A Systematic Review Exploring the Theories Underlying the Improvement of Balance and Reduction in Falls Following Dual-Task Training among Older Adults

Background: Balance impairment causes frequent falls in older adults, and preventing falls remains challenging. Dual-task (DT) training reduces falls by improving balance, but the precise theory is not fully understood. This review aims to explore the theories underlying the effectiveness of DT in improving balance and reducing falls in older adults. Methods: Eleven electronic databases were searched from database inception to June 2022. Two reviewers independently performed study screening and data extraction. The risk of bias (RoB) in the included studies was assessed using the Cochrane Collaboration RoB 2 tool. Results: The searches yielded 1478 citations, of which 30 studies met the inclusion criteria and were included in the review. Twenty-two of the 30 included studies utilized the motor-cognitive type of DT for training, while six used motor-motor and two utilized cognitive–cognitive DT. The included studies reported 20 different theories to explain the effectiveness of DT for improving balance and reducing falls in older adults. The predominant theory identified in the included studies was attention theory (n = 14). Overall, 26 studies reported improved balance and five studies found a reduction in fall incidence following DT training. Balance and falls improved significantly in 15 motor-cognitive DT intervention studies. Conclusion: Attention shifting between two tasks is reported to occur following DT training. Motor-cognitive DT training improves balance and reduces fall incidence in older adults by shifting attention based on the difficulty and priority of a task from the motor to the cognitive task.


Introduction
Globally, falls is the leading cause of unintentional death among older adults [1]. Moreover, falls is the 18th-leading cause of disability-adjusted life years in older adults [2,3]. Though the global incidence and prevalence of falls have dropped (by 3.7% and 6.5%, respectively) over the past two decades [4], managing falls remains a challenge among older adults. A common risk factor for falls is balance impairment [5]. One in five older Americans had balance impairment, and the condition was more common in women than in men [6]. One in four people with impaired balance has difficulty with daily activities [6]. Approximately 10% of falls result in fractures, which are also a significant source of morbidity and mortality in older adults [7,8]. Falls decrease quality of life and confidence, increase fear of falling, and limit functional ability and interpersonal interactions [9,10]. Even minor fall-related injuries are reported to cause pain, limited function, and high medical bills [7].
The causes of falls are multifactorial [11]. Poor balance is one of the crucial contributors and can result from white matter lesions in the brain that frequently lead to a decrease in multitasking activity [12][13][14][15]. White matter lesions have been linked to motor and cognitive dysfunction during multitasking, which degrades balance performance [15,16]. In multitasking, interference between two tasks reduces performance in one or both tasks [17], and

Risk of Bias
The Cochrane Collaboration Risk of Bias (RoB 2) tool was used to assess bias [40]. The RoB 2 tool analyzes randomization, intervention deviations, missing outcome data, outcome measurement bias, and result reporting bias. Each question was answered with "yes", "probably yes", "probably no", "no", or "no information" [40]. The bias risk of each domain was rated as "low", "some concerns", or "high." Similar to the individual domains, overall RoB 2 was also summarized as "low", "some concerns", or "high" risk of bias [40].

Statistical Analysis
We calculated the agreement between the two authors using the kappa value for the data screening process and quality appraisal. Values ≤ 0 indicated no agreement; 0.01-0.20 indicated no to little agreement; 0.21-0.40 indicated fair agreement; 0.41-0.60 indicated moderate agreement; 0.61-0.80 indicated substantial agreement; and 0.81-1.00 indicated nearly perfect agreement [41]. The RoB 2 tool was used to assess the bias and the methodological quality of specific results of RCTs. Since the focus of this review was to explore the theories, not the treatment effects, a quantitative analysis such as a meta-analysis or meta-regression was not considered necessary.

Search Results
The electronic searches yielded 1478 potentially relevant studies. Figure 1 summarizes the flow of studies through the review. After a stepwise screening process, 32 studies were found to be eligible for review. Two studies [42,43] were excluded after screening for full text. Two studies were excluded as they did not report any theories for explaining the treatment benefits of DT training. Therefore, this review included 30 studies. The agreement between the two review authors was near perfect (0.92) for full-text screening. Studies excluded at the full-text screening stage and the reasons for exclusion are reported in Appendix A Table A2.

