-task gait is largely based on studies examining straight-ahead walking, yet most daily activities require transition movements, such as turns and sit-tostand. Mechanics of turning deteriorate with age [9] wherein a simplified turning pattern can predict recurrent falls in the elderly [10]. The instability during turns is likely due to the unique physiologic and cognitive requirements of turns relative to straight-ahead walking [11,12]. For example, cognitive processing speed was found to be uniquely associated with curvilinear walking but not with straight-ahead walking [11]. It remains unclear how older adults manage dual-task mobility when the postural requirements of component tasks differ, as in linear and curved walking.
The Timed Up-and-Go test (TUG), a clinical test of mobility and fall risk in older adults, includes straightahead walking and transitions [13]. Dual-tasks have been integrated into the TUG (TUG dual-task ) [2,14]; however, these studies analyzed the total TUG duration rather than its individual phases, and thus may not be successful in assessing at-risk older adults. For instance, Shumway-Cook et al. [14] demonstrated that adding a dual-task challenge failed to enhance falls prediction more than that of the regular TUG. Although based on the conventional TUG, Mirelman et al. [15] demonstrated that the total TUG duration failed to differentiate older adults with and without cognitive impairment, but performance in specific phases did. Further, kinematic data during specific phases were found to be independent of the total duration [16]. No study thus far has examined phases of the TUG dual-task . Understanding dual-task behavior during individual phases may reveal subclinical mobility changes that may assist in early and targeted intervention.
The first objective was to examine age-related decrements in mobility during phases of the TUG dual-task . We hypothesized that older adults would demonstrate greater dual-task decrement of duration and peak velocity during transitions of the TUG compared to young adults. The second objective was to characterize dual-task performance during specific phases of the TUG when engaged in simple and complex secondary tasks. We hypothesized that dualtask decrements of duration and peak velocity will be greater during transitions and when secondary tasks are more complex.

Participants
Twelve healthy young adults (mean ± standard deviation, M ± SD: 26.13 ± 5.36 years) and 12 older adults (M ± SD: 74.18 ± 5.21 years) (see Table 1) were recruited from the community and university population. Physical therapists performed medical history taking, clinical screening of gross mobility, and vibration testing of the foot and ankle. Participants were included if they were able to independently ambulate in the community, follow instructions in English, and tolerate a 2-h testing session. Exclusion criteria were impaired vibration sense of feet or ankles, diagnosis of dementia, or any neurological, orthopedic, or medical condition that impaired walking. The Institutional Review Board approved this study, and participants provided signed informed consent.

Apparatus
Movements were recorded using six wireless inertial sensors (Opal TM , APDM, Portland, OR, USA), with dimensions of 48.4 9 36.1 9 13.4 mm. Sensors were secured to wrists and ankles bilaterally, and mid-thoracic and lower lumbar areas using Velcro straps. Trials were audio-video recorded for analysis of secondary tasks.

Tasks and procedures
Mobility was assessed using the 7-m TUG, instrumented with inertial sensors (iTUG) (Fig. 1) [17]. The single-task iTUG was performed as follows: upon cue, the subject stood up from a chair without hand support (Sit-to-Stand), walked straight-ahead 7 m, turned around (Turn), walked back to the chair, turned and sat down (Turn-to-Sit). Periods of walking to and from the 7-m mark were consolidated into a single straight-ahead walk (Straight-Walk). The phases of interest were Straight-Walk, Sit-to-Stand, Turn, and Turn-to-Sit. The single-task iTUG served as reference for each participant's dual-task performance. We selected four conditions for the TUG dual-task : (1) serialsubtraction by 3 0 s from a random number between 70 and 99 (COUNT); (2) carrying cup of water filled up to 1 cm below rim (CARRY); (3) combined COUNT and CARRY (CtCARRY); and (4) dialing home phone number with cell phone (DIAL). The DIAL condition served as alternative to CtCARRY, examining the effect of combined cognitivemanual processing, without structural interference from water dynamics seen in CtCARRY. Because integration of cognitive and manual tasks was required during CtCARRY and DIAL, these were considered complex tasks, while COUNT and CARRY were considered simple tasks. Subjects were instructed to walk as quickly as possible, and perform the secondary task as quickly and/or accurately as possible. Each subject performed three trials per condition in pseudo-randomized order.

