Interactive Cognitive-Motor Training in Older Adults – the Extra Boost for Cognitive Performance and Brain Function?

Research Collection

Doctoral Thesis

Interactive Cognitive-Motor Training in older Adults – The Extra Boost for Cognitive Performance and Brain Function?

Author(s): Eggenberger, Patrick

Publication Date: 2017

Permanent Link: https://doi.org/10.3929/ethz-b-000246161

Rights / License: Creative Commons Attribution 4.0 International

This page was generated automatically upon download from the ETH Zurich Research Collection. For more information please consult the Terms of use.

ETH Library DISS. ETH NO. 24680

INTERACTIVE COGNITIVE–MOTOR TRAINING IN OLDER ADULTS – THE EXTRA BOOST FOR COGNITIVE PERFORMANCE AND BRAIN FUNCTION?

A thesis submitted to attain the degree of

DOCTOR OF SCIENCES of ETH ZURICH

(Dr. sc. ETH Zurich)

presented by PATRICK EGGENBERGER Dipl. Natw. ETH

born on 04.06.1974 citizen of Grabs SG, Switzerland

accepted on the recommendation of

Prof. Dr. Eling D. de Bruin, ETH Zurich Prof. Dr. Nicole Wenderoth, ETH Zurich Prof. Dr. Jorunn L. Helbostad, NTNU Trondheim, Norway

2017

2 TABLE OF CONTENTS

1 Summary, Zusammenfassung 5

2 Thesis introduction 11

3 Gait speed in older adults (study 1) 27

4 Cognitive performance adaptations (study 2a) 51

5 Gait & physical performance adaptations (study 2b) 79

6 Brain functional adaptations (study 3) 119

7 Thesis conclusions 153

8 References 161

9 Acknowledgments 185

4 1

Summary, Zusammenfassung

5 1.1 SUMMARY

Emerging evidence indicates that age-related decline in higher order cognitive processing, e.g. attention and executive functioning, is associated with impaired gait and is important in relation to falls in older adults (Yogev-Selig- mann et al., 2008; de Bruin and Schmidt, 2010; Mirelman et al., 2012). There are strong indications that the most important risk factors for cognitive decline, dementia, impaired gait, and falls in older adults are interconnected and are modifiable through physical activity and training as well as partly through cognitive training (Stenhagen et al., 2013; Baumgart et al., 2015; Hortobagyi et al., 2015; Van Abbema et al., 2015; Liu et al., 2016). However, since to date, to the best of our knowledge, no interventions have studied synergistic or cumulative effects from multicomponent physical exercise programs with additional cognitive training to improve cognitive performance, brain function, gait, and prevent falls. Moreover, the neurophysiological mechanisms mediating the effects of physical training on cognition and brain function in older adults remain unclear.

Therefore, the aim of this thesis is to illuminate the effects of interactive simultaneous cognitive–motor training modalities on cognitive and brain function, cognitive–motor dual-task walking, and fall prevention in older adults. Study 1 (chapter 3) addresses the question of how good the cognitive and physical fitness status of Swiss older adults, as assessed with cognitive–motor dual-task gait speed measurements, is in relation to the requirements that are necessary to stay independent as a pedestrian in an urban environment. This cross-sectional study with 120 participants establishes the relevance of investigating effective training interventions for older adults, which are in the focus of the subsequent interventional studies 2a/b and 3. Thus, as a second step, a 6-month longitudinal training intervention study with a 1-year follow-up is performed, including 89 participants, to investigate both broad cognitive (study 2a, chapter 4) and physical adaptations, including dual-task gait measures and fall frequency (study 2b, chapter 5). Finally, in a third step, a shorter 8 weeks lasting training intervention study is conducted, including 42 participants, to address the question if the training-induced cognitive behavioral adaptations, that were observed in study 2a, would be reflected in brain functional adaptations in older adults (study 3, chapter 6).

6 The main findings from study 1 include that about every third (35.6%) older person at the age of 70–79 years and almost three-quarters (73.8%) of persons ≥80 years cannot walk faster than 1.2 m/s, which is required to cross streets safely within the green–yellow phase of pedestrian lights, under cognitively challenging conditions. Study 2a demonstrates, first, that the two interactive simultaneous cognitive–motor programs are partially advantageous to boost performance in two measures of executive function (switching attention and working memory) compared to an exclusively physical training program; and second, that cognitive performance, including executive functions, long-term visual memory (episodic memory), and processing speed, is maintained until 1-year follow-up after all three interventions. Study 2b shows, first, that the two interactive simultaneous cognitive–motor programs result in a significant advantage in dual-task costs of walking compared to the exclusively physical program but not in any other gait variables; second, that the two simultaneous cognitive–motor interventions lead to different training-specific adaptations in the rhythm and variability domains of gait; and third, that each of the three training programs very effectively reduced fall frequency for ~77%. Finally, Study 3 reports, first, that both the video game dancing and the balance interventions reduce left and right hemispheric prefrontal cortex (PFC) oxygenation during the acceleration of walking, while video game dancing showed a larger reduction at the end of the walking phase compared to balance training in the left PFC; and second, that the exercise training-induced modulations in PFC oxygenation are associated with improved executive functions.

The main conclusions from this thesis imply that the fitness status of many older adults is not appropriate to safely encounter the requirements for pedestrians in urban areas, which reinforces the need for regular cognitive and physical training in the older population (study 1). Interactive simultaneous cognitive–motor training should be integrated in training programs aiming to improve cognition, gait performance, physical functioning, and reduce fall frequency in older persons. Such programs may potentially counteract the large prevalence of cognitive and gait impairments, as well as reduce fall frequency, inherently leading to more independence and a better quality of life (studies 2a/b). Finally, training-induced brain functional adaptations in the PFC seem to reduce the need of prefrontal resources of executive function and attention involved in challenging treadmill walking. This effect may liberate

7 cognitive resources to focus on other processes while walking in attention demanding real-life situations such as crossing streets or walking while talking and could potentially reduce the risk of falling (study 3).

1.2 ZUSAMMENFASSUNG

Neuere Forschungsergebnisse zeigen, dass die altersbedingte Abnahme von höheren kognitiven Prozessen, wie Aufmerksamkeit und exekutive Funktio- nen, mit Beeinträchtigungen des Gehens zusammenhängt und eine wichtige Rolle spielt bei Stürzen von älteren Erwachsenen (Yogev-Seligmann et al., 2008; de Bruin and Schmidt, 2010; Mirelman et al., 2012). Es gibt klare Hin- weise dafür, dass die wichtigsten Risikofaktoren für abnehmende kognitive Fähigkeiten, Demenz, Gangstörungen und Stürze bei älteren Erwachsenen in Verbindung zu einander stehen und zudem durch körperliche Aktivität und körperliches Training, sowie teilweise auch durch kognitives Training, modifizierbar sind (Stenhagen et al., 2013; Baumgart et al., 2015; Hortobagyi et al., 2015; Van Abbema et al., 2015; Liu et al., 2016). Dennoch haben unse- res Wissens bis heute keine Studien die synergistischen und kumulativen Ef- fekte von körperlichen Multikomponenten-Trainings-programmen, ergänzt mit kognitivem Training, hinsichtlich Verbesserungen der kognitiven Leis- tungsfähigkeit, der Hirnfunktion, des Ganges und der Sturzprävention untersucht. Zudem sind die neurophysiologischen Mechanismen ungeklärt welche die Effekte von körperlichem Training auf Kognition und Hirnfunktion bei älteren Erwachsenen verursachen.

Das Ziel dieser Doktorarbeit ist es deshalb, die Effekte von interaktiven, simultanen kognitiv–motorischen Trainingsformen auf die kognitiven Fähig- keiten, die Hirnfunktion, das Gehen mit kognitiv–motorischer Doppelaufgabe und die Sturzprävention bei älteren Erwachsenen zu beleuchten. In Studie 1 (Kapitel 3) wird der Frage nachgegangen wie gut die kognitive und körper- liche Fitness von älteren Schweizerinnen und Schweizern ist, bezüglich den Anforderungen welche notwendig sind um als Fussgänger in einem städti- schen Umfeld unabhängig zu bleiben. Als Parameter der kognitiven und kör- perlichen Fitness wird dazu die Gehgeschwindigkeit bei einer kognitiv–motorischen Doppelaufgabe erhoben. Diese Querschnittsstudie mit 120

8 Teilnehmern liefert die Begründung dafür weshalb es von grosser Relevanz ist, effektive Trainingsinterventionen für ältere Erwachsene zu erforschen, welche bei den nachfolgenden Interventionsstudien 2a/b und 3 im Fokus stehen. Als zweiter Schritt wird somit eine 6-monatige Trainings-Interventions- studie durchgeführt, welche eine Nachfolgeuntersuchung nach einem Jahr beinhaltet. Anhand von 89 Teilnehmern werden dabei umfassende kognitive (Studie 2a, Kapitel 4) und körperliche Anpassungen untersucht, inklusive Messungen des Gehens mit kognitiv–motorischer Doppelaufgabe und Sturz- häufigkeit (Studie 2b, Kapitel 5). Schliesslich wird in einem dritten Schritt eine kürzere 8-wöchige Trainings-Interventionsstudie mit 42 Teilnehmern durchgeführt. Damit soll die Frage beantwortet werden, ob sich durch das Training verursachte kognitive, verhaltensbasierte Anpassungen, welche in Studie 2a beobachtet wurden, in funktionellen Anpassungen des Gehirns von älteren Erwachsenen wiederspiegeln (Studie 3, Kapitel 6).

Die wichtigsten Ergebnisse aus Studie 1 beinhalten, dass etwa jede dritte (35.6%) ältere Person im Alter von 70–79 Jahren und fast drei Viertel (73.8%) der Personen ≥80 Jahre unter kognitiv anspruchsvollen Bedingungen nicht schneller als 1.2 m/s gehen können, was erforderlich ist um eine Strasse wäh- rend der grün–gelb Phase der Fussgängerampel sicher zu überqueren. Stu- die 2a zeigt erstens, dass, die zwei interaktiven, simultanen kognitiv–motorischen Programme teilweise vorteilhaft sind gegenüber ausschliesslich kör- perlichem Training um die Leistung in zwei Messgrössen der exekutiven Funktionen zu verbessern (Aufmerksamkeitswechsel und Arbeitsgedächt- nis); und zweitens, dass, die kognitive Leistungsfähigkeit, inklusive exekutive Funktionen, visuellem Langzeitgedächtnis (episodisches Gedächtnis) und Verarbeitungsgeschwindigkeit, bis zur Nachfolgeuntersuchung nach einem Jahr in allen drei Interventionsgruppen erhalten bleiben. Studie 2b legt erstens dar, dass die zwei interaktiven, simultanen kognitiv–motorischen Programme im Vergleich zum ausschliesslich körperlichen Training einen signifikanten Vorteil im Zusammenhang bei den Leistungseinbussen mit der Doppelaufgabe beim Gehen aufweisen, jedoch nicht in allen anderen Gang- variablen; zweitens, dass die zwei simultanen kognitiv–motorischen Inter- ventionen zu unterschiedlichen trainings-spezifischen Anpassungen in den Bereichen des Rhythmus und der Variabilität des Gehens führen; und drit- tens, dass alle drei Trainingsprogramme die Sturzhäufigkeit sehr effektiv um ~77% senken. Schliesslich berichtet Studie 3 erstens, dass das Videospiel-

9 Tanztraining und auch das Gleichgewichtstraining die Sauerstoffanreiche- rung im linken und rechten präfrontalen Cortex (PFC) während der Be- schleunigungsphase des Gehens reduziert, während dem das Videospiel- Tanztraining im Vergleich zum Gleichgewichtstraining eine grössere Reduk- tion der Sauerstoffanreicherung am Ende der Geh-Phase zeigt; und zweitens, dass die trainingsbedingten Anpassungen der Sauerstoffanreicherung im PFC mit verbesserten exekutiven Funktionen zusammenhängen.

Die wichtigsten Schlussfolgerungen dieser Doktorarbeit implizieren, dass der Fitnesszustand von vielen älteren Erwachsenen nicht genügend ist um den Anforderungen an Fussgänger im städtischen Umfeld sicher zu begegnen. Dies bekräftigt die Notwendigkeit eines regelmässigen kognitiven und kör- perlichen Trainings in der älteren Bevölkerung (Studie 1). Interaktives, si- multanes kognitiv–motorisches Training sollte in Trainingsprogramme mit dem Ziel der Verbesserung von Kognition, Gehfähigkeit, körperlicher Funk- tionstüchtigkeit und zur Reduktion der Sturzhäufigkeit bei älteren Personen integriert werden. Solche Programme könnten der grossen Prävalenz von kognitiven und gangbezogenen Beeinträchtigungen entgegenwirken, sowie die Sturzhäufigkeit vermindern und somit zu mehr Unabhängigkeit und einer besseren Lebensqualität führen (Studien 2a/b). Schliesslich scheinen trainingsbedingte Anpassungen der Hirnfunktion im PFC den Bedarf an präfrontalen Ressourcen für die exekutiven Funktionen und die Aufmerk- samkeit zu verringern, welche während anspruchsvollem Gehen auf dem Laufband notwendig sind. Dieser Effekt könnte somit kognitive Ressourcen freigeben welche es ermöglichen, sich während dem Gehen auf andere Pro- zesse zu fokussieren. Beispielsweise beim Überqueren einer Strasse oder wenn während dem Gehen ein Gespräch geführt wird, also in Situationen wo erhöhte Aufmerksamkeit erforderlich ist. Zudem könnte dieser Effekt zu einer Reduktion des Sturzrisikos beitragen (Studie 3).

10 2

Thesis introduction

11 2.1 THE INTERRELATION OF COGNITION, GAIT, AND FALLS IN OLDER ADULTS

Three major health problems in the older population comprise cognitive decline, reduced walking ability, and increased risk of falling. First, advancing age is associated with a reduction of cognitive functioning, which is prevalent in almost every second elderly person (Scafato et al., 2010); thereby, cognitive decline threatens independence and quality of life of older adults (Williams and Kemper, 2010) and is recognized as a precursor of dementia (Baumgart et al., 2015). Second, reduced walking speed is predictive of community functioning (Studenski et al., 2003) and is regarded as the “sixth vital sign” (Fritz and Lusardi, 2009) due to its strong association with negative health outcomes (Liu et al., 2016). Third, about one-third of people aged 65 years or older fall at least once a year, and incidence rises with aging (Tinetti et al., 1988; Hausdorff et al., 2001; Gill et al., 2005). Notably, thinking, walking, and falling are highly associated (Alexander and Hausdorff, 2008) as depicted in Figure 1: cognitive decline is related to impaired gait (Liu et al., 2016) and a higher risk of falls (Muir et al., 2012), while impaired gait is also associated with an elevated fall risk (Abellan van Kan et al., 2009). However, the underlying neurophysiological mechanisms of these interrelations are still under debate and effective training approaches that counteract the risk factors related to cognitive decline, impaired gait, and falls are sparse.

FIGURE 1 | Associations between three major Thinking health problems in older adults. cognitive impairment cognitive impairment is is related to impaired related to higher risk gait (Liu 2016) of falls (Muir et al. 2012)

Walking Falling reduced usual walking speed is related to increased fall risk (Abellan van Kan 2009)

12 The strength of evidence for different factors that increase or decrease the risk for cognitive decline and dementia was reviewed by Baumgart et al. (2015) (Figures 2 and 3). Thereby, strong evidence was found for physical activity to be able to reduce the risk for cognitive decline and moderate evidence to reduce dementia. Furthermore, moderate evidence existed to support cognitive training as a measure to decrease the risk for cognitive decline and lower evidence to decrease dementia. Physical activity is generally recognized to promote healthy aging and even reduce morbidity and mortality rates (Byberg et al., 2009; Sun et al., 2010). Recent research demonstrated that physical activity may be particularly relevant for healthy brain aging and may protect from cognitive decline and dementia (Erickson et al., 2012; Lautenschlager et al., 2012; Gomez-Pinilla and Hillman, 2013; Gregory et al., 2013; Hotting and Roder, 2013).