Risk of Bias
The findings of the RoB assessment are illustrated in Figure 2. Overall, there was a low to moderate RoB across more than 38.5% of the studies. Twenty-six percent (n = 8) of the included studies were at low risk of bias, while 12.5% (n = 4) of the studies drew "some concerns" about outcome measurement, randomization, and deviation from intended intervention. "High" RoB was identified in 60% of studies (n = 18). The major methodological flaws were identified in measuring outcomes [23,[44][45][46][47], missing outcome data [29,48,49], or both [22,[50][51][52][53]. For methodological flaws for measuring outcome, firstly, flaws resulted because there was insufficient information available about whether outcome assessors were aware of the intervention that study participants had received. Secondly, the influence of the knowledge of the intervention on the assessment was addressed inadequately. For missing outcome data, the studies were reported "no, possibly no, or no information" if either the missing data was not reported or the statistical analysis for handling the missing data was not clearly reported.

Types of Dual-Task Training
The DT interventions delivered in the included studies were (1) motor-cognitive, (2) motor-motor, or (3) cognitive-cognitive DT, as reported in Table 1
Balance exercise with a verbal fluency task; Squire-stepping exercises in fixed-priority and variable-priority instruction

Cognitive-cognitive DT training
Exercise training with both cognitive tasks performed at the same time is intended to improve the executive function of the brain.
In a computer game, players see an animated truck and a road sign in the background before they fade away. Subsequently, two vehicles reappear to identify the correct vehicle. Meanwhile, a circle of cars appears around the edge, with one road sign. Participants need to find the location where the road sign first appeared on the edge.

Types of Exercises
The demographics, types of DT interventions, treatment dosage, effectiveness, outcome measurement, authors' conclusion, and theories of action proposed by the authors from the 30 included studies are reported in Table 2. Further detailed descriptions of the exercises, dosage of primary and secondary tasks, and control groups from the included studies are reported in Appendix A Table A3. Across the 30 included studies, different types of exercises were applied. Balance exercises appeared most frequently, as they were used in 12 studies [21,22,30,47,50,51,54,58,[61][62][63][64]. Balance exercises were performed using free-hand, low-tech systems with minimal technology support or technology-dependent, computerized balance systems. Balance exercise combined with resistance exercise was performed in four additional studies [48,49,57,60].

