Abstract
Objective:
This study compares dual task (DT) gait speed between robust and frail older adults in the setting of a community-based frailty screening program.
Methods:
A cross-sectional study was conducted involving older adults aged 50 years or older. Participants were categorized as robust, pre-frail or frail using the FRAIL phenotype. Gait speed was measured under single task (ST) and 2 DT conditions: arithmetic (DT1) and verbal fluency (DT2) tasks. For the arithmetic task, participants counted upwards in multiples of 2, starting from 16. For the verbal fluency task, they named different animals as they walked.
Results:
A total of 976 participants were recruited, of which 779 were robust, 181 were pre-frail and 16 were frail. Robust participants exhibited significantly faster DT gait speeds (DT1: 1.12 (0.25)m/s; DT2: 0.97 (0.27)m/s than pre-frail/frail (PFF) participants (DT1: 0.88 (0.30)m/s); DT2: 0.78 (0.30)m/s). Robust participants also demonstrated a significantly faster ST gait speed (1.31 (0.22)m/s) vs 1.05 (0.31)m/s).
Conclusion:
DT gait assessment combines both physical and cognitive domains and may serve as a more holistic tool for early frailty detection in the community. The significant differences observed in DT gait speeds highlight the need for integrating cognitive assessments into frailty screening protocols, supporting timely interventions to improve health outcomes.
Introduction
Frailty is a prevalent geriatric syndrome characterized by increased vulnerability to stressors due to diminished physiological reserves, leading to adverse health outcomes such as mortality, hospitalization, falls, and eventual institutionalization. 1 Interventions involving exercise and nutrition may slow or even reverse frailty progression in older adults. 2 Frailty screening is often conducted in the wider community. Here, individuals may appear outwardly healthy, leading to limited healthcare touchpoints. This poses practical challenges, as frailty assessments need to be objective, simple to perform in non-laboratory settings, yet simultaneously stressing inadequate reserves sufficiently to result in a measurable change.
Dual-task (DT) gait paradigms involve simultaneously performing a cognitive task while walking. The underlying hypothesis is that gait and cognitive tasks utilize similar cortical resources of the brain. This stresses the brain with lower reserves due to the simultaneous activation of multiple cognitive and motor brain areas. 3 There are 2 commonly proposed models to explain the effect of DT on performance: the central capacity sharing model 4 and the bottleneck model. 5 The central capacity sharing model posits that tasks are processed in parallel, 4 while the bottleneck model proposes that only 1 task can be processed at a time. 5 Both theories suggest that one’s ability to DT is limited by one’s central processing capacity.
Frailty and cognitive impairment share similar pathophysiological mechanisms, as both are linked to deficits in executive function, necessary for both gait and cognitive performance. Frail individuals with lower executive reserves may exhibit poorer DT performance, indicating an intertwined relationship between cognitive and motor decline.3,6,7 This relationship suggests that diminished DT performance can serve as an early marker for both motor and cognitive decline in frail populations.
The additional stress introduced by the DT can be quantified using DT cost, defined as the percentage change in gait speed observed from a single task (ST) to a DT scenario. 8 Two common categories of cognitive tasks utilized are mental tracking and verbal fluency tasks, which engage different cognitive processes, 9 and hence affect one’s DT performance to different extents.
While multiple studies have examined the effect of DT on gait performance and its relationship to frailty, there are significant gaps in the current literature. Some studies incorporated strict exclusion criteria10 -15 while others utilized specialized equipment like motion trackers12,15,16 or electronic walkways.13,14 These factors limit the extent to which the findings can be generalized to a community setting, where individuals often have multiple co-morbid conditions. Specialized equipment is often expensive and requires trained personnel to operate, making it inaccessible and impractical to incorporate into large-scale frailty screening programs.