FIGURE 2 | Strength of Increases risk evidence on risk factors Traumatic brain injury for cognitive decline Mid-life obesity History of Mid-life hypertension depression (adapted from Baum- Current smoking Sleep Hyper- Diabetes disturbances lipidemia

gart et al., 2015). strong lower unclear Cognitive decline of of evidence Physical activity Cognitive Moderate Social lower strong

training alcohol unclear engage-

Level Level Years of formal education moderate Mediterra- consump- ment nean diet tion

Reduces risk

FIGURE 3 | Strength of Increases risk evidence on risk factors History of Mid-life obesity depression for dementia (adapted Mid-life Sleep disturbances from Baumgart et al., hypertension Current smoking Hyper- Traumatic brain injury lipidemia 2015). strong moderate Diabetes unclear Dementia of of evidence Years of formal Physical Cognitive Moderate lower strong

activity training unclear alcohol

Level Level education moderate Mediterra- consumption nean diet Social engagement

Reduces risk

13 Impaired gait and particularly reduced usual walking speed is strongly associated with increased risk for disability, cognitive impairment, falls, and all- cause mortality (Abellan van Kan et al., 2009; Liu et al., 2016). A recent meta- analysis demonstrated that the risk of all-cause mortality is elevated by 89% in the older adults exhibiting the lowest preferred walking speeds (Liu et al., 2016) and is mostly associated with increased cardiovascular mortality risk (Dumurgier et al., 2009; Chen et al., 2012). Strength training (Van Abbema et al., 2015) and other exercise modalities (Hortobagyi et al., 2015) were shown to be effective measures to improve low walking speed.

The risk factors for falls in older adults, comprise impaired balance and gait, multimedication, a history of previous falls, advancing age, female sex, visual impairments, cognitive decline, and environmental factors (Ambrose et al., 2013). However, the three main components that predict falls are all related to physical fitness and include reduced mobility, heart dysfunction, and functional impairment (Stenhagen et al., 2013). Hence, many programs try to improve physical fitness and gait with the aim of preventing falls in older adults (Panel on Prevention of Falls in Older Persons and British Geriatrics, 2011).

Altogether, there are strong indications that the most important risk factors for cognitive decline, dementia, impaired gait, and falls in older adults are interconnected and are modifiable through physical activity and training as well as partly through cognitive training. Emerging evidence indicates that aging-related decline in higher order cognitive processing, e.g. in attention and executive functioning, are associated with impaired gait and are important in relation to falls in older adults (Yogev-Seligmann et al., 2008; de Bruin and Schmidt, 2010; Mirelman et al., 2012). Regardless of these findings, cognitive aspects are often neglected in fall prevention programs (de Bruin et al., 2011).

2.2 BENEFITS OF COGNITIVE AND PHYSICAL TRAINING IN OLDER ADULTS

Various physical exercise modalities have the potential to improve cognitive performance in older adults. Two meta-analytic studies reported that aerobic

14 exercise is effective in increasing cognitive performance in general, and executive function in particular (Colcombe and Kramer, 2003; Smith et al., 2010). More recent studies also found that strength and coordination training may positively affect cognitive abilities (Voelcker-Rehage et al., 2011; Chang et al., 2012; Liu-Ambrose et al., 2012). Physical exercise interventions that included aerobic training elicited both broad transfer and relatively large effects of cognitive performance measures (Lustig et al., 2009). In contrast, cognitive training studies have often shown highly task specific and small effects (Noack et al., 2009; Papp et al., 2009; Zelinski, 2009; Hindin and Zelinski, 2012; Oei and Patterson, 2014). These findings led to the assumption that the combination of cognitive and physical training might improve cognitive performance in old age more effectively than the training of an isolated ability (Lustig et al., 2009; Schaefer and Schumacher, 2011; Thom and Clare, 2011; Kraft, 2012; Gregory et al., 2013; Hotting and Roder, 2013; Law et al., 2014). Therefore, more and more studies pursue exactly this goal by administering a combined cognitive–motor training approach that includes cognitive and gross motor physical exercise components.

Some studies that implemented combined cognitive–motor training applied the training of the two components in a sequential manner (Oswald et al., 1996; Fabre et al., 2002; O'Dwyer, 2009; Legault et al., 2011; van het Reve and de Bruin, 2014; Fraser et al., 2017), whereas others performed the cognitive and physical training units simultaneously (Pichierri et al., 2012a; Pich- ierri et al., 2012b; Forte et al., 2013; Theill et al., 2013; Schättin et al., 2016). An advantage of simultaneous training designs might be that they include dual tasking and switching attention between the cognitive and physical activity. Thereby, the so called exergames (a neologism combining the words “exercise” and “video games”) represent a novel approach that simultaneously and interactively combine cognitive and physical training and have recently gained popularity. For instance, video game dancing, as a modality of exer- gaming, complemented with conventional strength and balance training, was shown to improve especially dual-task gait parameters which are related to brain function (Pichierri et al., 2012b), as well as higher cognitive processing as measured by standard neuropsychological tests (Fraser et al., 2014). Other exergames, e.g. interactive cognitive–motor step training games, have also led to improvements in cognitive functions in older adults (Schoene et al., 2015) and a systematic review by Shams et al. (2015) further highlighted the

15 positive effects that video games may have on cognition and brain structure in younger and older participants.

The interpretation of the existing literature on combined cognitive–motor training is, however, often limited due to small sample sizes (Fabre et al., 2002; O'Dwyer, 2009; Pichierri et al., 2012a; Pichierri et al., 2012b), inconsistent training exposures between intervention groups (Oswald et al., 1996; Fabre et al., 2002; Legault et al., 2011; Pichierri et al., 2012a; Pichierri et al., 2012b; van het Reve and de Bruin, 2014), or the lack of reference groups with only physical training (Pichierri et al., 2012a; Theill et al., 2013). Moreover, transfer to different cognitive domains was not assessed in some studies (Pichierri et al., 2012a; Pichierri et al., 2012b; Forte et al., 2013) and most interventions lasted no longer than 4 months (Fabre et al., 2002; O'Dwyer, 2009; Legault et al., 2011; Pichierri et al., 2012a; Pichierri et al., 2012b; Forte et al., 2013; Theill et al., 2013; Schättin et al., 2016; Fraser et al., 2017). This duration might be too short since physical training interventions of 6 months or longer have shown most consistent effects on cognition (Colcombe and Kra- mer, 2003; Erickson et al., 2012). Nevertheless, the promising findings from the previous research outlined above are well worth further investigations and since to date, to the best of our knowledge, no interventions have studied synergistic or cumulative effects from multicomponent physical exercise programs with additional cognitive training to improve cognitive performance, gait, and prevent falls.

Notably, not only behavioral cognitive performance improvements but also functional and structural brain plasticity has been observed after various exercise training modalities, including cardiovascular, strength, coordination, and balance training (Voss et al., 2010; Erickson et al., 2011; Voelcker-Rehage et al., 2011; Liu-Ambrose et al., 2012; Voss et al., 2013; ten Brinke et al., 2015). These studies illuminated the potential for plastic brain adaptations that occur after physical exercise training. However, up to date the training- induced functional brain adaptations were only observed during the execution of simple cognitive tasks, assessed with functional magnetic resonance imaging (fMRI). Therefore, it remains unclear if such adaptations are also measurable in real-life situations, such as challenging walking, and would ultimately be beneficial for older adults in their daily life activities.

16 2.3 NEUROPHYSIOLOGICAL MECHANISMS LINKING EXERCISE AND COGNITION

Physical activity or exercise induce alterations at the cellular and molecular level, which are likely to initiate structural and functional adaptations in the brain, and/or behavioral/socio-emotional changes that eventually influence cognitive functioning as depicted in Figure 4 below (Stillman et al., 2016).

Behavioral/ Physical Cellular and Structural and socio- Cognitive molecular functional activity/ emotional changes brain changes functioning exercise changes

Neurotrophic factors Brain morphology Sleep (BDNF, IGF-1, VEGF) Brain connectivity Mood Neurotransmitters and function Stress Angio-/Neurogenesis Perfusion Pain Inflammation Neuroelectric HPA axis activity potentials

FIGURE 4 | Various possible mediating pathways and bidirectional effects linking physical activity/exercise and cognition (adapted from Stillman et al., 2016).

Abbreviations: BDNF, brain-derived neurotrophic factor; HPA, hypothalamic-pituitary-adrenal; IGF- 1, insulin-like growth factor-1; VEGF, vascular endothelial growth factor.

2.3.1 Cellular and molecular mechanisms

Abundant evidence, mostly from animal studies, indicates that physical activity promotes angiogenesis and neurogenesis (van Praag et al., 2005), while reducing neuroinflammation (Ryan and Nolan, 2016). Two primary neuro- genic regions exist in the adult mammalian brain, which are the subventric- ular zone of the olfactory bulb and the subgranular zone of the dentate gyrus in the hippocampus (Gage, 2000; Kempermann et al., 2004; Ortega-Perez et al., 2007). In humans, substantial neurogenesis has been proven in the hippocampus and in the striatum which consists of caudate nucleus and puta- men and represents the input region of the basal ganglia. In contrast to mam- mals, no olfactory bulb neurogenesis is detectable in humans (Ernst et al.,

17 2014; Bergmann et al., 2015). Adult hippocampal neurogenesis is acknowl- edged to play a key role in cognition, brain function and disease (Bergmann et al., 2015; Ma et al., 2017). Cognitive impairment, that is evident with normal aging and in neurodegenerative diseases such as Alzheimer’s and Par- kinson’s disease, is attributed to the dysregulation of hippocampal neurogenesis, as well as to neuroinflammation which in turn abates hippocampal neurogenesis (Green and Nolan, 2014; Ryan and Nolan, 2016). Additionally, aerobic exercise also enhances efficient synaptic communication between neural cells (Kandola et al., 2016). Thereby, exercise promotes synaptic plasticity through facilitating long-term potentiation, which is a cellular mechanism of learning and memory, particularly in the hippocampal dentate gyrus (van Praag et al., 1999; Vivar et al., 2013).

The two most studied molecular pathways that link physical activity with cognition include brain-derived neurotrophic factor (BDNF) and insulin-like growth factor-1 (IGF-1) (Gomez-Pinilla and Hillman, 2013). BDNF supports neural survival, growth, and synaptic plasticity (Cowansage et al., 2010). Thereby, BDNF and IGF-1 gene expression and protein levels in different brain areas, mainly in the hippocampus, as well as in the periphery are enhanced with exercise (Duzel et al., 2016). In contrast, blocking of BDNF or IGF-1 signaling pathways attenuates exercise-induced cellular effects and behavioral learning and memory improvements (Vaynman et al., 2004; Ding et al., 2006). Vascular endothelial growth factor (VEGF) is another essential neurotrophin for angiogenesis and adult hippocampal neurogenesis related to physical exercise (Fabel et al., 2003). Increased growth of neurovasculature (angiogenesis) is less regionally specific (Cotman et al., 2007; Voss et al., 2011; Vivar et al., 2013) and precedes and stimulates neurogenesis (van Praag, 2008). Based on animal studies it appears that survival and integration of new neurons into existing cellular networks is a prerequisite for enhanced cognitive performance in association with exercise (Stillman et al., 2016).

Nevertheless, new neurons will die without providing novel experiences and learning opportunities along with increased aerobic exercise (van Praag et al., 1999; Kempermann et al., 2010). Human functional imaging data and rodent studies have demonstrated that aerobic exercise drives the vascular brain changes, whereas coordinative exercise and motor learning induce neuronal adaptations in the brain (Voelcker-Rehage and Niemann, 2013). Thereby, coordinative exercise might result in similar effects as so called “enriched

18 environments” in animal studies (van Praag et al., 1999; Voelcker-Rehage et al., 2011). Therefore, from a neurophysiological viewpoint, it seems most beneficial to include both types of exercise in programs aiming at improving cognition and brain function in older adults. However, in previously sedentary elderly people, aerobic exercise itself might as well include coordinative challenges that have effects of an enriched environment (Voelcker-Rehage and Niemann, 2013).

2.3.2 Brain structural and functional mechanisms

The main age-related losses in gray matter volume apply to the prefrontal cortex, the cingulate cortex, caudate nucleus of the basal ganglia, the medial temporal lobes (e.g. hippocampus), and the ventricles (Erickson et al., 2014; Fletcher et al., 2016). However, aging and physical fitness (or a lack thereof) affect the human brain structure differently. In fact, it was consistently demonstrated that higher cardiorespiratory fitness, as well as higher physical activity levels, are particularly correlated with a larger gray matter volume in the prefrontal cortex and the hippocampus (Erickson et al., 2014; Fletcher et al., 2016), as well the basal ganglia (Fletcher et al., 2016). In contrast, extensive regions of the frontal, parietal, and temporal cortex are exclusively influenced by aging, while several brain regions are under influence of both age and physical fitness (Fletcher et al., 2016). These findings provide back- ing to the hypothesis that effective interventions to boost brain structure in older adults should include additional cognitive training (and/or diet adaptations), rather than physical exercise alone, since these factors might promote brain regions that are not affected by physical exercise training (Fletcher et al., 2016).

The concept to assess gray and white matter brain structure as a mediator for effects of physical activity on cognition is a relatively new one (Stillman et al., 2016). Erickson et al. (2009) first detected in a cross-sectional analysis that higher aerobic fitness was related to larger hippocampus volume and better cognitive performance in elderly humans. Gray matter volume of prefrontal cortex and caudate nucleus of the basal ganglia represent additional brain regions that mediate the association between physical fitness and cognitive executive function and working memory, or cognitive flexibility, respectively, as has been reported in cross-sectional studies (Verstynen et al., 2012;

19 Weinstein et al., 2012). On the other hand, numerous randomized controlled intervention trials have examined exercise-induced effects on cognition or on brain outcomes independently. Particularly, randomized interventions including 6–12 month of moderate exercise have shown to increase hippocampus and prefrontal cortex volume (Erickson et al., 2014). Moreover, cognitive domains associated with the hippocampus were shown to improve after aerobic exercise, including spatial or relational learning and memory, object recognition, and avoidance learning (van Praag, 2008). Nevertheless, only few longitudinal studies assessed both cognitive and brain outcomes that allow to analyze causal interference (Stillman et al., 2016). For instance, Niemann et al. (2014) reported that basal ganglia volume was related to motor coordination and partially to cognitive performance, while, however, training-induced increases in basal ganglia volume were not associated with cognitive improvements (e.g. executive functions). Another seminal study provided first experimental evidence of increased hippocampal volume after 12 months of aerobic walking in previously inactive older adults, while no effects were found in the active control group performing stretching and toning exercises (Erickson et al., 2011). Thereby, changes in both anterior hippocampal gray matter volume and in spatial memory performance were related to changes in exercise performance. However, it was not tested if changes in hippocampal volume mediated the effects of exercise on cognition.

A possible explanation for the regional specificity of brain structural adaptations related to physical fitness or exercise, is that they are specifically involved in the planning, coordination, and execution of movements. Various “fitness sensitive” regions are in fact associated with motor functions such as the primary motor cortex with planning and execution of movements, the basal ganglia with regulation of movement and motor control, the superior temporal sulcus with perception of motion, and the anterior cingulate cortex with coordination of motor behavior (Fletcher et al., 2016). It is noteworthy that by means of brain imaging techniques, such as magnetic resonance imaging (MRI), it is not possible to conclude which microscopic structures have increased in volume (Voelcker-Rehage and Niemann, 2013). Particularly, more than half of the human brain’s gray matter consists of the neuropil, which is composed of axons, dendrites, and glial processes. Less than 20% consists of neural and glial cell bodies, while vasculature accounts for about 5%, and in- terstitial space takes up over 20% of gray matter volume (Thomas et al.,

20 2012). Typically, the increase of structural gray matter volume after physical exercise ranges between 1–8% (Draganski et al., 2004; Scholz et al., 2009; Erickson et al., 2011).

Another possible mechanism mediating exercise related effects on cognition is increased cerebral perfusion. Maass et al. (2015), for example, reported that hippocampal perfusion changes may link exercise effects with both hippocampal gray matter volume and memory performance. Notably, increased cerebral blood flow has also been recognized as a compensatory mechanism to maintain normal memory performance in response of pathologic amyloid-β accumulation in older adults at risk for dementia (Bangen et al., 2017).

Studies assessing white matter integrity found that older adults with high aerobic fitness may exhibit less age-related decline of axon myelination in parts of the corpus callosum (Johnson et al., 2012) and cingulum (Marks et al., 2007; Marks et al., 2011). Furthermore, white matter integrity has been demonstrated to mediate the relation between cardiorespiratory fitness and spatial working memory (Oberlin et al., 2016). The identified white matter tracts comprised the connections of the medial temporal with the prefrontal cortices, which represent the same regions that were related to physical fitness in studies assessing gray matter volume. A recent investigation by Fissler et al. (2017) found no effects of a short-term 10-week physical (or cognitive) training intervention on white matter integrity in older adults at risk for dementia, despite cognitive improvements. Nevertheless, the authors suggested potential long-term effects of engagement in physical (and cognitive) activities due to the positive relation between physical fitness (as well as cognitive abilities) with white matter integrity. This assumption is supported by a 1-year aerobic exercise intervention that demonstrated increases in white matter integrity in temporal and prefrontal regions, as well as in memory performance (Voss et al., 2013). However, no association was evident between these brain and cognitive changes, possibly due to the small sample size in this study. Finally, it is noteworthy that white matter changes also correlate with fall frequency, reduced gait speed, balance and gait impairments, and executive function deficits in older adults (Sartor et al., 2017).