Theories Reported in the Included Studies
Among the 30 included studies, 20 different theories to explain balance improvement following DT interventions were identified. A single theory was documented in 19 studies [22,[28][29][30][44][45][46][47][48][49]51,53,56,59,60,[63][64][65]67]. More than one theory was reported in 12 studies [21,23,50,52,[54][55][56][57][58]61,62,66], and of these, four studies [23,52,58,61] reported four theories. The attention theory was proposed to explain the improvement in balance and reduction in falls following the DT intervention in 14 studies [22,23,28,[44][45][46]48,51,[58][59][60]62,64,66]. The predominant theory invoked was the competition theory of attention, which was used in over a third [22,28,[44][45][46]48,51,[58][59][60]64] of all studies primarily (explained first, either alone or along with other theories) and in three studies secondarily (not explained as a primary theory, but presented along with other theories) [23,62,66]. The competition theory of attention has been extensively proposed as an underlying theory to explain improving balance, mostly in the application of motor-cognitive and motor-motor types of DT training. Figure 3a illustrates the pathway for improvement in balance following DT intervention using the attention theory model. In this model, the attention of the brain responds to a winning stimulus from multiple competing stimuli. In the context of DT, for example, the winning stimulus of the calculation or cognitive task draws attention during the balancing exercise or motor task. This shifting attention works to improve motor-cognitive capabili-ties, which contributes significantly to the improvement in balance. DT theories commonly suggested to improve balance were executive function in six studies [23,54,55,57,65,66] and a multicomponent training approach in five studies [29,30,50,55,67]. The executive function theory focuses on enhancing cognition to improve balance by loading on inhibiting, updating working memory, and task-set shifting [68], as illustrated in Figure 3b, while multicomponent theories targeting procedural memory work via multi-phase cognitive functions to improve balance [69,70], as shown in Figure 3c. The working memory model was invoked as a primary [21,49] or secondary theory [23,52] in two studies each. The working memory model focuses on neural efficacy to improve balance via the management of multiple task processes [23,71], as described in Figure 3d. Likewise, the divided attention theory, which focuses on rapidly shifting or splitting attentional focus between two tasks, was referred to as a primary [23,52] or secondary theory [23,52,57,58] in two and four studies, respectively. responds to a winning stimulus from multiple competing stimuli. In the context of DT, for example, the winning stimulus of the calculation or cognitive task draws attention during the balancing exercise or motor task. This shifting attention works to improve motor-cognitive capabilities, which contributes significantly to the improvement in balance. DT theories commonly suggested to improve balance were executive function in six studies [23,54,55,57,65,66] and a multicomponent training approach in five studies [29,30,50,55,67]. The executive function theory focuses on enhancing cognition to improve balance by loading on inhibiting, updating working memory, and task-set shifting [68], as illustrated in Figure 3b, while multicomponent theories targeting procedural memory work via multi-phase cognitive functions to improve balance [69,70], as shown in Figure  3c. The working memory model was invoked as a primary [21,49] or secondary theory [23,52] in two studies each. The working memory model focuses on neural efficacy to improve balance via the management of multiple task processes [23,71], as described in Figure 3d. Likewise, the divided attention theory, which focuses on rapidly shifting or splitting attentional focus between two tasks, was referred to as a primary [23,52] or secondary theory [23,52,57,58] in two and four studies, respectively.

Discussion
To the best of our knowledge, this is the first systematic review to summarize the possible theories underlying the effectiveness of DT interventions for improving balance and reducing falls in older adults. Identifying the spectrum of theories proposed by authors underlying the improvement of balance following DT was the objective of this review. Interestingly we found there were at least 20 theories explaining the improvement following the intervention. We notice that most of the studies reported more than one theory for the identified benefits. This is an indication that DT results in the improvement of multiple domains such as attention, procedural memory executive function and motor function.
The 30 included studies, filtered from a total of 1478 studies identified via database searches, proposed 20 distinct DT theories. These proposed theories were reported to explain the improvement in balance (n = 26) and the reduction in the number of falls (n = 5) following DT intervention among healthy older adults. The most frequently proposed theory was the competition theory of attention, which describes the shifting of attention from one task to a secondary task with DT practice. The DT interventions improved both motor and cognitive function compared to various controls.
The competition theory of attention was used to describe the improvement in balance and reduction in falls following DT training; it was cited primarily in 11 studies (36.7%) and secondarily in three additional studies (10%). This theory is derived from the definition of attention first proposed by William James in 1890 [72] and involves the central nervous system, where neurons are subjected to a wide range of internal and external stimuli at any given instance. Each stimulus competes for the attention of the nervous system [73]. The nervous system filters the stimuli by considering the challenges and attending to the prioritized task [74]. Based on the difficulty and priority of a task, attention shifts from one task to another. For instance, when an individual is sitting on an inflatable