There is also a strong case proposition for DT screening in frailty assessments as it is functionally relevant. DT gait activities are common in the older population, such as crossing a traffic junction while carrying grocery bags or conversing with others. 17 DT gait assessments become crucial, especially when these tasks are performed under time constraints, such as crossing a road. Also, DT gait impairments are amenable to rehabilitation by means of specific DT training. When integrated with other modalities of physical therapy, DT training has been shown to significantly improve one’s ST and DT gait performance, balance, and cognition, which are associated with decreased risk of falls and improved ADL performance. 18
The primary aim of this study is to compare the DT gait speed between robust and pre-frail/frail (PFF) older adults for 2 different cognitive tasks, with DT gait speed being the primary outcome measure. The working hypothesis is that there is a significant difference in DT gait speed between robust and PFF older adults for both DTs assessed. There are multiple secondary aims of this study, the first of which is to compare the DT cost between the 2 groups for both DTs. Secondly, the receiver operating characteristic (ROC) curves of ST and DT gait speed will be compared to assess the potential utility of DT gait speed in frailty screening in the community. Lastly, this study aims to examine the relationships between DT gait speed, cognition, mood, and various measures of functional status through the utilization of a correlation matrix analysis.
Materials and Methods
Study Description
Participants were recruited through the Individual Physical Proficiency Test for Seniors (IPPT-S), an ongoing comprehensive community-based frailty screening program aimed at promoting fitness, and preventing or delaying frailty progression in older adults. 19 From June 2018 to December 2023, 976 patients have since been recruited and included in this cross-sectional analysis. Ethics approval was obtained from the SingHealth Institutional Review Board (SingHealth CIRB 2018/2115, Singapore), and all participants provided written informed consent.
Screening sessions were held at various senior activity centers and community centers in the north-eastern region of Singapore. Eligible participants were ≥50 years old, community-dwelling, and could ambulate independently (with or without walking aids). There were no specific exclusion criteria, and all participants who were physically able to complete the ambulation task safely were eligible for the screening program. All participants underwent a multi-domain geriatric screening and physical fitness assessment, administered by a trained study team member.
Single Task (ST) Gait Speed
Gait speed was measured by a 10-m walk test, with an additional 2 m before and after to account for acceleration and deceleration respectively. This additional distance accounts for the gait initiation and acceleration phases, allowing one to achieve steady-state walking. 20 Participants were allowed to use a walking aid if necessary. A stopwatch was used to measure the time taken to walk 10 m, and gait speed was calculated accordingly.
For the ST, participants were instructed to walk at their comfortable walking speed. Participants performed 2 consecutive attempts to familiarize themselves with the testing procedure, and the faster gait speed was used for analysis. This preliminary trial also allowed for assessors to establish participants’ ability to ambulate safely.
Dual Task (DT) Gait Speed
The DT gait speed was conducted similarly to the ST gait speed, with the addition of a cognitive task. There are 2 categories of DT commonly used: mental tracking and verbal fluency. Mental tracking tasks require one to hold information in the mind while performing a mental process, and are commonly used to examine sustained attention, information processing speed, and executive function. 21 Verbal fluency tasks require spontaneous word production under specific search conditions and are also thought to examine executive function. 21 There are currently no established standards for DT gait assessment. Existing literature advises DT selection based on the tested cohort, taking into consideration background diseases and other socio-demographics such as education level.
To represent a mental tracking task, we chose an arithmetic task 22 (DT1). Standardized instructions were given to walk while counting upwards in multiples of 2, starting from 16. We believed that this DT was simple enough for the older adults to be familiar with in daily practice (eg, counting money) and independent of educational level while being sufficiently challenging in the attentional, concentration, and problem-solving domains of cognition. For the verbal fluency task (DT2), participants were asked to name different animals as they walked. This category was chosen for its universality, making it suitable for participants across diverse demographic backgrounds and educational levels. Participants attempted the DTs only once to prevent familiarity and weakening of the DT effect.