Functional activation or connectivity of distinct brain areas is also modulated through physical activity and exercise, either resulting from structural brain changes or independently. However, functional adaptations have been much

21 less studied than structural, and are mainly focusing on prefrontal cortex functioning, rather than the hippocampus (Stillman et al., 2016). For instance, Voss et al. (2010) demonstrated that functional connectivity was increased in two large-scale brain networks, the default mode and the frontal executive network, after a 1-year walking intervention with healthy older participants. Thereby, changes in the latter network were related to improved executive control. Similar findings were previously reported in a 6-month aerobic training intervention (Colcombe et al., 2004). Together, these results demonstrate the global effects of exercise on brain network efficiency and flexibility, which is known to be attenuated with aging. Therefore, it appears likely that different brain regions are collectively mediating exercise-induced cognitive improvements, rather than any single brain region alone (Stillman et al., 2016).

2.3.3 Behavioral and socioemotional mechanisms

Physical activity possibly also improves cognitive performance through its positive effects on sleep quality and efficiency (Wilckens et al., 2016). In a study including 121 older women, greater levels of physical activity effectively reduced negative effects of poor sleep on performance in executive functioning (Lambiase et al., 2014). Moreover, it was found that increased physical activity is related to improved mood and is effective in reducing symptoms of depression and anxiety (Bridle et al., 2012). On the other hand, low mood is associated with lower performance in various cognitive domains (Lichtenberg et al., 1995; Austin et al., 2001). Nevertheless, it remains unclear if mood mediates effects of physical exercise on cognition (Robitaille et al., 2014; Al- binet et al., 2016).

2.4 THESIS AIMS AND HYPOTHESES

The main aim of the present thesis is to illuminate the effects of interactive simultaneous cognitive–motor training modalities on cognitive and brain function, cognitive–motor dual-task walking, and fall prevention in older adults.

22 Study 1 (chapter 3) addresses the question of how good the cognitive and physical fitness status of Swiss older adults is in relation to the requirements that are necessary to stay independent as a pedestrian in an urban environment. This cross-sectional study with 120 participants also provides a current picture of the relevance of investigating effective training interventions that are tailored to the needs of older adults. Particularly, study 1 aims to identify in a convenience sample of Swiss older adults the proportion that doesn’t reach 1.2 m/s walking speed, which is required to safely cross streets within the green–yellow phase of pedestrian lights, while walking fast under a cognitively challenging dual-task condition. It is hypothesized, that under this fast speed dual-task walking condition the proportion of older adults walking slower than 1.2 m/s would be smaller compared to the preferred speed single- and dual-task walking conditions.

As a second step, a 6-month longitudinal training-intervention study with a 1-year follow-up is performed to investigate both broad cognitive and physical adaptations. The results of this second study are discussed in two separate publications and chapters of this thesis, respectively.

In particular, study 2a (chapter 4) aims to compare two variations of simultaneous cognitive–motor training with an exclusively physical multicomponent program and to evaluate the effects of these programs on cognition in healthy elderly people. It is hypothesized, first, that simultaneous cognitive– motor training may create additional beneficial effects on cognition, and second, that the two cognitive–motor training variations may lead to differential cognitive adaptations. Based on previous findings, reported earlier, cognition is expected to improve in all three programs. Furthermore, it is aimed to investigate the performance maintenance 1 year after the training interventions.

Study 2b (chapter 5) aims to compare two variations of multicomponent simultaneous cognitive–motor training with an exclusively physical program and to evaluate the effects of these programs primarily on DT gait performance in healthy elderly persons, whereas fall frequency and functional fitness represent secondary outcomes. It is hypothesized, first, that simultaneous cognitive–motor training may create additional enhancements on DT gait variables, and second, that the two cognitive–motor training variations may lead to differential gait adaptations. Based on previous findings, reported earlier,

23 improvements in gait, functional fitness, and a reduction in falls are expected after all three programs. Furthermore, it is aimed to investigate performance retention in gait and functional fitness 1 year after the training interventions.

In a third step, a shorter 8 weeks lasting training-intervention study is performed to address the question if the training-induced cognitive behavioral adaptations, that were observed in study 2, would be reflected in brain functional adaptations in older adults.

Therefore, the aim of study 3 (chapter 6) is to compare the effects of cognitive–motor video game dancing against conventional balance training on prefrontal cortex (PFC) activity during walking and on executive functions. It is hypothesized, first, that cognitive–motor video game dancing would elicit larger training-induced reductions of PFC oxygenation during walking than conventional balance training, and second, that training-induced changes in PFC oxygenation would correlate with changes in executive functions.

The interventional studies 2a/b and 3 are to be interpreted as exploratory in nature and serve to generate further hypotheses for future studies. Therefore, in the instances where preplanned hypotheses were tested, no statistical correction procedures to adjust for multiple comparisons were applied (e.g. Bon- ferroni), which reduce the chance of type I errors but at the expense of type II errors. Although the fact that these procedures are discussed with contro- versy, many researchers in the field of biomedical sciences recommend not to apply them under the aforementioned premises (Perneger, 1998; Streiner and Norman, 2011; Armstrong, 2014).

In summary, study 1 will provide current evidence about the cognitive and physical fitness status of Swiss older adults, as assessed with cognitive–motor dual-task gait speed measurements, in relation to the requirements for pedestrians in an urban environment. This first study establishes the exigence for the subsequent interventional studies 2a/b and 3, for which Figure 5 provides an overview within the previously discussed framework of cognition, gait, and falls, as well as the factors mediating effects of physical exercise on cognition.

24 Studies 2a and 3

Study 3 Study 3 Behavioral/ Cognitive- Cellular and Structural and socio- Cognitive molecular functional physical emotional changes brain changes test battery training changes

Study 2b

fNIRS measurment Fear of falling (FES-I) during challenging Geriatric depression walking scale (GDS)

Study 2b Gait analysis

Study 2b Fall frequency

FIGURE 5 | Overview of the targets and outcomes of the interventional studies 2a/b and 3 (adapted from Stillman et al., 2016).

Notes: The cross-sectional study 1 is not included in the graph, however, is providing the rational for the interventional studies depicted here. The parts of the graph printed in faded colors were not covered by any study in the present thesis. Abbreviations: FES-I, falls efficacy scale international; GDS, geriatric depression scale; fNIRS, functional near-infrared spectroscopy.

26 3

Gait speed in older adults (study 1)

27 OLDER ADULTS MUST HURRY AT PEDESTRIAN LIGHTS! A CROSS-SECTIONAL ANALYSIS OF PRE- FERRED AND FAST WALKING SPEED UNDER SINGLE- AND DUAL-TASK CONDITIONS

Patrick Eggenberger1, Sara Tomovic2,3, Thomas Münzer3, and Eling D. de Bruin1

1Institute of Human Movement Sciences and Sport, Department of Health Sciences and Technology, ETH Zurich, 2Institute of Physiotherapy, School of Health Professions, Zurich University of Applied Sciences, Winterthur, 3Ger- iatrische Klinik St.Gallen, Switzerland

Published in PLoS ONE, 31 July 2017: 12(7): e0182180. https://doi.org/10.1371/journal.pone.0182180

28 ABSTRACT

Background: Slow walking speed is strongly associated with adverse health outcomes, including cognitive impairment, in the older population. Moreover, adequate walking speed is crucial to maintain older pedestrians’ mobility and safety in urban areas. This study aimed to identify the proportion of Swiss older adults that didn’t reach 1.2 m/s, which reflects the requirements to cross streets within the green–yellow phase of pedestrian lights, when walking fast under cognitive challenge.

Methods: A convenience sample, including 120 older women (65%) and men, was recruited from the community (88%) and from senior residences and divided into groups of 70–79 years (n=59, 74.8 ± 0.4 y; mean ± SD) and ≥80 years (n=61, 85.5 ± 0.5 y). Steady state walking speed was assessed under single- and dual-task conditions at preferred and fast walking speed. Additionally, functional lower extremity strength (5-chair-rises test), subjective health rating, and retrospective esti- mates of fall frequency were recorded.

Results: Results showed that 35.6% of the younger and 73.8% of the older participants were not able to walk faster than 1.2 m/s under the fast dual-task walking condition. Fast dual-task walking speed was higher compared to the preferred speed single- and dual-task conditions (all p<.05, r= .31 to .48). Average preferred single-task walking speed was 1.19 ± 0.24 m/s (70–79 y) and 0.94 ± 0.27 m/s (≥80 y), respectively, and correlated with performance in the 5-chair-rises test (rs=−.49, p<.001), subjective health (�=.27, p<.001), and fall frequency (�=−.23, p=.002).

Conclusions: We conclude that the fitness status of many older people is inadequate to safely cross streets at pedestrian lights and maintain mobility in the community’s daily life in urban areas. Consequently, training measures to improve the older population’s cognitive and physical fitness should be promoted to enhance walking speed and safety of older pedestrians.

29 3.1 INTRODUCTION

In the older population, reduced usual or preferred walking speed is strongly associated with increased risk for disability, cognitive impairment, falls, and all-cause mortality (Abellan van Kan et al., 2009; Liu et al., 2016). A recent meta-analysis demonstrated that the risk of all-cause mortality is elevated by 89% in the older adults exhibiting the lowest preferred walking speeds (Liu et al., 2016). Walking speed was therefore proposed as a simple geriatric assessment to identify older adults with increased mortality risk and was suggested to be “the sixth vital sign” (Abellan van Kan et al., 2009; Fritz and Lusardi, 2009; Studenski, 2009; Peel et al., 2013; Turner et al., 2014; Liu et al., 2016). The increase in all-cause mortality risk in relation to slow walking speed is mostly associated with increased cardiovascular mortality risk (Du- murgier et al., 2009; Chen et al., 2012). Moreover, reduced walking speed is predictive of community functioning (Studenski et al., 2003) and is an issue for older adults attempting to maintain their mobility and safety as pedestrians in urban areas (Tournier et al., 2016). In contrast to what intuitively might be assumed, aging itself is not a strong explanatory factor for the ob- servable slowing of preferred gait speed in the older population. Healthy older adults walk at a speed that exceeds standards for crossing urban streets; furthermore, they are able to adopt a significantly faster gait in response to a crosswalk signal albeit this comes at the cost of increased gait variability (Brown et al., 2015). It seems that especially potentially modifiable factors such as impairment of Instrumental Activities of Daily Living, physical inac- tivity and cardiovascular disease, are related to the observed slowing of walking speed in many older adults (Busch et al., 2015).

Preferred walking speed of a substantial proportion of older adults is slower than 1.2 m/s (Langlois et al., 1997; Romero-Ortuno et al., 2010; Asher et al., 2012; Bollard and Fleming, 2013; Donoghue et al., 2016). This speed is required, however, at pedestrian lights in Switzerland and many other coun- tries, including the United States, Canada, the United Kingdom, Ireland, and South Africa, to cross streets safely within the green–yellow phase (VSS, 1992; Amosun et al., 2007; Asher et al., 2012; Brown et al., 2015; Donoghue et al., 2016). For instance, a recent comprehensive investigation from Ireland reported that 31% of older adults in the age group of 65–74 years and 61% of those older than 75 years were walking slower than 1.2 m/s at their preferred

30 speed (Donoghue et al., 2016). Another study, from the United Kingdom, found that up to 84% of men and 93% of women older than 65 years of age did not reach 1.2 m/s with normally paced walking (Asher et al., 2012). Recent research indicates that older pedestrians’ mobility and safety is not only negatively affected by physical frailty, which is characterized by slow walking speed, low physical activity, unintentional weight loss, exhaustion, and muscle weakness (Fried et al., 2001), but also by attention deficits (and visual impairment) (Tournier et al., 2016). This finding is supported by other investigations linking executive control and attention deficits with increased fall risk (Herman et al., 2010; Mirelman et al., 2012). Thus, it appears question- able if the measurement of single-task preferred walking speed adequately reflects the physical and cognitive requirements that are co-existing when crossing a street (Nagamatsu et al., 2011; Neider et al., 2011).

Several studies support the notion that cognitive–motor dual-tasking and divided attention play an important role in street crossing behavior of older adults. For instance, older adults were more susceptible to dual-task impairments compared to younger adults in a simulated virtual reality street crossing task. In contrast, younger adults did mostly not show any dual-task costs from listening to music or talking on a cell phone while street crossing (Neider et al., 2011). Dual-task costs are referred to as percentage of loss, relative to the single-task street crossing or walking condition (Neider et al., 2011; Eg- genberger et al., 2015b). Previous research has shown that high dual-task costs for different gait parameters, including stride time, stride velocity and stride length at fast walking speed, are associated with poorer divided attention performance (Dommes et al., 2013). In addition, a divided attention experiment on a simulated street demonstrated that older adults with poorer perceptual, physical, and cognitive functions were prone to make risky street crossing decisions (Butler et al., 2016). Finally, another simulation study identified several predictors of dangerous street-crossing choices which included low walking speed, as well as visual processing speed, visual attention, and attention shifting ability (Dommes et al., 2013). Therefore, measuring dual-task walking speed, including a cognitive task, may represent a better approximation of real-life conditions for crossing streets at pedestrian lights.

To the best of our knowledge, to date only one study has assessed preferred dual-task walking speed as a measure to estimate the proportion of older adults that is at risk when crossing streets at pedestrian lights (Donoghue et

31 al., 2016). The authors demonstrated that the percentage of older adults (≥75 years of age) walking slower than 1.2 m/s at preferred speed rose from 61% under single-task to 91% under dual-task walking conditions, respectively (Donoghue et al., 2016). Furthermore, a recent systematic review identified the lack of research examining the effect of cognitive dual-tasks on gait speed in community-dwelling older adults (Smith et al., 2016). Particularly, studies are inexistent that include a fast speed dual-task walking condition which might reflect time pressure and cognitive challenge when crossing streets at pedestrian lights even more appropriately (de Bruin and Schmidt, 2010).

Therefore, this study aimed to identify in a convenience sample of Swiss older adults the proportion that doesn’t reach 1.2 m/s walking speed, which is required to safely cross streets within the green–yellow phase of pedestrian lights, while walking fast under a cognitively challenging dual-task condition. We hypothesized, that under this fast speed dual-task walking condition the proportion of older adults walking slower than 1.2 m/s would be smaller compared to the preferred speed single- and dual-task walking conditions.

3.2 MATERIALS AND METHODS

3.2.1 Study design and participants

This study represents a cross-sectional evaluation of preferred and fast walking speed under single- and dual-task conditions among 120 older adults living in a Swiss city with about 80’000 residents (Stadt-SG, 2016). We used baseline gait data of 90 older adults enrolled for a six-month training intervention as described previously (Eggenberger et al., 2015a; Eggenberger et al., 2015b). In addition, 30 participants were recruited for this cross-sectional study to increase sample size in the older age group (≥80 y) in order to balance age distribution in the whole sample. We aimed to approximate the proportion of older adults living in senior residences in our sample with the Swiss population average (CURAVIVA, 2014). Data collection was performed at Geriatrische Klinik St.Gallen, Switzerland. The study protocol was approved by the local ethics committee of the canton St.Gallen, Switzerland (study-

32 number: EKSG 14/005, SNCTP 000000039) and the initial training intervention trial was registered at Current Controlled Trials ISRCTN70130279. Our reporting adheres to the STROBE recommendations on what should be included in an accurate and complete report of an observational cross-sectional study (von Elm et al., 2007).

The participants that enrolled to the aforementioned training intervention trial were recruited in August and September 2012 through a newspaper article, a local seniors organization (Pro-Senectute, 2017), senior residence facilities, primary care physicians, and via the websites of the city’s geriatric hospital (Gesundheit-und-Alter, 2017) and the department of sports of the canton St.Gallen (Kanton-SG, 2017); testing sessions succeeded in October 2012. The additional 30 participants were recruited in February and March 2014 among patients receiving ambulatory or inpatient physical therapy at the city’s geriatric hospital and were tested in the same period. For eligibility, participants had to be older than 70 years, live independently or at senior residence facilities, and sign informed consent. Participants had to be able to walk at least 20 meters, with or without walking aids, for gait analysis. Older adults with Alzheimer’s disease, other type of dementia, or recent head injury were excluded. Judgment by their primary care physician was required in the case of acute or instable chronic diseases (e.g. stroke, diabetes) and rapidly progressing or terminal illnesses before accepting a person for participation.