Outcome measures
Our outcome measures were phase-specific duration and peak velocities of the iTUG dual-task , because these measures were available for each phase, thus allowing comparison of similar mobility constructs [18] across phases. Measures were obtained through Mobility Lab TM software (APDM, Eugene, OR) utilizing algorithms by Salarian et al. [17,19]. Duration (s) refers to the time to complete each phase, while peak velocity (°/s) refers to 95 % of peak angular velocity of trunk per individual phase. Next, we calculated the dual-task cost (DTC), defined as the percent change in performance relative to an individual's single-task performance [20]. This normalized the data and distinguished dual-task processing from usual agerelated changes. The DTC was computed for duration (Eq. 1) and peak velocity (Eq. 2), with negative and positive multipliers used, respectively, for directionality of performance decrement.
The greater DTC value in the negative direction implied greater performance decrements.
To assess manual performance, we recorded the number of spills per phase during simple (CARRY) and complex (CtCARRY) manual conditions. To assess cognitive performance, we examined response rate (Eq. 3), and response accuracy (Eq. 4) during simple (COUNT) and complex (CtCARRY) cognitive conditions. Both outcomes were expressed in percent for simplicity of interpretation.
Recent studies [20][21][22] examined both single-and dualtask performance of the secondary tasks to assess interplay of gait and secondary tasks. Our study did not adopt this methodology because we were interested in phase-specific performance rather than DTC of the entire TUG, and this methodology may not necessarily assist in answering questions related to phase-specific performance. Instead, we compared phase-specific change in performance of the secondary tasks between simple and complex conditions.

Statistical analysis
SPSS (Version 22) was used. Group means are reported as M and its 95 % confidence interval (CI). To determine the effect of age (Young, Old) and phase (Straight-Walk, Sitto-Stand, Turn, Turn-to-Sit) on DTC of mobility measures, linear mixed models were utilized on duration and peak velocity per phase. We used a mixed-design univariate analysis of variance (ANOVA) with random-nested factor of subject, and fixed factors of age, group and phase. To examine differences in phases, post hoc pairwise comparisons were performed using Bonferroni corrections. Separate analyses were performed for each secondary task. Similar univariate ANOVA, as described above, was used to examine manual and cognitive performance, with analysis performed separately for each outcome per condition.

Subjects
As summarized in Table 1 Fig. 3). Thus, duration-related decrements were higher for older adults when secondary tasks were complex, but not during simple conditions. Peak velocity decrements were comparable between young and older adults.
Post hoc analysis revealed that dual-task decrement on duration was particularly evident for older adults during Straight-Walk and Turn phases during CtCARRY and DIAL (Fig. 2c, d), but not during COUNT and CARRY conditions (Fig. 2a, b). Furthermore, DTC on peak velocity during Straight-Walk was worse for older than younger adults during CtCARRY and DIAL (Fig. 3c, d). Taken together, this suggests that older adults dampened trunk velocities and extended duration during Straight-Walk, as well as lengthened Turn phases when secondary tasks required integration of cognitive and manual processing.

Discussion
To our knowledge, this is the first study that investigated aging-related dual-task behavior during specific phases of the iTUG. Our results demonstrated pronounced aging-related decrements impacting duration of Straight-Walk and Turn phases, and peak velocity during Straight-Walk, particularly during complex conditions requiring cognitive-manual integration. These findings suggest that attentional processing is different across phases of an activity.