Discussion
To the best of our knowledge, this is the first systematic review to summarize the possible theories underlying the effectiveness of DT interventions for improving balance and reducing falls in older adults. Identifying the spectrum of theories proposed by authors underlying the improvement of balance following DT was the objective of this review. Interestingly we found there were at least 20 theories explaining the improvement following the intervention. We notice that most of the studies reported more than one theory for the identified benefits. This is an indication that DT results in the improvement of multiple domains such as attention, procedural memory executive function and motor function.
The 30 included studies, filtered from a total of 1478 studies identified via database searches, proposed 20 distinct DT theories. These proposed theories were reported to explain the improvement in balance (n = 26) and the reduction in the number of falls (n = 5) following DT intervention among healthy older adults. The most frequently proposed theory was the competition theory of attention, which describes the shifting of attention from one task to a secondary task with DT practice. The DT interventions improved both motor and cognitive function compared to various controls.
The competition theory of attention was used to describe the improvement in balance and reduction in falls following DT training; it was cited primarily in 11 studies (36.7%) and secondarily in three additional studies (10%). This theory is derived from the definition of attention first proposed by William James in 1890 [72] and involves the central nervous system, where neurons are subjected to a wide range of internal and external stimuli at any given instance. Each stimulus competes for the attention of the nervous system [73]. The nervous system filters the stimuli by considering the challenges and attending to the prioritized task [74]. Based on the difficulty and priority of a task, attention shifts from one task to another. For instance, when an individual is sitting on an inflatable exercise ball and catching a tennis ball simultaneously, attention needs to shift from catching the tennis ball to sitting on the inflatable exercise ball to ensure the safety of the user. The attention theory was used to explain balance improvement, as the participants were able to attend to two simultaneous tasks efficiently with repeated practice, and the number of falls was reduced due to their capacity to increase their multi-tasking ability. Talwer et al. [59] delivered a square stepping exercise that involved switching one's attention while passing a ball under variable priority instructions. The significant difference in balance measurement after training revealed that gradual practice decreases the reaction time of passing balls [59].
The competition theory of attention led to the development of the divided attention theory and the selective attention theory. These theories are mentioned in four studies [23,52,57,58]. The divided attention concept addresses the limitations of multitasking in information processing. The inability to process all information simultaneously demands division of attention, which splits or rapidly switches the attentional focus [75]. DT results in enhanced divided attention with the use of optimal attention resources in motor control [19]. This enables splitting attention between the primary and secondary tasks, which improves the outcomes of the functional task. The selective attention theory explains how multiple dynamic events and static sources of input are filtered and subsequently perceived, cognitively processed, and ultimately responded to [76,77]. Balance is considered a dynamic event [78], and the motor-cognitive DT interventions were intended to improve multiple dynamic events, including balance [78]. This intervention promotes parallel information processing and manipulating information [76]. Selective attention to task-relevant inputs and decision-making about balance performance leads to balance improvement [78,79]. One study [52] reported that motor-cognitive DT intervention was effective in improving balance and invoked the theories of both the divided and the selective attention theories as an explanation. Three more studies invoked the divided attention theory as an explanation for the improvement in balance [23,58] and reduction in falls [57] following DT intervention.
The multicomponent or multimodal training approach was proposed to explain the reduction in fall incidence following DT intervention in five studies [29,30,50,55,67]. This theory emphasizes that the DT intervention could activate procedural memory, which in turn enhances cognitive function [70]. Procedural memory is crucial for activities of daily life [80] and plays a role in multi-phase motor and cognitive functions. Simultaneous activity of the striatum and caudate nucleus improves procedural memory, targeting both motor and cognitive functions in one training program [69]. Thus, a multicomponent training program is appropriate for procedural memory improvement [70]. Based on this theory, the DT interventions are hypothesized to improve cognitive function, attention, cognitive control, memory, reasoning, and executive functions simultaneously [67,70]. Multiphasebased DT intervention improves specific functions by targeting balance through repeated exercises [81]. Repeated processes enhance cognitive function, improve balance [29], and reduce the number of falls [30] in older adults. One study [67] examined both outcomes and found reduced falls, but not improved balance for tandem scale. This may be because this study used the infrequently tandem balance measurement scales, which are not appropriate tests for aging participants (mean age of 81.9 years).
The executive function theory was reported in six studies [23,54,55,57,65,66]. It covers a spectrum of three key cognitive processes for improving balance by enhancing cognition: increasing the load on task inhibition, updating working memory, and task-set shifting [68,82]. Improvement is achieved through motor-cognitive DT training, DT exercise increases the load on inhibiting primary tasks or motor tasks, updates it on the working memory, and eases the motor-to-cognitive task switching [54,68]. Together, they improved executive function, which improved balance and reduced falls. Four studies [23,55,65,66] used this theory to explain an improvement in balance, while one study used it to explain a reduction in falls [57] following DT intervention in older adults.
Two additional theories based on the executive function theory have been proposed to explain how the motor-cognitive type of DT training helps to improve balance: the working memory model and the cognitive flexibility theory.
Cognitive flexibility theory explains the ability to spontaneously rearrange information in an adaptive response to substantially altering situational demands [83]. Task-switching activities, including DT, require cognitive flexibility [84]. This flexibility is achieved in DT exercises, which activate the prefrontal, anterior cingulate, and posterior parietal cortices and basal ganglia, enhancing cognitive flexibility [84]. This theory was proposed in one study [52] that demonstrated substantial improvement in balance after DT training.
Working memory is necessary for complex task processes, including reasoning, comprehension, and learning [71]. The working memory model was proposed to explain the results in four different studies [21,23,49,52]. Balance improvement occurred due to the neuronal efficacy resulting from the motor-cognitive DT training, which is responsible for comprehensive learning via amplifying cognition [23].
Six theories in three studies explained how DT performance improves balance and reduces falls. In one study [58], the following three theories were described: (1) a taskoriented approach emphasized improving balance or movement strategies within a given environment using motor-cognitive exercise [85,86]; (2) a task-automation model led to complete task automation [87]; and (3) a task integration model introduced effective integration of two tasks, minimizing the sharing capacity to improve DT performance [87]. These three models propose that balance is enhanced by automating one of the two DT exercises. Silsupadol et al. combined one of the previously automated tasks into fixed or variable priority instruction, with balance as the primary task and cognitive task training as the secondary task [58]. This increased the balance function in older adults with repeated practice.
In addition to the theories described above [58], three studies proposed the integrated motor and cognitive theory, the theory of reduced resources, the capacity-sharing theory, and the limited resource theory to explain the balance improvement following DT intervention. One study [56] evaluating the advantages of motor-cognitive DT training invoked an integrated motor and cognitive theory to explain the theory of fall reduction. According to this theory, motor, cognitive, and physical development are all influenced by the biological predispositions that are most necessary for safe movement [88]. Similarly, motor-cognitive DT aims to enhance balance and reduce falls in both domains. Mirelman et al. postulated that combining a treadmill exercise with a cognitive exercise would mitigate fall risk significantly during the six months after training [56]. Integrated DT allows tasks to become habitual and minimizes competition to improve balance.
The reduced resource theory explains that the repetitions in DT training increase cognitive capacity, which in turn improves balance [62]. The task coordination theory describes the coordination and management of DT performance to optimize stability during the performance of concurrent tasks [89]. The capacity-sharing theory states that effective DT integration promotes DT performance by sharing brain resources between tasks [90,91]. In other words, sharing capacity improves through task integration, as brain resources are shared between two tasks, resulting in better DT performance [90]. According to the limited resource theory, the tasks in DT training compete for limited neural resources [91]. After performing the DT exercise, DT performance improves because each task becomes automatic and competition reduces [91]. Thus, complex motor-cognitive training makes tasks automatic and less competitive, thereby improving balance control in older adults [61].