3.2.2 Measurements

Walking speed In the 90 participants that were also enrolled for the training intervention trial, walking speed (among other spatio-temporal gait parameters) was assessed at baseline using the GAITRite electronic walkway system (CIR Sys- tems, Havertown, USA) with the Platinum Version 4.0 software. The validity and reliability of the GAITRite system has been well established (Bilney et al., 2003; van Uden and Besser, 2004; Webster et al., 2005). Walking was initiated two meters before the 7.3-meter active area of the walkway and ended 2 meters thereafter to allow for steady state gait assessment (Ng et al., 2013).

In the group of 30 participants that were recruited additionally for this study, walking speed was assessed using photo-electric barriers (Brower Timing

33 Systems, Draper, USA) since other gait parameters were not required. The two photo-electric barriers were set up at a height of 50 cm and measured steady state walking time over 8 meters distance. Therefore, participants started walking 2 meters before the first photo-electric barrier and were asked to walk steadily until the finish line 2 meters after the second photo- electric barrier. The difference in walking distance (7.3 vs. 8 meters) for speed assessment in the two groups of participants described above is not expected to be a confound, since previous research has shown a high level of validity and comparability for walking speed assessment at differing distances between 5 and 10 meters in older adults (Ng et al., 2013; Middleton et al., 2015; Kim et al., 2016).

The testing protocol comprised single- and dual-task walking, in which participants were instructed to walk under four different conditions: at self-selected preferred walking speed, at a fast walking speed (as fast as possible without running), and each with or without a concurrent cognitive task. Each test condition was repeated three times and the mean value was used for statistical analyses. With participants deemed to be less resilient by the investi- gator, only two repetitions per walking condition were performed to prevent bias from potential fatigue (van het Reve and de Bruin, 2014). Two repetitions per walking condition still represent high measurement reliability for walking speed assessments (Bilney et al., 2003; Goldberg and Schepens, 2011). Trials were repeated when a participant stopped walking or performing the cognitive task. The cognitive–motor dual-task condition was adjusted to the participant’s cognitive abilities to ensure a similar relative cognitive load among all participants. This approach is supported by a meta-analysis that advised to fit the complexity of the cognitive dual-task to the capacity of the target population (Chu et al., 2013). The three levels of cognitive task difficulty that were added to the walking task consisted of either a) counting backwards in steps of seven from a random number between 200 and 250, or b) the same calculation task with steps of three, or c) enumerating objects (e.g. flowers, country names, or first names). The purpose of this procedure was to quantify cognitive–motor interference while walking (Al-Yahya et al., 2011). In order to define the appropriate level of cognitive task difficulty, the participants were first asked to perform the most difficult task (counting backwards in steps of seven) in a standing position without walking. If they were able to accomplish this calculation task in a continuous manner, it was chosen to be

34 performed during the walking assessments. Otherwise, the next easier task (counting backwards in steps of three) was tested without walking, etc. Meta- analytic data showed no differential effects on walking speed among cognitive tasks involving internal interfering factors, such as the mental tracking and verbal fluency tasks applied in the present study (Al-Yahya et al., 2011). These tasks may also share complex neural networks (Fuster, 2003; Gazzaley and D’Esposito, 2006), which interfere with those of gait control (Al-Yahya et al., 2011). Participants were instructed not to prioritize either task and were allowed to use assistive walking devices. Relative dual-task costs (DTC) were calculated as percentage of loss relative to the single-task walking performance, according to the formula DTC (%) = 100 × (dual-task score − single- task score) ⁄ single-task score (Eggenberger et al., 2015b).

Secondary outcomes To assess the association between walking speed and functional lower extremity strength a 5-chair-rises test was performed in accordance with the criteria described for the Short Physical Performance Battery (Guralnik et al., 1994). In addition, the following data were registered: date of birth, sex, educational level (including four options: secondary school = 9 years, appren- ticeship = 13 years, grammar school = 13 years, university/higher education = 17 years), habitation (including four options: independent, apartment for older adults, senior residence, special-care home), subjective health rating (including four options: very good, good, medium, poor), retrospective fall frequency for the 6-month period prior to the study (including three options: no falls, one fall, more than one fall; a fall was defined as “an event which results in a person coming to rest inadvertently on the ground or floor or other lower level” [(WHO, 2016)]), and walking aids (including three options: no walking aid, cane, walking frame).

3.2.3 Statistical analyses

Group differences in the socio-demographic measures of age and education years were compared with student’s independent t-tests. Intra-individual differences between walking conditions were analyzed using student’s paired t- tests. Inter-individual differences within walking conditions between age groups and sexes were analyzed using student’s independent t-tests.

35 Pearson’s correlation for parametric data or Spearman’s correlation and Ken- dall’s � for non-parametric data, were applied to analyze the relationships between walking speed and other parameters. Missing data were excluded from the analysis and are specified in the results section. Statistical calculations were performed with IBM SPSS Statistics software for Macintosh, version 23.0 (IBM Corp., Armonk, NY) with a significance level of α=.05. Effect size r from t-tests, was defined as small at r=.10, medium at r=.30, and large at r=.50 and above (Cohen, 1988).

3.3 RESULTS

TABLE 1 | Socio-demographic characteristics of the two age groups

Variable 70–79 years ≥80 years

Women Men Women Men

N, n (%) 34 (57.6%) 25 (42.4%) 44 (72.1%) 17 (27.9%) Age (years), mean ± SD 74.3 ± 0.5 75.5 ± 0.5 85.0 ± 3.8t 86.8 ± 4.0 Habitation, n (%) independent 32 (94.1%) 25 (100%) 33 (75.0%) 12 (70.6%) apartment for older adults 1 (2.9%) 0 2 (4.5%) 1 (5.9%) senior residence 1 (2.9%) 0 9 (20.5%) 4 (23.5%) special-care home 0 0 0 0 Education (years), mean ± SD 13.0 ± 0.4 13.6 ± 0.4 12.5 ± 2.0* 13.9 ± 2.7

Notes: Bold values indicate significance or trend. t = trend for statistical difference between sexes within age group (p=.097). * = significant statistical difference between sexes within age group (p=.023). Ab- breviations: SD, standard deviation.

36 TABLE 1, continued | Socio-demographic characteristics of the two age groups

Variable 70–79 years ≥80 years

Women Men Women Men

Subjective health, n (%) very good 4 (11.8%) 4 (16.0%) 3 (6.8%) 1 (5.9%) good 18 (52.9%) 11 (44.0%) 19 (43.2%) 10 (58.8%) medium 12 (35.3%) 7 (28.0%) 22 (50.0%) 5 (29.4%) poor 0 2 (8.0%) 0 1 (5.9%) Falls last 6 months, n (%) 0 19 (55.9%) 19 (76.0%) 28 (63.6%) 11 (64.7%) 1 11 (32.4%) 5 (20.0%) 13 (29.5%) 4 (23.5%) >1 4 (11.8%) 1 (4.0%) 3 (6.8%) 2 (11.8%) Walking aids, n (%) none 27 (79.4%) 23 (92.0%) 27 (61.4%) 9 (52.9%) cane 5 (14.7%) 0 9 (20.5%) 5 (29.4%) walking frame 2 (5.9%) 2 (8%) 7 (15.9%) 3 (17.6%)

Table 1 shows the socio-demographic characteristics of the two age groups. The age ranges in the group ≥80 years were 80–96 years for women and 80– 94 years for men, respectively. In each of the following variables one value was missing: education level (women 70–79 y), subjective health (men 70–79 y), and walking aids (women ≥80 y).

3.3.1 Walking speed

Figure 1 illustrates the percentage of women and men that were walking slower than 1.2 m/s under the four walking conditions. Differences in walking speed between sexes were not statistically significant. The highest percentage of participants, comprising women and men, walking slower than 1.2 m/s was evident in the preferred speed dual-task (DT) walking condition (age group 70–79 years: 62.7%, age group ≥80 years: 88.5%), followed by preferred speed single-task (ST) walking (70–79 y: 50.8%, ≥80 y: 82.0%), fast speed DT walking (70–79 y: 35.6%, ≥80 y: 73.8%), and fast speed ST walking (70–79 y: 10.2%, ≥80 y: 42.6%).

FIGURE 1 | Percentage of women and men walking slower than 1.2 m/s.

Notes: The graphs represent each of the four walking conditions (A–D). Abbreviations: DT, dual- task; ST, single-task.

Figure 2 shows the boxplots for the walking speed measurements under the four walking conditions, separated by age group. The younger participants were walking significantly faster under all respective walking conditions (all p<.001, r range from .41 to .43). Walking speed was significantly reduced in the DT walking conditions compared to ST walking (age group 70–79 years, preferred speed DT vs. ST: t(58) = −3.34, p=.001, r=.40; fast speed DT vs. ST: t(58) = −8.86, p<.001, r=.76; age group ≥80 years, preferred speed DT vs. ST: t(60) = −4.26, p<.001, r=.48; fast speed DT vs. ST: t(60) = −12.35, p<.001,

38 r=.85). Additionally, walking speed was significantly lower in preferred ST walking than in fast DT walking (70–79 y: t(58) = −3.09, p=.003, r=.38; ≥80 y: t(60) = −2.50, p=.015, r=.31). No missing values were present for any walking speed measures.

FIGURE 2 | Walking speed under the four walking conditions.

Notes: Mean walking speed was higher in the younger age group in all four respective walking conditions (all p<.001). Mean values of adjacent boxplots in the respective age group are significantly different (all p<.05). The dashed green line represents the required 1.2 m/s walking speed to cross streets safely within the green–yellow phase of pedestrian lights. Abbreviations: DT, dual-task; ST, single-task.

Dual-task costs (DTC) were not different between age groups for preferred speed walking (p=.624), whereas for fast speed walking a trend for significantly smaller DTC in the younger group existed (t(118) = 1.84, p=.069, r=.17). Fast speed DTC were significantly higher than preferred speed DTC in both age groups (70–79 y: t(58) = 7.68, p<.001, r=.71; ≥80 y: t(60) = 4.38,

39 p<.001, r=.49). No differences in DTC between sexes were found. Performance data for all four walking conditions, as well as DTC, are summarized in Table 2.

3.3.2 Correlation with the 5-chair-rises test

A significant association was found between better performance (= shorter time duration) in the 5-chair-rises test and higher preferred and fast ST walking speeds, respectively (rs=−.49, 95% BCa CI [−.628, −.350], p<.001; rs=−.51, 95% BCa CI [−.636, −.355], p<.001, Figure 3). No difference in 5-chair-rises test performance was evident between sexes within the age groups. The younger age group was significantly faster at completing the 5-chair-rises test (t(113) = −2.54, p=.012, r=.23). Missing values in the 5-chair-rises test were present for one male participant in age group 70–79 years, and for two female and two male participants in age group >80 years. Performance data for the 5-chair-rises test are presented in Table 2.

TABLE 2 | Walking speed, dual-task costs, and 5-chair-rises-test performance

Variable 70–79 years ≥80 years

Women Men Women Men

Walking speed (m/s), mean ± SE Preferred DT 1.08 ± 0.05 1.13 ± 0.05 0.83 ± 0.04 0.86 ± 0.06 Preferred ST 1.17 ± 0.04 1.21 ± 0.05 0.94 ± 0.04 0.96 ± 0.06 Fast DT 1.25 ± 0.05 1.33 ± 0.06 0.99 ± 0.05 1.03 ± 0.06 Fast ST 1.52 ± 0.05 1.63 ± 0.06 1.28 ± 0.05 1.31 ± 0.08 Dual-task costs (%), mean ± SE Preferred −7.3 ± 3.2% −6.3 ± 2.4% −8.0 ± 4.0% −10.4 ± 3.6% Fast −17.0 ± 2.5% −18.2 ± 2.2% −22.8 ± 2.3% −20.6 ± 3.0% 5-chair-rises test (s), mean ± SE 11.2 ± 0.5 10.4 ± 0.7 12.9 ± 0.7 11.9 ± 0.9

Abbreviations: SD, standard deviation; DT, dual-task; ST, single-task.

40 FIGURE 3 | Association between fast single-task walking speed and performance in the 5-chair-rises test.

Notes: The gray line represents the regression line (y = 20.53 − 6.09 × x, R2 linear = 0.310).

3.3.3 Correlation with subjective health and fall frequency

Better subjective health ratings were significantly related to higher preferred and fast ST walking speeds, respectively, (�=.27, 95% BCa CI [.134, .404], p<.001; �=.29, 95% BCa CI [.143, .425], p<.001). The same result was found for the relation with the 5-chair-rises test (�=−.22, 95% BCa CI [.072, .352], p=.003). Lower retrospective fall frequency was significantly associated with higher preferred and fast ST walking speeds, and shorter 5-chair-rises test time, respectively (�=−.23, 95% BCa CI [−.360, −.096], p=.002; �=−.20, 95% BCa CI [−.334, −.053], p=.009; �=.27, 95% BCa CI [.123, .414], p<.001). Data for subjective health ratings and retrospective fall frequency are presented in Table 1.

3.4 DISCUSSION

This study aimed to identify in a convenience sample of Swiss older adults the proportion that doesn’t reach 1.2 m/s walking speed, which is required to safely cross streets within the green–yellow phase of pedestrian lights, while walking fast under a cognitively challenging dual-task condition. Our main

41 finding is that 35.6% of Swiss older adults of our convenience sample in the age group of 70–79 years and 73.8% in the age group older than 80 years are not able to walk faster than 1.2 m/s under fast speed dual-task walking. Therefore, it seems feasible to conclude that a similar proportion of older adults is not capable of crossing streets safely within the green–yellow phase of pedestrian lights. These results, first, suggest that the fitness status of many older adults seems not to be appropriate to safely encounter the requirements for pedestrians in urban areas and, second, reinforce the need for regular cognitive and physical training in the older population (Kuh et al., 2014) to keep up with the demands of daily life in the community.

Notably, to date no other study that stated a discrepancy between older adults walking ability and the requirements to cross streets at pedestrian lights, has interpreted their findings in the view of older adult’s lack of fitness. In contrast, these researchers recommended that policy makers should think about reducing the walking speed requirements at pedestrian lights (Langlois et al., 1997; Amosun et al., 2007; Romero-Ortuno et al., 2010; Asher et al., 2012; Bollard and Fleming, 2013; Donoghue et al., 2016). However, we would argue that this recommendation is tackling the problem of impaired walking and street crossing ability in older adults only indirectly. Generally, there are strong indications that cognitive and physical factors are related to the observed slowing of walking speed in older adults (Kuh et al., 2014; Busch et al., 2015), and recent research demonstrated that aging-related sensorial, cognitive and physical changes, have a major negative impact on older pedestrian’s mobility (Tournier et al., 2016). Furthermore, walking speed and cognitive measures (e.g. processing speed and spatial planning) are predictive of unsafe street crossing behavior in old adults (Geraghty et al., 2016). There- fore, the most direct way of enhancing older pedestrians’ safety would include training of their cognitive and physical capacities (Geraghty et al., 2016; Tournier et al., 2016). Particularly, walking speed is modifiable through strength training (Van Abbema et al., 2015) and other physical exercise modalities (Hortobagyi et al., 2015) and is trainable until old age (Colcombe and Kramer, 2003; Smith et al., 2010; Van Abbema et al., 2015). Cognitive components are improvable, for instance, through aerobic training (Colcombe and Kramer, 2003; Smith et al., 2010) or recent combined cognitive–motor training approaches (Eggenberger et al., 2015a). Consequently, older adults should invest time in modifying cognitive and physical capacities to improve their

42 street crossing behavior, which represents one of the main components of pedestrian activity that older adults should be able to perform in order to maintain independence in daily functioning (Tournier et al., 2016).

3.4.1 Proportion of older adults walking slower than 1.2 m/s

A wide range of proportions of older adults walking slower than 1.2 m/s have been described previously (Langlois et al., 1997; Amosun et al., 2007; Romero- Ortuno et al., 2010; Asher et al., 2012; Bollard and Fleming, 2013; Donoghue et al., 2016), ranging from 31% (Donoghue et al., 2016) to over 99% (Langlois et al., 1997). These proportions varied depending on the heterogeneity of the included samples; e.g. participants’ age and health status, walking test conditions (preferred or fast speed, single- or dual-task), measurement methods (steady-state speed or speed including an acceleration phase), and sample size. Regarding these factors, the study by Donoghue and colleagues (2016), examining an Irish nationally representative sample of 4909 participants, seems best suited for comparisons with our study. They similarly applied steady state walking speed measurements and were the only researchers to date that included a preferred speed dual-task walking condition. Nonethe- less, this is the first study applying a fast speed dual-task walking condition to approximate the real-life situation of crossing streets at pedestrian lights under cognitive challenges and time pressure. We found that under this test condition the proportion of older adults walking slower than 1.2 m/s is smaller compared to our preferred speed single- and dual-task walking results, thus confirming our hypothesis. In addition, our novel approach resulted in a more conservative estimation of the proportion of older adults that are at risk when crossing streets at pedestrian lights compared to the preferred speed dual- tasks walking approach applied by Donoghue and colleagues (2016).