From this, DT cost was calculated with the formula: DT cost = [(ST gait speed − DT gait speed)/ST gait speed] × 100, to obtain a percentage change in gait speed with the addition of DT. A more positive DT cost value represents a greater decline in performance with the addition of the DT. For example, an individual who has a ST gait speed of 1.0 m/s and DT gait speed of 0.8 m/s would have a DT cost of 20%, indicating a 20% reduction in gait performance when performing the DT.
Frailty Status
Frailty status was determined by the FRAIL score, a simple 5-item self-reported questionnaire that has been validated in multiple populations.23,24 The 5 domains assessed were
Measures of Other Domains
Cognition was measured with the Chinese Mental Status Examination (CMMSE), a 28-point assessment validated to screen for cognitive impairment in Singapore’s geriatric population. 25 Cognitive impairment was defined as CMMSE ≤ 21. 25 Mood was assessed with the 15-item Geriatric Depression Scale (GDS), a self-reported questionnaire designed to screen for depression in older individuals. 26 We used a cut-off threshold of GDS ≥5 to signify at least mild depression. Functional status was measured with the Barthel Index (BI) and Lawton and Brody’s Instrumental Activities of Daily Living (IADL) scale, which quantify one’s level of physical disability and functional decline respectively.27,28 The number of falls in the past year was documented based on participants’ self-reports.
Statistical Analysis
The sample size was calculated with G*Power version 3.1.9.6. To achieve an effect size (Cohen’s d) of 0.23 with a statistical power of 80% and a significance level (α) of 5%, the calculated sample size using a 1:4 allocation ratio is 186 participants in the PFF group and 744 participants in the robust group. This sample size allocation ensures that the study is adequately powered to detect the hypothesized effect size within the specified statistical parameters.
Statistical analysis was conducted using IBM SPSS Statistics version 29. Independent sample t-tests were used to compare the mean ST and DT gait speeds and DT cost between groups based on frailty status. Cohen’s d was used to estimate the effect size between the robust and PFF groups. Descriptive statistics for ST and DT gait speed and DT cost in each group were calculated and reported as means with standard deviations (SD).
A receiver operating characteristic (ROC) curve was generated to evaluate the ability of DT gait speed to predict frailty status based on the FRAIL score, which was dichotomized as PFF or robust. This was created by plotting the sensitivity (true positive rate) against (1-specificity) false positive rate across various threshold values, providing a visual assessment of DT gait speed’s discriminative performance. The area under the curve (AUC) is interpreted as the probability to correctly discriminate between 2 individuals with different frailty status, where 1.0 represents perfect discrimination and 0.5 represents no discrimination. Additionally, associations between ST and DT gait speed and measures of cognition, mood, and function were examined using the Pearson correlation coefficient (r).
Results
Participant Characteristics
The main characteristics of the study population are presented in Table 1. A total of 976 participants were recruited, with a mean age of 68.0 (6.8) years, of which 727 (74.5%) were female. There were 779 robust participants (79.8%), 181 (18.5%) pre-frail participants, and 16 frail participants (1.6%). Cognitive impairment was found in 97 (9.9%) participants, while 160 (16.4%) participants had at least mild depression.
Summary of Participant Demographics and Characteristics, Presented in Mean (SD).
Abbreviations: BMI, Body Mass Index; CMMSE, Chinese Mini Mental Status Examination; GDS, Geriatric Depression Scale; IADL, Instrumental Activities of Daily Living.
Significant P-values (<.05) are indicated in bold.
Calculated with an independent sample t-test.
Single Task (ST) and Dual Task (DT) Gait Speed and DT Cost in Frailty
Table 2 displays the primary outcome measures of DT1 and DT2 gait speeds, along with ST gait speed and DT cost. Robust participants demonstrated significantly faster gait speeds than their PFF counterparts, evident in all 3 conditions (P < .001). The PFF participants also experienced greater DT cost compared to the robust participants. Of note, this difference in DT cost between both groups was only found to be significant for DT1, but not for DT2.
Gait Speed and DT Cost in Frailty, Presented in Mean (SD).