Implications
The findings of this review provides a better understanding of the underlying theories for improvement following DT training. Our Table 2 reports a summary of the type of exercise, dosage of intervention, outcomes on balance and falls reduction and the proposed theory underlying the improvement. These findings benefit researchers by helping them to plan future intervention studies that could bridge the literature gap such as including cognitive-cognitive DT exercises and for the clinicians, these findings will assist in making a clear choice on the type and dosage of DT intervention for achieving specific health benefit among the older adults.
This review has several limitations. (1) Most of the studies did not mention the repetition and allocated time for DT separately (i.e., the dosage for the primary and secondary tasks). The load and complexity of the exercise, as well as the clinician's skills, are key to achieving a successful outcome from intervention; however, they were addressed inadequately in the included studies. Future RCTs should address these inadequacies appropriately. (2) Our included studies were restricted to English and Chinese language publications; therefore, it is possible that potentially relevant studies [92,93] were not considered. (3) The findings of this study must be interpreted with caution due to the heterogeneity in participant characteristics, study methods, type of DT intervention, outcome measures used, dosage of intervention, study setting and mode of intervention delivery. (4) More than half (60%) of the studies were classified as having poor methodological quality, and as quality impacts the study results, this must be considered when interpreting the findings. (5) Not all RCTs necessarily support the theories. Among the potential studies for inclusion, two studies [42,43] that did not record an improvement in balance or fall reduction did not report a theory to support and therefore they were excluded in the fulltext screening stage. In addition, one study [60], though did not record an improvement mentioned a theory for supporting the benefits of the intervention. (6) We restricted studies among healthy older adults without pathological conditions and therefore generalizing these findings to all older adults is limited. Future reviews are warranted to study special groups of older adults with pathological conditions and (7) we did not attempt to explore the reasons for improvement among the studies that did not report the theory (n = 2). This strategy is in line with our review protocol. This review also has the following merits: (1) extensive searches for relevant studies were performed across 11 databases using a systematic methodology and (2) since this review exclusively included RCTs, our conclusions are based on high-quality evidence.