Comparing the preferred speed dual-task walking conditions, fewer of our Swiss 70–79 years old group were slower than 1.2 m/s (62.7%) than in the aforementioned Irish 65–74 years old group (76%) (Donoghue et al., 2016). In the older age groups, this difference between the two studies was not present with almost identical values (88.5% Swiss vs. 91% Irish <1.2 m/s) (Donoghue et al., 2016). We hypothesize that the discrepancy in the dual-task condition might have occurred due to disparate cognitive tasks that were added. In our study, a serial subtracting task was applied and adapted to each participant’s

43 abilities. On the other hand, Donoghue et al. (2016) asked their participants to recite alternate letters of the alphabet (e.g. A–C–E etc.). Apparently, in the younger participants, the letters task slowed walking speed more than the subtracting task. This effect might not have come into play in the older participants due to a similarly high percentage of around 90% in both studies. Nevertheless, Al-Yahya and colleagues (2011) asserted a lack of evidence regarding differential effects of specific cognitive tasks on walking speed. Based on meta-analytical data, it was concluded that in general, tasks involving internal interference (e.g. mental tracking tasks, such as those described above) would disturb walking performance more than tasks involving external interference (e.g. reaction time tasks) (Al-Yahya et al., 2011; Chu et al., 2013).

The younger participants in our study (70–79 y) tended to exhibit lower fast walking speed dual-task costs compared to the older participants (≥80 y). This finding might reflect a cognitive advantage, since dual-task effects on walking speed, induced by mental tracking tasks, are strongly related with cognitive status as was shown by meta-analytical evidence (Al-Yahya et al., 2011). In addition, our result is in line with the notion that younger adults (<40 y) demonstrated consistently less interference on walking speed from concurrent cognitive tasks than healthy older adults (Al-Yahya et al., 2011). Simi- larly, Zito and colleagues (2015) reported that older adults looked more at the ground while crossing a virtual street and to a lesser extent at the other side of the street compared to younger adults. Together, these cognitive–motor dual-tasking effects could be explained with the resource-based attentional model which proposes that the total capacity of attentional resources is limited and declines at older age, causing problems in dual-task situations where attentional resources are shared between competing tasks (Woollacott and Shumway-Cook, 2002; Leisman et al., 2016).

In the preferred speed single-task walking condition, about 20% more older adults in our sample walked slower than 1.2 m/s in both age groups (70–79 y and ≥80 y) when compared to the Donoghue et al. (2016) study. This discrepancy might be explained by the fact that participants in the latter study were divided into groups that were five years younger (65–74 y and ≥75 y). Fast speed single-task walking was assessed by one previous study. Amosun and colleagues (2007) tested 47 South African participants aged 65–93 years and reported that 34% walked slower than 1.2 m/s which is between the percent- ages of our two age groups showing 10.2% and 42.6%, respectively.

44 3.4.2 Implications for cognitive and physical training

Recently, Leisman and colleagues (2016) argued that cognitive and motor processes are functionally connected, providing ample clinical and neural evidence supporting their dynamic bidirectional influences. For instance, early impairments of cognitive processes, including attention, executive function, and working memory, were shown to be associated with slowed single- and dual-task walking speed and instability (Montero-Odasso et al., 2012; Gonza- les et al., 2016). The interrelation of cognition and gait in combination with the cognitive–motor dual-task characteristics of crossing streets at pedestrian lights asks for combined cognitive and physical training measures to improve older people’s mobility and safety on public streets (Forte et al., 2015; Zito et al., 2015).

Several studies attained positive effects from physical training on walking speed. For instance, Plummer and colleagues (2015) reported significant effects of physical exercise on single- and dual-task walking speed in their meta-analysis. The interventions included varied types of exercise with or without dual-task components. The authors concluded that improvements in dual-task walking were primarily on account of an increase in walking speed under dual-task conditions and not due to reduced dual-task costs. Another recent meta-analysis by Van Abbema and colleagues (2015) demonstrated that progressive strength training (at 70–80% of the one repetition maximum) was most effective to improve preferred walking speed, followed by exercise with a rhythmic component, and combined strength, balance, and endurance training. This concurs with our finding that higher walking speed correlated significantly with better functional lower extremity strength as measured with the 5-chair-rises test.

Moreover, in a recent combined cognitive–motor multicomponent training intervention with 71 older adults we have shown a 12% increase in preferred walking speed (1.16 ± 0.24 m/s to 1.30 ± 0.25 m/s from pre- to post-test, respectively; mean ± SD) and a concurrent improvement of 13% in preferred dual-task walking speed (1.05 ± 0.29 m/s to 1.19 ± 0.31 m/s) (Eggenberger et al., 2015b). Thereby, it seems noteworthy that an increase in preferred walking speed of 0.1 m/s was established as substantial and meaningful (Perera et al., 2006; Abellan van Kan et al., 2009; Kwon et al., 2009). Interestingly, when translating the data of the latter study into the context of crossing

45 streets at pedestrian lights, 42.6% of the participants (mean age 78.9 ± 5.5 years) were not able to walk faster than 1.2 m/s under dual-task conditions at baseline, whereas after 6 months of training this percentage decreased to 25.5% (calculation based on original data from (Eggenberger et al., 2015b)]). Additionally, the substantial improvements in walking speed were accompanied by broad improvements in various cognitive domains (including executive functions and attention) (Eggenberger et al., 2015a). Such training-induced cognitive improvements appear to be mediated by functional brain adaptations in the prefrontal cortex during challenging walking (Eggenberger et al., 2016). These adaptions might free up attentional resources that could be potentially useful when crossing streets in addition to improvements in walking speed.

3.4.3 Walking speed and health-related parameters

Further support for the benefits of an adequate walking speed level and appropriate lower extremity strength is provided by their association with health-related parameters. Our results indicate a significant association between the subjective health rating and both, walking speed and functional lower extremity strength (5-chair-rises test), respectively. This finding is in line with other studies reporting a strong relationship between reduced walking speed and adverse health outcomes in older adults (Abellan van Kan et al., 2009; Franklin et al., 2015; Liu et al., 2016). Thereby, preferred walking speed below 0.8 m/s (in a 4-meter-walk) was established as a sensible and often used cut-point to identify persons at risk of adverse outcomes (Abellan van Kan et al., 2009). Moreover, our correlation analyses also show significant relations of retrospective fall frequency with both, walking speed and functional lower extremity strength. Similarly, a meta-analysis demonstrated that single- and dual-task walking speed tests were equally applicable to predict falls in older people (Menant et al., 2014). However, findings appear to be controversial in this context, since a recent systematic review concluded that future fall risk is stronger related to dual-task than single-task gait testing (Muir-Hunter and Wittwer, 2016).

46 3.4.4 Strengths and limitations

A methodological strength of this study is the integration of four different walking conditions, comprising single- and dual-task walking at preferred and fast speed, to approximate real-life street crossing behavior at pedestrian lights. However, some limitations should be considered. First, the applied measurements of gait speed do only reflect an approximation of the conditions at real pedestrian lights, since the measurements were performed indoors in a protected environment, which implies a lack of ecological validity (Corrigan and McBurney, 2008). A second limitation of this study is that it includes a convenience sample of participants that either intended to participate in a prospective training intervention or that were additionally recruited for cross-sectional walking speed assessments. Therefore, it is possibly not representative for the general population of (Swiss) older adults and the data does not reflect normative aging. However, because average preferred single- and dual-task walking speeds in our heterogenic sample were comparable to values reported in two recent meta-analyses (Smith et al., 2010; Bohannon and Williams Andrews, 2011), we assume that our measurements are reason- ably representative. Moreover, the distribution of participants living independently vs. those living in senior residences was approximately in the range of the distribution in the Swiss older population (70–79 y: 1.7% of our sample living in senior residences vs. 1.7–4.8% of Swiss population, respectively; ≥80 y: 21.3% vs. 8.5–30.3%) (CURAVIVA, 2014). Another, third, limitation of the presented data is that actual crosswalk speed requirements do not always follow the 1.2 m/s guideline, but appear to vary widely (Amosun et al., 2007; Bollard and Fleming, 2013; Salbach et al., 2014). A recent systematic review, for instance, reported mean walking speed requirements of 0.44, 0.73–0.78, and 1.32 m/s, in the large cities of Melbourne, Singapore, and Los Angeles, respectively (Salbach et al., 2014). However, considering the limited number of crossroads that were analyzed in these studies, it still seems adequate to compare older adult’s walking speed with the 1.2 m/s standard recommended by national traffic authorities, including the United States, Canada, the United Kingdom, Ireland, South Africa, and Switzerland (VSS, 1992; Langlois et al., 1997; Amosun et al., 2007; Romero-Ortuno et al., 2010; Asher et al., 2012; Bollard and Fleming, 2013; Brown et al., 2015; Donoghue et al., 2016).

47 3.5 CONCLUSIONS

This study demonstrates that about every third (35.6%) older person at the age of 70–79 years and almost three-quarters (73.8%) of persons ≥80 years cannot walk faster than 1.2 m/s, which is required to cross streets safely within the green–yellow phase of pedestrian lights, under cognitively challenging conditions (fast speed dual-task walking). We propose that fast speed dual-task walking is more reflective of the time pressure and cognitive challenge at real pedestrian lights than the previously applied assessments of single- or dual-task walking at preferred speed. This novel approach led to a more conservative estimation of the proportion of older adults that are unable to walk as fast as 1.2 m/s compared to a previous investigation assessing dual- task walking at preferred speed (Donoghue et al., 2016). Nonetheless, this proportion is still alarmingly high, given the fact that slowed walking speed is related to various adverse health outcomes (Abellan van Kan et al., 2009; Montero-Odasso et al., 2012; Gonzales et al., 2016; Liu et al., 2016).

We therefore strongly recommend acknowledging the results of this and other similar studies as evidence that the fitness status of many older people is inadequate for maintaining their mobility-related independence in daily life and their safety as pedestrians in urban areas. In the long run, it clearly appears more fruitful to promote direct measures that improve the older population’s cognitive and physical fitness instead of tackling the problem indirectly by adapting pedestrian light’s settings as proposed by preceding studies (Langlois et al., 1997; Amosun et al., 2007; Romero-Ortuno et al., 2010; Asher et al., 2012; Bollard and Fleming, 2013; Donoghue et al., 2016). Slow walking speed, as the main cause of the problem, is modifiable particularly through strength and other training modalities (Hortobagyi et al., 2015; Van Abbema et al., 2015), while the cognitive, attentional components are improvable through aerobic (Colcombe and Kramer, 2003; Smith et al., 2010) and other exercise training modes (Voelcker-Rehage et al., 2011; Chang et al., 2012), or recent promising combined cognitive–motor training approaches (Eggenberger et al., 2015a; Eggenberger et al., 2016). Future research should, therefore, investigate combinations of cognitive and physical training strategies targeting the dual-task requirements of safe street crossing behavior related to gait speed and assess their effectiveness in ecologically valid settings.

48 Acknowledgments This work was supported by the Zürcher Kantonalbank within the framework of sponsoring of movement sciences, sports, and nutritional sciences at ETH Zurich. Zürcher Kantonalbank had no influence on the study design and the analyses presented in this paper, had no access to the data, and did not contribute to this manuscript in any way. The authors would like to thank the management of Geriatrische Klinik St.Gallen, Switzerland, for supporting the study and providing room for data acquisition. Furthermore, we thank our postgraduate students, Marius Angst and Stefan Holenstein for helping with data acquisition. We highly appreciated the support of the team of physiotherapists at Geriatrische Klinik St.Gallen. We would also like to thank all participants for their kindness and patience during the testing sessions.

Author contributions Conceptualization: PE, ST, TM, EDB Formal Analysis: PE, ST Investigation: PE, ST Methodology: PE, ST, TM, EDB Resources: PE, ST, TM, EDB Supervision: TM, EDB Visualization: PE Writing – Original Draft Preparation: PE Writing – Review & Editing: PE, ST, TM, EDB

Disclosure The authors report no conflicts of interest in this work.

Copyright notice This work was initially published by PLoS ONE and licensed under Creative Commons Attribution 4.0 International (CC BY 4.0) https://creativecom mons.org/licenses/by/4.0/, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. The following changes were applied to the original manuscript: the term “cognitive–physical” was replaced by the term “cognitive–motor” to use consistent terminology throughout the manuscript of this doctoral thesis.

50 4

Cognitive performance adaptations (study 2a)

51 DOES MULTICOMPONENT PHYSICAL EXERCISE WITH SIMULTANEOUS COGNITIVE TRAINING BOOST COG- NITIVE PERFORMANCE IN OLDER ADULTS? A 6- MONTH RANDOMIZED CONTROLLED TRIAL WITH A 1-YEAR FOLLOW-UP

Patrick Eggenberger1, Vera Schumacher2,3, Marius Angst1, Nathan Theill4,5, and Eling D. de Bruin1

1Institute of Human Movement Sciences and Sport, Department of Health Sciences and Technology, ETH Zurich, 2Department of Gerontopsychology and Gerontology, 3University Research Priority Program “Dynamics of Healthy Aging”, University of Zurich, 4Division of Psychiatry Research, Uni- versity of Zurich, Schlieren, 5Center for Gerontology, University of Zurich, Switzerland

Published in Clinical Interventions in Aging, 17 August 2015: 10, 1335–1349. http://dx.doi.org/10.2147/CIA.S87732

52 ABSTRACT

Background: Cognitive impairment is a health problem that concerns almost every second elderly person. Physical and cognitive training have differential positive effects on cognition, but have been rarely applied in combination. This study evaluates synergistic effects of multicomponent physical exercise complemented with novel simultaneous cognitive training on cognition in older adults. We hypothesized that simultaneous cognitive–motor components would add training specific cognitive benefits compared to exclusively physical training.

Methods: Seniors, older than 70 years, without cognitive impairment, were randomly assigned to either: 1) virtual reality video game dancing (DANCE), 2) treadmill walking with simultaneous verbal memory training (MEMORY), or 3) treadmill walking (PHYS). Each program was complemented with strength and balance exercises. Two 1-hour training sessions per week over 6 months were applied. Cog- nitive performance was assessed at baseline, after 3 and 6 months, and at 1-year follow-up. Multiple regression analyses with planned comparisons were calculated.

Results: Eighty-nine participants were randomized to the three groups initially, 71 completed the training, while 47 were available at 1-year follow-up. Advantages of the simultaneous cognitive–motor programs were found in two dimensions of executive function. “Shifting attention” showed a time intervention interaction in favor of DANCE/MEMORY versus PHYS (F(2, 68) = 1.95, trend p=.075, r=.17); and “working memory” showed a time intervention interaction in favor of DANCE versus MEMORY (F(1, 136) = 2.71, trend p=.051, R2=.006). Performance improvements in executive functions, long-term visual memory (episodic memory), and processing speed were maintained at follow-up in all groups.

Conclusions: Particular executive functions benefit from simultaneous cognitive– motor training compared to exclusively physical multicomponent training. Cogni- tive–motor training programs may counteract widespread cognitive impairments in the elderly.

Keywords: elderly, executive function, transfer, cognitive impairment, dance, video game

53 4.1 INTRODUCTION

A decrease in cognitive performance in old age is predominant in most individuals. This was confirmed by a large Italian epidemiological study demonstrating that aging-associated cognitive decline has a prevalence rate of 28% for people from 65 years to 84 years (Scafato et al., 2010). Additionally, another 17% of this Italian population (N=4’785) showed objective evidence of cognitive decline without cognitive complaints, which add up to a total of 45% of people showing some kind of cognitive impairment without dementia. Since cognitive decline potentially threatens independence and quality of life for older adults, prevention and treatment of cognitive impairment in the elderly has assumed increasing importance (Williams and Kemper, 2010). Two factors that may positively affect cognition in the elderly are physical activity and cognitive training.

Research has pointed out recently that physical activity may be relevant for healthy brain aging and may protect from cognitive decline and dementia (Er- ickson et al., 2012; Lautenschlager et al., 2012; Gomez-Pinilla and Hillman, 2013; Gregory et al., 2013; Hotting and Roder, 2013). Most physical intervention studies that focused on adaptations in cognitive performance, brain function, or brain structure, applied aerobic type exercise. Two meta-analytic studies reported that aerobic exercise is effective in increasing cognitive performance, in general, and executive function in particular (Colcombe and Kra- mer, 2003; Smith et al., 2010). More recent studies also found that strength and coordination training may positively affect cognitive abilities (Voelcker- Rehage et al., 2011; Chang et al., 2012). Voelcker-Rehage et al. (2011) demonstrated training specific functional plasticity in the brain based on functional magnetic resonance imaging data. Thereby, aerobic training increased activation in the sensorimotor network and coordination training led to a higher activation of the visuospatial network (Voelcker-Rehage et al., 2011), whereas strength training changed the hemodynamic activity of brain regions associated with response inhibition processes (Liu-Ambrose et al., 2012).