Abbreviations: ST, single task; DT1, arithmetic dual task; DT2, verbal fluency dual task.
Significant P-values (<.05) are indicated in bold.
Calculated with an independent sample t-test.
Cohen’s d used to estimate the standardized effect size between the robust and pre-frail/frail groups.
Receiver Operating Characteristic (ROC) Curve Analysis
ROC curve analysis was employed to compare the accuracy of ST and DT gait speeds in predicting frailty status (Figure 1). The ST gait speed achieved an AUC of 0.760 (95% CI = 0.719-0.801), while DT1 gait speed had an AUC of 0.722 (95% CI = 0.681-0.763) and DT2 gait speed showed an AUC of 0.681 (95% CI = 0.638-0.725). Higher AUC values indicate greater discriminative ability of gait speed in predicting frailty status within our cohort.
Receiver operating characteristic (ROC) curve for prediction of frailty based on ST and DT.
Correlations Between Gait Speed and Measures of Cognition, Mood, and Functional Status
The correlation matrix illustrates that ST and DT gait speeds were strongly positively correlated (DT1: r = .825, DT2: r = .706, P < .01; Table 3). DT gait speeds were also significantly correlated with age (DT1: r = −.359, DT2: r = −.285, P < .01), BMI (DT1: r = −.134, DT2: r = −.089, P < .01), measures of cognition (DT1: r = .243, DT2: r = .149, p < .01), depressed mood (DT1: r = −.237, DT2: r = −.179, P < .01), basic ADLs (DT1: r = .200, DT2: r = .143, P < .01), instrumental ADLs (DT1: r = .275, DT2: r = .234, P < .01), and the number of falls (DT1: r = −.071, P < .05; DT2: r = −.093, P < .01).
Pearson Correlation Coefficients of Gait Speed, Measures of Cognition, Mood, and Functional Status.
Abbreviations: BI, Barthel Index; BMI, Body Mass Index; CMMSE, Chinese-Mini Mental Status Examination; DT1, arithmetic dual task; DT2, verbal fluency dual task; Falls, number of falls in 1 year.; GDS, Geriatric Depression Scale; IADL, Instrumental Activities of Daily Living; ST, single task.
95% Confidence Interval for individual measures and ST gait speed.
95% Confidence Interval for individual measures and DT1 gait speed.
95% Confidence Interval for individual measures and DT2 gait speed.
P < .05. **P < .01.
Discussion
The primary objective of this study was to compare and evaluate DT gait speed changes between robust and PFF participants to improve frailty detection and its potential in functional rehabilitations. PFF participants showed significantly slower gait speed in both DT conditions compared to their robust counterparts, with higher DT costs for both cognitive tasks, indicating greater performance decline during multitasking. Notably, the arithmetic task had a preferentially greater impact on gait speed in the PFF group compared to the verbal fluency task, reflected by significant differences in DT1 cost but not in DT2.
Our findings underscore the significance of DT gait assessments as they reveal functional vulnerabilities in PFF older adults who may have decreased cognitive and motor reserves. In line with prior research linking frailty and cognitive impairment,3,6,7 the poorer DT performance observed in our PFF cohort supports the idea that frailty impacts the brain’s capacity for multitasking. Thus, DT assessments not only help in identifying physical frailty but also provide valuable insights into cognitive function, particularly in domains like executive control, which is often diminished in frailty. We thus advocate for the integration of DT gait assessments in comprehensive frailty screening programs as PFF older adults may have diminished cognitive reserves which are also functionally important in mobility.
While DT impairment is more pronounced in PFF older adults compared to their robust counterparts, ROC curve analysis shows that ST gait speed remains a better screening tool for predicting frailty status when used in isolation. This aligns with the recognition that frailty is a multifaceted construct that encompasses more than just physical and cognitive impairments. 1 Cognition likely contributes the least to the physical frailty phenotype when compared to other domains like strength, mobility, and energy. 29 While cognitive impairment and physical frailty often coexist, they remain distinct entities with similar underlying pathophysiological processes. 7
The literature further reports inconsistent findings when analyzing DT gait performance in the detection of frailty. When analyzing DT cost, some studies found significantly different DT cost between groups,13,15,30 while others did not detect this difference.31 -33 We also had different results with the DT1 and DT2 performances, which further emphasizes that the selection of cognitive tasks in gait testing is important, due on their variable impact on cognitive reserves.