Conclusions
This review identified 20 possible theories to explain the improvement in balance and reduction in falls following DT training among older adults. The dominant theory invoked to explain the effectiveness of DT training was the attention theory, which is commonly proposed in motor-cognitive DT studies. In motor-cognitive DT training, attention is shifted based on the complexity and priority of a task from the motor to the cognitive task to improve balance and reduce falls incidence among older adults.   Table A2. Reasons for excluded study after full-text reading.

2, 3
Data collection process 9 Specify the methods used to collect data from reports, including how many reviewers collected data from each report, whether they worked independently, any processes for obtaining or confirming data from study investigators, and if applicable, details of automation tools used in the process.

Data items 10a
List and define all outcomes for which data were sought. Specify whether all results that were compatible with each outcome domain in each study were sought (e.g., for all measures, time points, analyses), and if not, the methods used to decide which results to collect.

10b
List and define all other variables for which data were sought (e.g., participant and intervention characteristics, funding sources). Describe any assumptions made about any missing or unclear information.

N/A
Study risk of bias assessment 11 Specify the methods used to assess risk of bias in the included studies, including details of the tool(s) used, how many reviewers assessed each study and whether they worked independently, and if applicable, details of automation tools used in the process.

5
Effect measures 12 Specify for each outcome the effect measure(s) (e.g., risk ratio, mean difference) used in the synthesis or presentation of results. N/A Describe the processes used to decide which studies were eligible for each synthesis (e.g., tabulating the study intervention characteristics and comparing against the planned groups for each synthesis (item #5)).

13b
Describe any methods required to prepare the data for presentation or synthesis, such as handling of missing summary statistics, or data conversions. Describe any methods used to assess risk of bias due to missing results in a synthesis (arising from reporting biases). N/A Certainty assessment 15 Describe any methods used to assess certainty (or confidence) in the body of evidence for an outcome. N/A

Study selection 16a
Describe the results of the search and selection process, from the number of records identified in the search to the number of studies included in the review, ideally using a flow diagram.

16b
Cite studies that might appear to meet the inclusion criteria, but which were excluded, and explain why they were excluded. 21 Study characteristics 17 Cite each included study and present its characteristics. 6 Risk of bias in studies 18 Present assessments of risk of bias for each included study. 6

Results of individual studies 19
For all outcomes, present, for each study: (a) summary statistics for each group (where appropriate) and (b) an effect estimate and its precision (e.g., confidence/credible interval), ideally using structured tables or plots.

Other Information
Registration and protocol 24a Provide registration information for the review, including register name and registration number, or state that the review was not registered.
2 24b Indicate where the review protocol can be accessed, or state that a protocol was not prepared. 2 24c Describe and explain any amendments to information provided at registration or in the protocol. N/A Support 25 Describe sources of financial or non-financial support for the review, and the role of the funders or sponsors in the review. 19 Competing interests 26 Declare any competing interests of review authors. 19 Availability of data, code and other materials 27 Report which of the following are publicly available and where they can be found: template data collection forms; data extracted from included studies; data used for all analyses; analytic code; any other materials used in the review.