Cognitive training studies have often shown highly task specific effects (No- ack et al., 2009; Papp et al., 2009; Zelinski, 2009; Hindin and Zelinski, 2012; Oei and Patterson, 2014). More widespread transfer effects were found when different cognitive abilities were combined in complex interventions or lifestyle changes (Lustig et al., 2009). Nevertheless, effects were often small,

54 while aerobic training elicited both broad transfer and relatively large effects (Lustig et al., 2009). These findings led to the assumption that not only the combination of different cognitive abilities but also the combination of cognitive and physical training improves cognitive performance in old age to a greater extent than the training of an isolated ability (Lustig et al., 2009; Schaefer and Schumacher, 2011; Thom and Clare, 2011; Kraft, 2012; Gregory et al., 2013; Hotting and Roder, 2013; Law et al., 2014). Therefore, more and more studies pursue exactly this goal by administering a combined cognitive– motor training approach including cognitive and gross motor physical exercise components.

Some studies applied the physical and cognitive training sessions in a sequential manner (Oswald et al., 1996; Fabre et al., 2002; O'Dwyer, 2009; Legault et al., 2011), whereas others performed the physical and cognitive training units simultaneously (Pichierri et al., 2012a; Pichierri et al., 2012b; Forte et al., 2013; Theill et al., 2013). An advantage of simultaneous training designs might be that they include dual tasking and switching attention between the cognitive and physical activity. For instance, Theill et al. (2013) investigated the effects of simultaneous memory training and treadmill walking and revealed benefits in cognitive–motor dual-task walking compared to a single cognitive training and a passive control group. Virtual reality video game dancing represents a novel mode of simultaneous cognitive–motor training and has been applied by Pichierri et al. (2012a; 2012b). Intervention groups demonstrated increased cognitive–motor dual-task performance in a stepping accuracy task (Pichierri et al., 2012a) or during fast walking (Pichierri et al., 2012b), respectively. Nonetheless, interpretation of the existing studies on combined cognitive–motor training is often limited due to small sample sizes (Fabre et al., 2002; O'Dwyer, 2009; Pichierri et al., 2012a; Pichierri et al., 2012b), inconsistent training exposures between intervention groups (Oswald et al., 1996; Fabre et al., 2002; Legault et al., 2011; Pichierri et al., 2012a; Pichierri et al., 2012b), or the lack of reference groups with only physical training (Pichierri et al., 2012a; Theill et al., 2013). Moreover, transfer to different cognitive domains was not assessed in some studies (Pichierri et al., 2012a; Pichierri et al., 2012b; Forte et al., 2013) and most interventions lasted for 4 months at most (Fabre et al., 2002; O'Dwyer, 2009; Legault et al., 2011; Pichierri et al., 2012a; Pichierri et al., 2012b; Forte et al., 2013; Theill et al., 2013). This duration might be too short since physical training interventions

55 of 6 months or longer have shown most consistent effects on cognition (Col- combe and Kramer, 2003; Erickson et al., 2012). Therefore, we suggest that the promising findings in previous research are worth further investigation.

This study aims to compare two variations of simultaneous cognitive–motor training with an exclusively physical multicomponent program and to evaluate the effects of these programs on cognition in healthy elderly people. We hypothesize, first, that simultaneous cognitive–motor training may create additional beneficial effects on cognition, and second, that the two cognitive– motor training variations may lead to differential cognitive adaptations. Based on previous findings, reported earlier, we expect cognition to improve in all three programs. Furthermore, we aim to investigate the performance maintenance 1 year after the training interventions.

4.2 MATERIALS AND METHODS

4.2.1 Study design and participants

This study was a randomized, controlled trial (RCT), including a three groups parallel 6-month training intervention and a 1-year nonintervention follow- up. Assessments of cognitive performance were performed four times: pretraining, after 3 months, 6 months (posttraining), and at 1-year follow-up. Data collection and training were performed at Geriatrische Klinik St.Gallen, Switzerland. The study protocol was approved through the local ethics committee of the canton St.Gallen, Switzerland (study number: EKSG 12/092) and registered at Current Controlled Trials ISRCTN70130279. No changes were made to the planned methods after trial commencement. Our reporting in the manuscript adheres to the CONSORT 2010 guidelines (Moher et al., 2010).

Participants were recruited through a newspaper article, a local seniors organization (Pro Senectute St.Gallen), senior residence facilities, primary-care physicians, and via the websites of the city’s geriatric hospital and the department of sports of the canton St.Gallen. Interested persons were invited to an information event. We included male and female participants because

56 both sexes are similarly affected by age-related cognitive decline (Scafato et al., 2010). For eligibility, participants had to be older than 70 years, live independently, or at senior residence facilities, and had to sign the informed consent. Residents of retirement homes classified as 0, 1, or 2 within the Swiss classification system for health care requirements (BESA-levels, Ger- man abbreviation for Bewohner-Einstufungs- und Abrechnungs-System) could enroll in the study. Level 0 means the person does not need care or treatment; levels 1–2 means, the person only needs little care or treatment. Seniors with diagnosed Alzheimer’s disease, dementia, recent head injury, or a score < 22 points (MacKenzie et al., 1996) on the Mini Mental Status Exam- ination (MMSE, (Folstein et al., 1975), which indicates cognitive impairment were excluded. Judgment by their primary care physician was required in the case of acute or instable chronic diseases (e.g., stroke, diabetes), rapidly progressing or terminal illnesses before accepting a person for participation.

A priori power analysis (G*Power 3.1.3 Software, (Faul et al., 2007) revealed a sample size of 75 participants in order to achieve 80% power for a three group pretest, 3- and 6-month test design (25 participants per group). The α- level was set at 0.05 and the effect size f at 0.3. The randomization scheme was generated with the website Randomization.com (Randomization, 2017). Applying block randomization to achieve three groups with a ratio of 1:1:1. Participants were blinded to the expected study outcome, while blinding of the investigators was not possible since they supervised and conducted training and testing sessions.

4.2.2 Training programs

Two 1-hour training sessions per week were performed in groups of five to six participants, under the instruction of two trained postgraduate students. At least 1 day was included between sessions for recovery. Training programs were based on current recommendations for physical fitness and fall prevention for the elderly (Chodzko-Zajko et al., 2009; Sherrington et al., 2011; Gra- nacher et al., 2012). The three multicomponent programs consisted of 20 minutes aerobic endurance training (either video game dancing, treadmill memory training, or treadmill walking) and complementary strength and balance exercises (20 minutes each). The exercise training principles of progression and overload were applied for every training component (Ammann et al.,

57 2014), and they were adapted to each participant’s abilities such that a moderate to vigorous intensity was achieved (Chodzko-Zajko et al., 2009). In total, 52 sessions were performed within 6 months (26 weeks), with some participants missing certain sessions due to personal reasons. Sessions 25–32 (4 weeks) were performed individually according to a home exercise plan, due to Christmas holiday and 3-month test sessions. The home exercise plan comprised the same strength and balance exercises as instructed during normal training sessions, but no video game dancing and treadmill memory training. Compliance to the home exercise plan was assessed with a training diary.

Video game dancing (DANCE) Program DANCE included virtual reality video game dancing as a simultaneous cognitive–motor training (Figure 1A). This training component combines an attention demanding cognitive task with a simultaneous motor coordination aspect. We used two Impact Dance Platforms (Positive Gaming BV, Haarlem, the Netherlands) and created various levels of difficulty in step patterns and frequency with the StepMania Software (StepMania, 2016). Sev- eral styles of music were selected to add variety and meet preferences of participants. Participants stood on the one-by-one meter platform, which contained four pressure sensitive areas to detect steps forward, backward, to the left, and to the right, respectively. Stepping sequences were cued with arrows appearing on a large screen and had to be performed exactly when an arrow reached a highlighted area on the screen in order to achieve best scores in the game. Participants were holding on to ropes for security reasons. Training difficulty was adapted to each individual’s coordination ability and was increased progressively.

Treadmill memory training (MEMORY) Program MEMORY comprised treadmill walking with verbal memory exercise as a simultaneous cognitive–motor training (Figure 1B). Verbal memory training consisted of a computer-based serial position training that was presented on a computer screen in front of the treadmill, with a standard computer mouse as an input device. E-Prime 2.0 Professional software (Psychol- ogy Software Tools, Pittsburgh, PA, USA) was used to program the training. Participants were asked to memorize the correct sequence of 3–20 words lighting up one after the other for 3 seconds on the computer screen.

58 Thereafter, a distraction task was followed where participants had to define if three presented words had a meaning or not. Then, the initially memorized words were presented again, either in the same or another sequence, and participants had to decide if the sequence remained the same or not, by pressing the mouse button. The initial level for this training was set at a sequence of three words and was extended by one word as soon as the participants reached 80% of correct answers within the level. Treadmill speed and inclination were set individually for each participant, such that a subjective rate of perceived exertion of five to seven points on the ten-point Borg scale was reached as recommended by the American College of Sports Medicine (ACSM) position stand on exercise with older adults (Chodzko-Zajko et al., 2009).

FIGURE 1 | Simultaneous cognitive–motor training components.

Notes: A; video game dancing. B; treadmill memory training. In (A) two participants perform steps on a pressure sensitive platform to the rhythm of the music. Step timing and direction are cued with arrows on a screen. In (B) a participant is walking on a treadmill while performing verbal memory exercises presented on a computer screen.

Treadmill walking (PHYS) Program PHYS included aerobic treadmill walking without an additional cognitive task and acted as a reference group with exclusively physical training components. Participants were walking or running at a constant pace. Tread- mill speed and inclination were set individually for each participant, such

59 that a subjective rate of perceived exertion of five to seven points on the ten- point Borg scale was reached (Chodzko-Zajko et al., 2009).

Complementary strength and balance exercises In addition to one of the three different aerobic training components described earlier, muscular strength and balance exercises complemented each program (Figure 2). Four to five strength exercises for lower and upper extrem- ities and trunk stabilization were performed using own body weight, resistive rubber bands, and weight vests of maximally 10 kg (1–3 sets, with 8–12 repetitions, at slow to fast movement speed). Number of sets and repetitions were adapted individually for each participant, such that a subjective rate of perceived exertion of five to seven points on the ten-point Borg scale was reached (Chodzko-Zajko et al., 2009). Balance training consisted of different exercises including two- and single-leg stance variations, either on the floor or on various types of instable surfaces (e.g., foam and air pads, ropes, etc.) (de Bruin and Murer, 2007). Exercise level, volume, and intensity were chosen according to the participants’ individual abilities and increased progressively.

FIGURE 2 | Examples of complementary balance (A) and strength (B) exercises.

Notes: The participant in (A) tries to maintain balance while stepping from one object to the next (objects are soft rubber “stones” and a skipping rope) and (B) shows a participant performing split leg squats wearing a weight vest.

60 4.2.3 Measurements

Cognitive tasks (primary outcome) Cognitive performance was measured by applying a test battery including nine “paper-and-pencil” tasks to assess transfer to different cognitive domains. Four of these tests were repeated at 1-year follow-up, while the other tests were excluded to reduce test time for participants. Executive function was measured with the Trail Making Test Part B (TMT-B) (Lezak et al., 2012), working memory was assessed with the Executive Control Task (Baller et al., 2006), and long-term visual memory was tested with three different parallel versions of the Paired-Associates Learning task (Baller et al., 2006); furthermore, long-term verbal memory was assessed with the German version (Härting et al., 2000) of the Logical Memory subtest (Story Recall) from the Wechsler Memory Scale-Revised (WMS-R) (Wechsler, 1987), whereby only one of two different stories from the original test was presented (story A) and no delayed recall after 30 minutes was performed; moreover, short-term verbal memory was measured with the Digit Forward and Backward Tasks from WMS-R (Wechsler, 1987), attention was tested with three different parallel versions of the Age Concentration Tests A and B (Gatterer, 2008) (as an adaptation to the original test, we calculated “number of correct figures” divided by “time” as the test result); and finally, information processing speed was assessed with the Trail Making Test Part A (TMT-A) (Lezak et al., 2012) and the Digit Symbol Substitution Task (DSST) from Wechsler Adult Intelli- gence Scale-Revised (WAIS-R) (Wechsler, 1981).

Training enjoyment (secondary outcome) Overall training enjoyment was assessed at 6-month test using the German eight-item version of the Physical Activity Enjoyment Scale (PACES) (Mullen et al., 2011; Jekauc et al., 2013). The average score of the eight items was used for statistical analysis. Additionally, we asked participants specifically about their enjoyment of the balance and the strength training, as well as the video game dancing, the treadmill memory training, or the treadmill walking. Thereby, we used the same scoring system from one to seven points (least to most enjoyment) as in the PACES. We assumed that training with cognitive

61 elements would be enjoyed more than treadmill walking and that video game dancing would be enjoyed more than treadmill memory training (Graves et al., 2010).

4.2.4 Statistical analyses

Group differences in the baseline demographic and performance data were compared with one-way analysis of variance (ANOVA). Multiple regression analysis with planned comparisons, including orthogonal contrast and polynomial trend coding, were applied to investigate training effects on the cognitive test battery for the 6-month training period. We produced contrast coding variables based on the hypotheses. The first contrast was set to compare the two combined cognitive–motor training groups with PHYS. The second contrast compared the two cognitive–motor training groups (DANCE versus MEMORY). According to the study design, comprising three time points of measurement, we created polynomial trend coding variables to assess the linear and quadratic trend. Effect code variables were produced for each group’s individuals to account for subject effects. Repeated measures ANOVA with Bonferroni correction was applied for post hoc comparisons from pretest to 3- month test and from 3- to 6-month tests (p=.025 for two comparisons). Re- peated measures of ANOVA were also used to assess differences between 6- month test and 1-year follow-up. Missing values from participants who completed the full 6-month trial but missed single test items due to health constraints or social obligations were replaced by the group mean value at the respective time point of measurement. One-way ANOVA with planned contrasts was performed to compare group differences in the training enjoyment questionnaire. Statistical calculations were performed with IBM SPSS Sta- tistics software for Macintosh, version 22.0 (IBM Corp., Armonk, NY, USA) with a significance level of �=.05. Effect sizes, represented as R2-change in the multiple regression analysis, were considered as small for R2-change=.01, medium for R2-change=.06 and large for R2-change=.14 and above; effect size r from one-way ANOVA, was defined as small at r=.10, medium at r=.30, and large at r=.50 and above (Cohen, 1988).

62 4.3 RESULTS

Out of 89 participants initially enrolled, 71 participants completed the 6- month training intervention (20.2% attrition) and were included in the analysis of the outcomes derived at pretest, 3- and 6-month tests. Time points and reasons for dropouts are presented in Figure 3.

FIGURE 3 | Trial design and participants’ flow.

Notes: Participants were randomly assigned to one of two simultaneous cognitive–motor training groups (DANCE and MEMORY) or an exclusively physical multicomponent training group (PHYS) and were trained over 6 months twice weekly for 1 hour. Nine cognitive tests were assessed at pretest, 3-month test, and 6-month test. Four tests were repeated at 1-year follow-up. Abbreviations: DANCE, virtual reality video game dancing; MEMORY, treadmill walking with simultaneous verbal memory training; PHYS, treadmill walking.

63 Dropouts were equally distributed between groups, and therefore, the final analyses were performed only in individuals who completed the 6-month intervention. Forty-seven participants were available for the 1-year follow-up test session and were included in the analysis of these outcomes. The following missing values from persons who completed the 6-month training were replaced by the group mean value: at pretest, three persons from DANCE and two persons from MEMORY missed TMT A and B, one person from DANCE missed four other items, and one person from PHYS missed seven test items; at 3-month test, one person from MEMORY missed all nine tests (this person did not miss any pretests). No missing values were evident at 6 months and follow-up tests. One cognitive task, the Digit Backward Task, was not analyzed because some participants applied a strategy that defeated the idea of the test. Participants’ recruitment lasted from August 2012 until the end of September 2012, when pretests were performed. The training intervention lasted from October 2012 until the end of March 2013, with 3-month test at the beginning of January 2013 and 6-month test at the beginning of April 2013. One year later, in April 2014, follow-up test was performed.