Dual tasking is an integral part of community ambulation, which requires individuals to perform cognitive and motor tasks simultaneously in a variety of dynamic environments. These findings support the practical value of DT gait assessments in predicting real-world mobility challenges for PFF individuals. Although their mean ST gait speed of 1.1 (0.3) m/s was adequate, their slower DT gait speed (DT1: 0.9 (0.3) m/s; DT2: 0.8 (0.3) m/s) fell below the 1.0 m/s threshold required for safely crossing pedestrian crossings in Singapore. 34 In contrast, the robust group had gait speeds that met the demands of a pedestrian crossing in all 3 conditions (ST: 1.3 (0.2) m/s; DT1: 1.1 (0.3) m/s; DT2: 1.0 (0.3) m/s). This highlights the limitations of relying solely on ST assessments, which may provide a false sense of security regarding safe ambulation in complex environments.
In addition to ensuring safety on an individual level, a deeper understanding of DT gait speed in the older population may facilitate the design of neighborhood environments to meet their needs. An important component of a geriatric-friendly neighborhood is walkability, which encompasses the domains of traffic and personal safety. Improving walkability in a neighborhood promotes physical activity, leading to improved health outcomes, both physical and mental. 35 Similarly, this information can also be applied clinically when setting individual rehabilitation goals to ensure safety in the community.
Correlation analysis shows that slower DT gait speed was associated with poorer cognition, mood, functional status, and increased falls. The significant correlations between DT gait speed, cognition, and mood are in tandem with the findings discussed above. One’s functional status is determined by the ability to carry out basic and instrumental activities of daily living (ADLs), which often requires executive function. Poorer DT gait performance is associated with decreased functional status and ADL ability. 36 Similarly, impaired DT ability and decreased executive function appear to be predictors of falls.37,38 These relationships demonstrate the potential of DT gait assessments to be used as a surrogate marker of executive function.
Beyond screening, DT gait assessment may play a crucial role in frailty interventions. Programs incorporating DT exercises in older adults have been shown to improve one’s walking speed and balance to a greater extent compared to ST exercises only, 18 consequently reducing one’s risk of falling. DT exercise programs may even reverse one’s frailty status, in addition to improving executive function, cognition, and attention.39,40 Furthermore, DT training is especially clinically relevant as most ADLs require one to perform 2 or more tasks simultaneously. 17 Screening and early identification of DT impairments in the frail older individuals may allow for more targeted interventions to improve one’s frailty status and associated outcomes.
Strengths of our Study
There are multiple strengths of our study that address gaps present in prior research. A large sample size was recruited with minimal exclusion criteria, providing a more accurate representation of frail adults commonly encountered in the community. Unlike previous studies with stringent exclusion criteria, our inclusion of participants with multiple comorbid conditions and/or walking aids offers a realistic reflection of frailty in the community. This study was conducted as part of a comprehensive frailty assessment, which allowed for a holistic assessment of each participant. This enables us to evaluate their physical performance in the context of their overall health, which is often complex with impairments in multiple domains that encompass a large number of subjective and objective variables.
Our study was also conducted in a community setting, allowing for in situ assessment of the participants’ performance. This is in contrast to an artificial laboratory environment, which is often unfamiliar and does not accurately simulate real-life scenarios. This also highlights the accessibility of DT gait speed assessments, which can be easily carried out without specialized gait assessment equipment that is often costly and requires trained personnel to operate. DT gait speed is a simple, safe, and inexpensive assessment tool that can be implemented with minimal training.