Dual-task processing depends on phase of iTUG
Previous studies have shown that dual-task performance declines when postural challenge increases. Attentional demands were shown to progressively increase from sitting, standing, to walking [24]; or from standing to stair negotiation [25]. These studies, however, examined dualtask performance under separate tasks of varying postural challenge. Notably in our study, dual-task performance was assessed based on a sequence of tasks with different postural challenges, a scenario consistent with daily activity. Our findings suggest that attention processing is not at steady state throughout an activity; rather, it is different across phases of an activity. Consistent with our hypothesis, older adults had pronounced decrements in duration during Turn particularly during complex conditions; however, we also find pronounced decrements in Straight-Walk. The high dual-task deficit in Straight-Walk seems less intuitive because straight-path walking is a well-learned task, thus should not highly tax attention resources. This may be explained by task prioritization. Straight-Walk may be regarded as low threat to stability relative to other transition phases. Thus, older adults may have diverted attentional resources to the secondary task at the expense of walking just because they can afford to do so [26]. Although we did not examine prioritization tradeoff, our results show that cognitive performance in COUNT and CtCARRY was improved during Straight-Walk relative to other phases, but at the expense of gait. Similarly, Patel et al. [22] demonstrated improved cognitive performance in young adults but at the expense of gait speed.
Age-related changes in grasp control can make Straight-Walk more complex. Diermayr et al. [27] demonstrated that coordination of grasp forces was compromised in older adults during challenged gait (obstacles) but not during regular walking, suggesting resultant deficits when other factors such as balance and attention were challenged. Similarly, in the current study, while not a grasp experiment, older adults had pronounced decrements in duration and peak velocity during Straight-Walk when conditions were complex (CtCARRY, DIAL). Manual performance across phases did not differ per group or condition (CARRY, CtCARRY), suggesting that dualtasking mostly impacted gait and not the manual task. The findings of Diermayr et al. [27] along with our results suggest that modulation of gait and grasp control could be challenging even during straight-ahead walking when attentional loading is increased.
The longer time interval for Straight-Walk phase may be another factor for worse DTC. Straight-Walk, by design, was of longer duration than other phases in the TUG, and thus greater time may have provided more opportunity for errors or corrective responses (as in the subtraction task). It is possible that the DTC is time-sensitive, such that a more comprehensive assessment can be made when observations occur over a longer time interval.

Dual-task processing depends on secondary task
Earlier reports have suggested that the type of task does not uniquely determine the extent of DTC [28]. For instance, previous studies demonstrated that cognitive (serial subtractions) and manual (coin transfer) secondary tasks rendered similar effects on gait [28,29]. Our findings, however, revealed differentiation, such that CARRY generally caused intermediate DTC between COUNT and the combined tasks (CtCARRY, DIAL). Our study further expanded the conditions by integrating cognitive and manual modalities, allowing use of a broader range of secondary tasks common in daily activities. The greatest dual-task costs to the TUG occurred during CtCARRY and DIAL conditions, both of which comprised manual and cognitive components. Thus, the minimal to moderate DTC during COUNT or CARRY became additive when the secondary task involved both cognitive and manual components.
A significant interaction of age and phase on Straight-Walk duration was found in CtCARRY and DIAL. However, peak velocity had a significant interaction only for CtCARRY, but not for DIAL, likely due to differences in structural interference (water versus rigid object such as a phone). Overall, the greatest dual-task cost to iTUG was seen during more complex tasks, suggesting that type and complexity of secondary tasks matter in their effect on walking.

Secondary task performance
Cognitive tasks may impact gait, just as gait may perturb cognitive performance [1]. We found that subtraction rate and accuracy was greatest during Straight-Walk, and was most impaired during Sit-to-Stand regardless of complexity of the secondary task (COUNT, CtCARRY). Overall, cognitive performance was similar across age groups. Hall et al. [30] demonstrated that cognitive factors could explain dual-task walking performance only when the cognitive challenges were sufficiently complex. Our older subjects had lower MoCA scores compared to young; nevertheless, both performed similarly in the subtraction task. Therefore, the cognitive task may not have been complex enough to load the cognitive systems in older adults. What may differentiate young and older adults in this study is dual-task performance during the iTUG.

Limitations
Performance during the dialing task was not assessed due to technology limitations. This study examined a relatively small sample size; therefore, findings need to be interpreted cautiously.

Conclusion
Attentional processing is different across phases of complex functional activities like the TUG. Older adults are more susceptible to dual-task mobility decrements during straightahead walking and turning, particularly when secondary tasks require integration of cognitive and manual modalities. Examination of these phases during clinical testing may assist clinicians in identifying at-risk individuals.