TABLE 1 | Baseline demographic characteristics and training compliance

Variable DANCE MEMORY PHYS p-value, two-tailed

N 24 22 25 Sex, female 14, 58.3% 16, 72.7% 16, 64.0% .602 Age, years 77.3 (6.3) 78.5 (5.1) 80.8 (4.7) .079t MMSE, score 28.4 (1.4) 28.3 (1.2) 28.0 (1.7) .533 Education, years 13.7 (1.5) 13.9 (2.1) 12.0 (2.1) .002** Total training compliance 84.3% (12.7%) 86.1% (9.1%) 87.1% (7.9%) .633 (52 sessions) Home-training compliance 79.9% (23.0%) 90.0% (14.8%) 83.5% (18.4%) .201 (eight sessions)

Notes: Data are means (± standard deviation in brackets) or numbers. Bold values indicate significance or trend, ** p<.01, t p<.10 trend. Abbreviations: MMSE, Mini Mental State Examination; DANCE, virtual reality video game dancing; MEMORY, treadmill walking with simultaneous verbal memory training; PHYS, treadmill walking.

64 Table 1 shows baseline demographic characteristics and training compliance of the three intervention groups. Baseline cognitive performance data did not show significant differences between intervention groups for any of the nine cognitive transfer tests (TMT-A p=.351; TMT-B p=.334; Executive Control p=.652; Paired Associates p=.156; Story Recall p=.655; Digit Forward p=.458; Age Concentration A p=.390; Age Concentration B p=.346; DSST p=.548; all p-values two-tailed).

4.3.1 Cognitive tasks

Figures 4 and 5 depict performance development for the nine cognitive tasks. Statistical details of the multiple regression analysis over the first three time points of measurement, including two planned comparisons or contrasts, are provided in Table 2. In eight of nine cognitive tasks, except Digit Forward Task, linear global time effect showed significant performance improvement from pretest to 6-month test in each of the three intervention groups. The analysis of performance maintenance from 6 months to follow-up test is shown in Table 3. Performance remained unchanged until 1-year follow-up test in three cognitive tasks and increased significantly in TMT-B.

The first contrast in the multiple regression analysis tested if the two simultaneous cognitive–motor interventions performed better compared to PHYS. Thereby no significant time intervention interaction was found. Additional post hoc comparison for performance development in the TMT-B from pretest to 3-month test showed a small to moderate effect with a trend to significance for the time intervention interaction between DANCE/ MEMORY versus PHYS: the two groups with a cognitive training component reduced their time to complete the task, while PHYS was performing slower (F(2, 68) = 1.95, trend p=.075 (one-tailed for directional hypothesis), r=.17). No other trend or significant time intervention interaction was found for post hoc comparisons of the two separate 3-month training periods (data are not presented).

FIGURE 4 | Cognitive performance developments in the four tests that included a 1-year follow-up measurement.

Notes: Significant overall improvements were shown in all tests over the 6-month training period (graphs A–D all p<.05, one-tailed). In Trail Making B (graph B), only the two groups with a cognitive training component (DANCE and MEMORY) improved from pretest to 3-month test (trend p=.075, one-tailed). In Executive Control (graph C), different time courses of adaptation between DANCE and MEMORY were found (trend p=.051, one-tailed). From 6-month test to 1-year follow-up test Trail Making B improved significantly (graph B, p=.015), while performance was maintained in the three other tests (graphs A, C, and D). Error bars indicate ± standard error of the mean. Abbreviations: DANCE, virtual reality video game dancing; MEMORY, treadmill walking with simultaneous verbal memory training; PHYS, treadmill walking.

The second contrast tested differences between the two cognitive–motor interventions. There was a trend to a significant linear time intervention interaction between DANCE and MEMORY in the Executive Control Task from pretest to 6-month test, reflecting different time courses of adaptation: DANCE improved continuously, while MEMORY showed an improvement over the first 3 months and a decrease of performance, back to baseline level, after the second 3 months of training (F(1, 136) = 2.71, trend p=.051 (one- tailed for directional hypothesis), R2-change =.006). Additional post hoc comparison of the development in the Executive Control Task from 3-month to 6-

66 month tests revealed a significant time intervention interaction with a small to moderate effect, also reflecting the aforementioned improvement for DANCE and the decline in MEMORY (F(2, 68) = 3.20, p=.024 (one-tailed for directional hypothesis), r=.21).

FIGURE 5 | Cognitive performance developments in the five tests that did not include a 1- year follow-up measurement.

Notes: Significant overall improvements were shown in the tests in graphs (A, C, D, and E) (all p<.05, one-tailed) over the 6-month training period. No improvement was found in Digit Forward (graph B). Error bars indicate ± standard error of the mean. Abbreviations: DANCE, virtual reality video game dancing; MEMORY, treadmill walking with simultaneous verbal memory training; PHYS, treadmill walking.

67 TABLE 2 | Multiple regression for the linear global time effect (from pretest to 3- and 6-month tests, N=71) and the interaction between orthogonal contrasts and time effect for the cognitive test battery

Dependent variable Predictor b 95% CI SE b β p, one- R2 - (cognitive domain) tailed change

Trail Making Part A ABC -4.93 -6.81 -3.05 0.95 -.22 <.001*** .049 (information processing speed) AB x C 0.78 -0.53 2.09 0.66 .05 .120 .003 A x B 0.06 -2.27 2.40 1.18 .00 .479 .000 Trail Making Part B ABC -5.57 -11.46 0.32 2.98 -.08 .032* .006 (executive function, shifting) AB x C -0.93 -5.03 3.18 2.08 -.02 .328 .000 A x B -1.76 -9.07 5.56 3.70 -.02 .318 .000 Executive Control ABC 0.70 0.14 1.27 0.29 .11 .008** .013 (executive function, working AB x C -0.19 -0.58 0.20 0.20 -.04 .170 .002 memory) A x B 0.58 -0.12 1.28 0.35 .08 .051t .006 Paired-Associates Learning ABC 0.51 0.26 0.76 0.13 .21 <.001*** .043 (long-term visual memory) AB x C 0.05 -0.12 0.23 0.09 .03 .272 .001 A x B -0.03 -0.34 0.28 0.16 -.01 .426 .000 Story Recall ABC 0.55 0.19 0.92 0.18 .13 .002** .017 (long-term verbal memory) AB x C -0.04 -0.29 0.22 0.13 -.01 .389 .000 A x B 0.11 -0.35 0.56 0.23 .02 .319 .000 Digit Forward ABC 0.00 -0.23 0.23 0.12 .00 .493 .000 (short-term verbal memory) AB x C 0.00 -0.16 0.16 0.08 .00 .487 .000 A x B -0.11 -0.39 0.18 0.14 -.04 .225 .002 Age Concentration Test A ABC 0.03 0.02 0.05 0.01 .20 <.001*** .040 (concentration, attention) AB x C 0.00 -0.01 0.01 0.01 -.02 .366 .000 A x B 0.00 -0.02 0.02 0.01 -.01 .407 .000 Age Concentration Test B ABC 0.01 0.00 0.03 0.01 .08 .036* .007 (concentration, attention) AB x C 0.00 -0.01 0.01 0.01 -.01 .433 .000 A x B 0.01 -0.01 0.03 0.01 .05 .140 .002 Digit Symbol Substitution ABC 2.20 1.62 2.77 0.29 .18 <.001*** .033 (information processing speed) AB x C 0.16 -0.24 0.56 0.20 .02 .216 .000 A x B -0.44 -1.15 0.27 0.36 -.03 .113 -.001

Notes: ABC, linear global time effect; AB×C, linear time×intervention interaction DANCE/MEMORY versus PHYS; A×B, linear time×intervention interaction DANCE versus MEMORY. A, DANCE; B, MEMORY; C, PHYS. Bold values indicate significance or trend, * p<.05, ** p<.01, *** p<.001, t p<.10 trend. Abbreviations: DANCE, virtual reality video game dancing; MEMORY, treadmill walking with simultaneous verbal memory training; PHYS, treadmill walking; CI, confidence interval; SE, standard error of the mean.

68 TABLE 3 | Repeated measures ANOVA from 6-month test to follow-up test, N=47

Dependent variable Effect F(2, 44) p, r (cognitive domain) two-tailed

Trail Making Part A Time 0.104 .748 .05 (information processing speed) Time×intervention 0.664 .520 .12 Trail Making Part B Time 6.444 .015* .36 (executive function, shifting) Time×intervention 0.372 .691 .09 Executive Control Time 0.110 .741 .05 (executive function, working memory) Time×intervention 1.086 .346 .16 Paired-Associates Learning Time 1.133 .293 .16 (long-term visual memory) Time×intervention 0.216 .807 .07

Notes: * p<.05. Bold values indicate significance or trend. Abbreviation: ANOVA, analysis of variance.

4.3.2 Training enjoyment

FIGURE 6 | Comparison of training enjoyment in the three interventions.

Notes: No group differences were shown for overall training enjoyment (PACES), strength, and balance training (all p>.05). The two cognitive–motor training components (video game dancing and treadmill memory) tended to be enjoyed more than treadmill walking (trend p=.069, one- tailed). Scores system is from one to seven points (least to maximal enjoyment), tp<.10 trend, error bars indicate ± standard error of the mean. Abbreviations: PACES, Physical Activity Enjoyment Scale; TE, training enjoyment; DANCE, virtual reality video game dancing; MEMORY, treadmill walking with simultaneous verbal memory training; PHYS, treadmill walking.

69 Training enjoyment was measured at 6-month test in the 71 participants who completed the 6-month training. Results are presented in Figure 6. One-way ANOVA and planned contrasts did not show significant group differences for overall training enjoyment (PACES, p=.606), training enjoyment of balance (p=.979), and strength training (p=.972). A trend to a significant contrast was found between the enjoyment of the two cognitive–motor training components (video game dancing and treadmill memory training) and the treadmill walking (t(68) = 1.503, trend p=.069 (one-tailed for directional hypothesis), r=.18). Participants seemed to favor the two cognitive–motor components over the treadmill walking.

4.4 DISCUSSION

This study aimed to compare two simultaneous cognitive–motor training interventions with an exclusively physical multicomponent training program and to evaluate effects on cognition. The study comprised a 6-month training intervention and a 1-year follow-up. The two main findings were first, that the cognitive–motor programs were partially advantageous to boost performance in two measures of executive function (switching attention and working memory), thereby video game dancing resulted in transfer to an untrained cognitive domain (working memory); and second, that cognitive performance, including executive functions, long-term visual memory (episodic memory), and processing speed, was maintained until 1-year follow-up. These findings are important since executive functions, episodic memory, and processing speed are particularly affected by aging-related decline (Deary et al., 2009). Therefore, we suggest that simultaneous cognitive–motor training should be integrated in training programs aiming to improve cognition in the elderly.

4.4.1 Does simultaneous cognitive–motor training boost cognitive performance?

We found one indication in our results that supported the hypothesis that the simultaneous cognitive–motor programs (DANCE, MEMORY) had advantages over an exclusively physical intervention (PHYS) in terms of

70 cognitive adaptations. Both cognitive–motor interventions showed larger improvements in the TMT-B compared to PHYS within the initial 3-month training period (Figure 4B). This result showed a trend to statistical significance but seems worth mentioning due to the small to moderate effect size. The TMT-B reflects the ability of shifting attention, which is a dimension of executive function. This ability might have been trained through the simultaneous performance of cognitive and physical activities in DANCE and MEMORY in the way that attention had to be shifted continuously between the two activities. The dual-task situation in the treadmill memory training possibly had some impact on cognitive shifting ability because treadmill walking itself required a certain amount of attention from the elderly participants to be executed safely. A similar result related to cognitive shifting was demonstrated in an investigation that compared effects on cognition after contemporary dancing, Tai Chi, or balance training (Coubard et al., 2011). Thereby, only contemporary dancing, which can be regarded as a modality of simultaneous cognitive–motor activity, had an effect on cognition and particularly on switching attention as assessed with the Rule Shift Cards Sorting Test (Wil- son et al., 1998). Interestingly, a recent extensive investigation with 182 participants by van het Reve and de Bruin (2014) did not show this additional effect on shifting attention after 3 months of sequential cognitive and physical training compared to exclusively physical training (strength and balance exercises). This observation supports the benefits from simultaneously performed cognitive–motor training over sequential cognitive and physical training programs. In our study, participants of PHYS had about 2 years less school education compared to the other groups despite randomization. How- ever, we would argue that this difference had no influence on the development of cognitive outcomes, since baseline cognitive measures and MMSE scores were not statistically different. Furthermore, adaptation patterns were similar in PHYS compared to the other groups in several cognitive outcomes. We conclude that additional cognitive functions, particularly switching attention, are promoted by the dual-task situation in simultaneous cognitive–motor programs and further research is warranted to substantiate or refute this assumption.

The expected differential adaptation patterns from DANCE versus MEMORY were confirmed in the results of the Executive Control Task (Figure 4C), which reflects working memory as another dimension of the executive

71 functions: different time courses of adaptation from pretest, to 3-month, and 6-month tests for DANCE versus MEMORY were found, with superior performance in DANCE after 6-month training. This result was supported by a significant time intervention interaction with small to moderate effect size in the specific analysis of the second 3-month training period. Within this period DANCE improved, whereas MEMORY deteriorated and PHYS remained unchanged. Our finding confirms the previously noted importance of applying longer training durations (6 months or longer) to assess cognitive adaptation patterns and to achieve larger training gains from physical interventions (Colcombe and Kramer, 2003; Erickson et al., 2012; Cai et al., 2014). More importantly, the result represents an adaptation from video game dancing in an untrained cognitive domain (working memory) or a so-called transfer effect. Previous studies on combined cognitive–motor training failed to produce cognitive transfer effects but reported training specific adaptations: for instance, Theill et al (2013) demonstrated performance gains in the Executive Control Task after simultaneous cognitive–motor and single cognitive training, which both contained specific working memory exercises. Similarly, van het Reve and de Bruin (2014) reported a training specific adaptation after a 3-month computerized divided attention training, which was contained in a sequential cognitive–motor program. In summary, the present study provides first indications that simultaneous cognitive–motor training boosts particular executive functions (shifting attention and working memory) depending on the duration of the intervention, and that the video game dancing leads to cognitive transfer in working memory. However, further investigations are necessary to substantiate this finding. Improvements of executive functions in seniors are clinically important because they are critical for the regulation of gait, are related to fall risk (Mirelman et al., 2012), and are prone to aging- related decline in general (Deary et al., 2009).

4.4.2 Are cognitive training effects maintained after cessation of the training intervention?

Training gains were preserved in our study in three out of four follow-up tests over 1 year without any further training intervention being applied. Surpris- ingly, performance kept increasing in the TMT-B from 6-month test to follow- up test in all groups, which may reflect a delayed response to the intervention.

72 We did not systematically assess the amount of training that participants might have taken up individually after cessation of the intervention. There- fore, we cannot estimate a possible effect of additional individual training on cognitive measures at follow-up. Maintenance of cognitive performance was reported previously after different kinds of training interventions. For instance, a 1-year follow-up cognitive assessment after 6 months of either in- door cycling or stretching and coordination training demonstrated maintenance of selective attention (d2 test) and episodic memory learning (Hotting et al., 2012). However, only the subgroup with a high level of cardiovascular fitness, measured at follow-up, was able to preserve performance in episodic memory recognition, while the low-fit subgroup deteriorated from post intervention to 1-year follow-up. This finding indicates the importance of cardiovascular fitness as a mediator to maintain certain cognitive abilities. A 5-year follow-up study also reported sustained effects in a composite “cognitive function” score after sequential cognitive–motor training (effect size d+=.75) (Os- wald et al., 2006), but no performance maintenance was found after exclusively physical training comprising balance and coordination exercises. This finding stands in contradiction to our own result and may support the neces- sity of multicomponent physical programs containing aerobic endurance and muscular strength exercises for long-term performance maintenance of cognition. Finally, a meta-analysis including seven studies with exclusively cognitive interventions found persistent cognitive enhancements over different follow-up periods from 3 months up to 5 years (Valenzuela and Sachdev, 2009). Considering the existing literature and our own results, it may be summarized that long-term cognitive performance maintenance can be evident after both, exclusively cognitive or multicomponent physical training containing aerobic endurance and strength exercises, as well as after sequential or simultaneous cognitive–motor interventions. However, which type of intervention might be superior in this respect needs further investigation.

4.4.3 Can simultaneous cognitive–motor and exclusively physical multicomponent training programs elicit broad cognitive adaptations?

A significant global linear time effect in the regression analyses of eight out of nine cognitive tests was found in this study for all three interventions taken

73 together. Therefore, our results might extend the findings from the majority of interventions and meta-analyses with exclusively physical training demonstrating improvements in different cognitive dimensions (Colcombe and Kra- mer, 2003; Cassilhas et al., 2007; Liu-Ambrose and Donaldson, 2009; Wil- liamson et al., 2009; Liu-Ambrose et al., 2010; Smith et al., 2010; Voelcker- Rehage et al., 2011; Chang et al., 2012; Roig et al., 2013). However, due to the lack of a passive control group, to account for learning effects from repeated measurements, we are not able to display training effects exclusively. None- theless, based on similar results from the literature, the long intervals of 3 months between test sessions and the application of parallel versions in some of the cognitive tests, we assume that performance improvements can at least partly be accounted for as training effects.