Limitations
While this study contributes additional insights in this field, it is essential to acknowledge its limitations. Participant recruitment was done through sign-ups at senior activity centers or community centers, possibly resulting in selection bias. These individuals tended to be more physically active and socially engaged, and may falsely underrepresent the proportion of frail individuals. Despite this potential limitation, this recruitment strategy was chosen because it was deemed to be the most efficient and effective method of outreach. These centers provide a platform for engagement with the older residents, and serve as a place of familiarity and social networks. They have a higher likelihood of being successful in contacting and recruiting these participants, 41 and are also able to help disseminate publicity materials to a wider population.
A total of 18.5% and 1.6% of our participants were found to be pre-frail and frail respectively, which is lower than the prevalence of 31.2% to 45.7% and 5.7% to 6.2% previously found in our local Singaporean population.42,43 However, these studies only involved older populations with participants aged 65 years and older, possibly accounting for the increased rates of frailty. While frailty is indeed more commonly identified in the older population, pronounced physiological changes and sarcopenia may manifest as early as 50 years old. 44 Furthermore, given that prefrailty/frailty is more likely to be amenable to interventions at an early stage, it is critical to identify at-risk individuals when they are younger. 45
Future Research
Currently, DT gait assessments lack standardization and have no established testing protocol. The DT performed can be either physical or cognitive, and the DT influence can vary widely, depending on the task chosen. While there are comprehensive gait assessment guidelines available, 8 the heterogeneity of currently available studies on DT highlights the lack of uniformity in this domain. To improve standardization of DT assessments amongst practitioners, more research needs to be done to establish validated testing protocols that are clinically relevant and can be easily replicated in a community setting. Building on this, normative and cut-off values can then be determined, facilitating use in daily clinical practice. Similarly, there is a paucity of information surrounding specific DT training parameters. More research is required to determine the most effective modality of DT exercises.18,40 This would include parameters like type and difficulty of DT, exercise frequency and duration, as well as the recommended progression of exercises.
To further expand on the current findings and literature, future research should also explore the longitudinal relationship between DT gait speed and frailty. Long-term studies could investigate whether changes in DT gait performance precede the development of frailty, which is observed in cognitive impairment, 3 or if they are occurring in parallel. This would provide additional insight into the utility of DT as a predictor of frailty, and shape its function to be implemented in a public health screening program.
Conclusion
In conclusion, this study advances our understanding of frailty assessment in community-dwelling older adults, underscoring the potential of DT gait speed as a practical and more holistic tool for frailty screening. While PFF individuals indeed demonstrated slower DT gait speed and greater DT cost compared to robust individuals, ST gait speed still remained the better assessment tool for predicting frailty when used in isolation. Consequently, DT gait speed does not serve to replace ST gait speed as a screening tool. Instead, it complements it by incorporating the additional aspect of cognition, mirroring the complex nature of ADLs and community ambulation. The integration of DT gait assessments into routine clinical practice may open new avenues for early detection and targeted interventions, ultimately enhancing the well-being of older adults in the community. Moving forward, there is a need for more work to establish standardized guidelines in the realm of DT assessment and interventions, and this remains a promising aspect of frailty screening and rehabilitation.
Footnotes
Acknowledgements
We thank the study participants, staff of the Senior Activity Centres, and Resident Committees in the northeast region of Singapore for their logistical and manpower support.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research is supported by the National Medical Research Council (NMRC) through the SingHealth PULSES II Centre Grant (CG21APR1013) and the AM-ETHOS Duke-NUS Medical Student Fellowship (AM-ETHOS01/FY2023/13-A13) Grant. The grants funded the research staff, assessment equipment, and on-site conduct of the trial, and the researchers were independent from funders.
Ethical Approval
Ethics approval was obtained from SingHealth Institutional Review Board (SingHealth CIRB 2018/2115, Singapore).
Consent to Participate
All participants provided written informed consent during enrolment into the study.
Consent for Publication
Not applicable.