Several neurobiological and physiological mechanisms have been suggested to link physical training with benefits on cognitive performance: these are increased neurogenesis and synaptogenesis in the cortical structure, promo- tion of cerebral metabolism, alterations of neurotransmitter and neurotrophic factor levels, availability of cerebral oxygen and glucose, and reduced oxida- tive stress (Marmeleira, 2013). In particular, the cardiovascular fitness hypothesis has been promoted to relate aerobic fitness and cognition. Thereby, aerobic training was found to affect certain mechanisms, such as cerebral blood flow, brain-derived neurotrophic factor, and cerebral structure, which are also associated with increased cognitive performance (Marmeleira, 2013). Nevertheless, a meta-analysis by Etnier et al. (2006) failed to support the relation between aerobic fitness and cognition. The authors argued that aerobic fitness itself might not be sensitive enough to indicate cognitive adaptations from aerobic exercise training, whereby the underlying adaptations of aerobic training might be more sensitive (Etnier et al., 2006). For instance, the cerebral circulation hypothesis relies on studies that have found elevated oxygen and glucose transport to the brain, leading to improved cognitive performance (Chodzko-Zajko et al., 1992). Furthermore, the neurotrophic stimu- lation hypothesis suggests that training-induced enhancement of brain-derived neurotrophic factor stimulates neurogenesis and thereby positively affects learning and mental performance (Marmeleira, 2013). Finally, the neu- roadrenergic hypothesis proposed that cardiovascular training promoted neurotransmitter availability, such as noradrenaline, adrenaline, and serotonin, which are thought to be related to memory storage and retrieval (Zornetzer,

74 1985; Etnier et al., 2006). As indicated earlier, some recent studies also demonstrated that strength (Liu-Ambrose et al., 2012) and coordination (Voelcker-Rehage et al., 2011) training-induced changes in hemodynamic brain activity or elevated activation of certain brain networks, respectively, which were associated with improved cognition.

The importance of the physical part of a simultaneous cognitive–motor training intervention in older adults was supported by Theill et al. (2013) who reported improvements in long-term visual memory (Paired-Associates Task) after simultaneous treadmill walking and memory training, but not after exclusively cognitive training. Additionally, two meta-analytic studies pointed out that exclusively cognitive training programs increased performance only on related or training-specific tasks and no cognitive transfer effects were evident (Noack et al., 2009; Papp et al., 2009). This was also confirmed by the recent review from Oei and Patterson (2014) who investigated transfer effects in video game training studies. However, some cognitive training studies reported broad improvements from cognitive training, particularly from extended but not from strategy training approaches (Zelinski, 2009). Extended practice refers to the training of basic cognitive abilities, such as choice response time or phoneme span, which are used in different cognitive activities. Additionally, a recent meta-analysis by Hindin and Zelinski (2012) found similar effect sizes in extended cognitive training compared to aerobic exercise training, although different neurophysiological mechanisms would likely have led to these effects. Nevertheless, it appears that physical and combined cognitive–motor training interventions may be more beneficial than exclusively cognitive training interventions for older adults to enhance a broad range of cognitive abilities. Such training programs should therefore be implemented in the clinical prevention of cognitive impairments, which are widely prevalent in older adults (Scafato et al., 2010).

4.4.4 Strengths and limitations

Methodological strengths of this study were the comparably large number of participants, the long training period with follow-up measurements, and the broad cognitive testing battery to assess several dimensions of cognition. Some limitations have to be considered as well. First, the specific effects of the two simultaneous training modalities could not be identified exactly

75 because of the combination with the multiple physical components (strength and balance training). However, this was not the focus of this study, since we explicitly aimed at evaluating effects from different multicomponent programs. Second, the conclusions and recommendations from this study are limited to physically and mentally healthy seniors, because following the selection criteria such participants were recruited. Training effects might have been even larger in a population of lower physical and mental status. This assumption is based on the exercise training principle “Initial Values” stating that improvement in the outcome of interest will be greatest in those with lower initial values (Ammann et al., 2014). Those with lowest levels of fitness theoretically have greatest room for improvement. It seems, therefore, important and warranted to repeat the study design in a more vulnerable population exhibiting impairments in fitness and/or cognitive domains. Further, as mentioned earlier, we did not include a passive control group in the design of the study, which means that we could not exactly differentiate between training effects and learning effects from repeated testing. However, this was not the main focus of the present study. Although participants were blinded to the expected study outcome, blinding of the investigators was not possible since they also supervised and conducted training and testing sessions. This is an additional limitation to this study.

4.5 CONCLUSIONS

We demonstrated that multicomponent simultaneous cognitive–motor training programs have the potential to boost particular executive functions (including shifting attention and working memory) in healthy older adults compared to an exclusively physical multicomponent program. Importantly, performance levels in executive functions, long-term visual memory (episodic memory), and processing speed were maintained over 1 year after all three programs. The novel training concepts of simultaneous cognitive–motor activity tended to be enjoyed more by seniors than traditional training and led to training specific as well as to transfer adaptations in cognition. Therefore, we recommend multicomponent simultaneous cognitive–motor training programs to enhance particular executive functions in older adults. Such programs may potentially counteract the large prevalence of cognitive

76 impairments and decline in the elderly, inherently leading to more independence and a better quality of life. Future studies should also investigate the neurological background of cognitive behavioral performance enhancements to shed light on the interconnection between plasticity of cognition, brain function, and brain structure.

Acknowledgments This work was supported by the Zürcher Kantonalbank within the framework of sponsoring of movement sciences, sports, and nutritional sciences at ETH Zurich. Zürcher Kantonalbank had no influence on the study design and the analyses presented in this paper, had no access to the data, and did not contribute to this manuscript in any way. The authors would like to thank PD Dr. med. Thomas Münzer, chief physician, and the management of Geriat- rische Klinik St.Gallen, Switzerland, for supporting the study and providing room for training and data acquisition. Furthermore, we thank our postgraduate students, Stefan Holenstein, Fabienne Hüppin, Manuela Kobelt, Alex- andra Schättin, and Sara Tomovic for training instruction and helping with data acquisition. We highly appreciate the support of the team of physiotherapists at Geriatrische Klinik St.Gallen, and particularly the support from Carmen Fürer and Carole Scheidegger who initially put forward the idea of performing the study at their institution. Last but not the least, we would like to thank all the participants for their enthusiasm, kindness, and patience during our extensive training and testing interventions.

Author contributions PE, study preparation and conception, participants’ recruitment, data acquisition, statistical analysis, data interpretation, drafting manuscript; VS, study conception, conception of cognitive test battery, data interpretation, revising manuscript; MA, study preparation, training instruction, data acquisition, data interpretation, revising manuscript; NT, study conception, conception of serial position training, supporting statistical analysis, data interpretation, revising manuscript; EDB, study conception, data interpretation, revising manuscript. All authors read and approved the final manuscript.

77 Disclosure The authors report no conflicts of interest in this work.

Copyright notice This work was initially published by Dove Medical Press Limited and licensed under Creative Commons Attribution 3.0 Non-Commercial Unported (CC BY- NC 3.0) http://creativecommons.org/licenses/by-nc/3.0/. Non-commercial uses of the work are permitted without any further permission from Dove Medical Press Limited, provided the work is properly attributed. The following changes were applied to the original manuscript: the term “cognitive–physical” was replaced by the term “cognitive–motor” to use consistent terminology throughout the manuscript of this doctoral thesis.

78 5

Gait & physical performance adaptations (study 2b)

79 MULTICOMPONENT PHYSICAL EXERCISE WITH SIM- ULTANEOUS COGNITIVE TRAINING TO ENHANCE DUAL-TASK WALKING OF OLDER ADULTS: A SEC- ONDARY ANALYSIS OF A 6-MONTH RANDOMIZED CONTROLLED TRIAL WITH 1-YEAR FOLLOW-UP

Patrick Eggenberger1, Nathan Theill2,3, Stefan Holenstein1, Vera Schuma- cher3,4, and Eling D. de Bruin1

1Institute of Human Movement Sciences and Sport, Department of Health Sciences and Technology, ETH Zurich, 2Division of Psychiatry Research, Uni- versity of Zurich, Schlieren, 3Center for Gerontology, University of Zurich, 4Department of Gerontopsychology and Gerontology, 5University Research Priority Program “Dynamics of Healthy Aging”, University of Zurich, Swit- zerland

Published in Clinical Interventions in Aging, 28 October 2015: 10, 1711–1732. http://dx.doi.org/10.2147/CIA.S91997

80 ABSTRACT

Background: About one-third of people older than 65 years fall at least once a year. Physical exercise has been previously demonstrated to improve gait, enhance physical fitness, and prevent falls. Nonetheless, the addition of cognitive training components may potentially increase these effects, since cognitive impairment is related to gait irregularities and fall risk. We hypothesized that simultaneous cognitive–motor training would lead to greater improvements in dual-task (DT) gait compared to exclusively physical training.

Methods: Elderly persons older than 70 years and without cognitive impairment were randomly assigned to the following groups: 1) virtual reality video game dancing (DANCE), 2) treadmill walking with simultaneous verbal memory training (MEMORY), or 3) treadmill walking (PHYS). Each program was complemented with strength and balance exercises. Two 1-hour training sessions per week over 6 months were applied. Gait variables, functional fitness (Short Physical Performance Battery, 6-minute walk), and fall frequencies were assessed at baseline, after 3 months and 6 months, and at 1-year follow-up. Multiple regression analyses with planned comparisons were carried out.

Results: Eighty-nine participants were randomized to three groups initially; 71 completed the training and 47 were available at 1-year follow-up. DANCE/MEMORY showed a significant advantage compared to PHYS in DT costs of step time variability at fast walking (p=.044). Training-specific gait adaptations were found on comparing DANCE and MEMORY: DANCE reduced step time at fast walking (p=.007) and MEMORY reduced gait variability in DT and DT costs at preferred walking speed (both trend p=.062). Global linear time effects showed improved gait (p<.05), functional fitness (p<.05), and reduced fall frequency (−77%, p<.001). Only single-task fast walking, gait variability at preferred walking speed, and Short Physical Performance Battery were reduced at follow-up (all p<.05 or trend).

Conclusions: Long-term multicomponent cognitive–motor and exclusively physical training programs demonstrated similar potential to counteract age-related decline in physical functioning.

Keywords: elderly, dance video game, gait, falls, functional fitness, detraining, sex

81 5.1 INTRODUCTION

Falls are a significant health problem in the elderly population. About one- third of people aged 65 years or more fall at least once a year, and incidence rises with aging (Tinetti et al., 1988; Hausdorff et al., 2001; Gill et al., 2005). Physical injuries from falls in elderly persons often lead to disability and loss of independence and may increase the risk of premature death. Psychological consequences include depression and fear of falling, which are factors that also elevate the risk of falls in the future (Stel et al., 2004; Tinetti and Kumar, 2010). Furthermore, considerable health care costs result from falls, sum- ming up to US$23.3 billion annually in the USA and US$1.6 billion in the UK. The costs per fall associated with injury range from US$3’476 to US$10’749 and rise up to US$26’483 when hospitalization is necessary (Davis et al., 2010). Several risk factors exist for falls, comprising impaired balance and gait, multimedication, a history of previous falls, advancing age, female sex, visual impairments, cognitive decline, and environmental factors (Am- brose et al., 2013). Stenhagen et al. (2013) found three main components that predict falls: reduced mobility (odds ratio (OR) = 2.12), heart dysfunction (OR = 1.66), and functional impairment (OR = 1.38), whereby each component is related to physical fitness. Hence, many programs try to improve physical fitness and gait with the aim of preventing falls in older adults (Panel on Prevention of Falls in Older Persons and British Geriatrics, 2011).

Physical exercise programs were shown to have similar effects on fall prevention as interventions aiming at multiple fall risk factors (Campbell and Rob- ertson, 2007; Gillespie et al., 2012). A recent meta-analysis among elderly persons living in the community reported significantly reduced fall rate (rate ratio = 0.71) and risk of falling (relative risk = 0.85) after physical multicomponent group exercise (Gillespie et al., 2012). It was suggested that multicomponent exercise programs should particularly include strength, balance, gait, and coordination training and should last >12 weeks (one to three sessions per week) to effectively reduce falls (Gardner et al., 2000; Chang et al., 2004; Sherrington et al., 2011; Gillespie et al., 2012). However, a meta-analysis by Muir et al. (2012) reported that fall risk in community-dwelling older adults was elevated considerably when global measures of cognition were impaired (summary estimate of OR = 2.13). Therefore, the efficacy of physical exercise programs in preventing falls may be raised further by adding cognitive

82 training components. In fact, various cognitive domains, including attention, executive function, information processing, and reaction time, are related to balance, gait, and fall risk (Holtzer et al., 2007; Herman et al., 2010; Muir et al., 2012). Thereby, dual-task (DT) ability, as a part of the executive functions, appears to be especially important.

Regardless of these findings, cognitive aspects are often neglected in fall prevention programs, and to date, few studies have applied multicomponent exercise programs with additional cognitive training to improve DT ability in older persons. Van het Reve and de Bruin (2014), for instance, reported improvements in DT walking after 12 weeks of balance and strength exercise with sequential computerized cognitive training. Other studies applied cognitive and physical training simultaneously in order to target DT ability more specifically (de Bruin et al., 2011; Pichierri et al., 2012b; Forte et al., 2013; Theill et al., 2013). Although three of these studies successfully improved measures of DT walking (de Bruin et al., 2011; Pichierri et al., 2012b; Theill et al., 2013), none assessed long-term retention of gait performance and effects on fall frequency. Nevertheless, simultaneous cognitive–motor DT exercise appears to be promising in complementing multicomponent physical exercise interventions to further enhance DT walking and prevent falls in the elderly persons. However, clearly more extensive research is necessary to substantiate this assumption.

This study aims to compare two variations of multicomponent simultaneous cognitive–motor training with an exclusively physical program and to evaluate the effects of these programs primarily on DT gait performance in healthy elderly persons, whereas fall frequency and functional fitness represent secondary outcomes. We hypothesize, first, that simultaneous cognitive–motor training may create additional enhancements on DT gait variables, and second, that the two cognitive–motor training variations may lead to differential gait adaptations. Based on previous findings, reported earlier, we expect improvements in gait, functional fitness, and a reduction in falls after all the three programs. Furthermore, we aimed to investigate performance retention in gait and functional fitness 1 year after the training interventions. This study is a secondary analysis of a 6-month randomized controlled trial with a 1-year follow-up, where we initially investigated training effects on cognitive performance (Eggenberger et al., 2015a).

83 5.2 MATERIALS AND METHODS

5.2.1 Study design and participants

This study was a randomized controlled trial, including three groups that underwent parallel 6-month training interventions and a 1-year nonintervention follow-up. Assessments were performed four times: pretraining, after 3 months, 6 months of training (posttraining), and at 1-year follow-up. Data collection and training were performed at Geriatrische Klinik, St.Gallen, Switzerland. The study protocol was approved through the local ethics committee of the canton St.Gallen, Switzerland (study number: EKSG 12/092) and was registered at Current Controlled Trials ISRCTN70130279. No changes were made to the planned methods after trial commencement. Our reports adhere to the CONSORT (Consolidated Standards of Reporting Tri- als) 2010 guidelines (Moher et al., 2010).

Participants were recruited through a newspaper article, a local organization for the elderly (Pro Senectute St.Gallen), residence facilities for the elderly, primary care physicians, and via the Web sites of the city’s geriatric hospital and the department of sports of the canton St.Gallen. Interested persons were invited to an information event. For eligibility, participants had to be older than 70 years, live independently or at residence facilities for the elderly, and sign informed consent. Participants had to be able to walk at least 20 m, with or without walking aids, for gait analysis. Residents of retirement homes classified 0, 1, or 2 within the Swiss classification system for health care requirements (BESA levels, German abbreviation for: Bewohner-Einstufungs- und Abrechnungssystem) could enroll in the study. Level 0 means the person does not need care or treatment and levels 1–2 mean that the person only needs little care or treatment. Seniors diagnosed with Alzheimer’s disease, dementia, recent head injury, or a score <22 points (MacKenzie et al., 1996) on the Mini-Mental State Examination (Folstein et al., 1975), which indicates cognitive impairment, were excluded. Judgment by their primary care physician was required in the case of acute or instable chronic diseases (e.g. stroke and diabetes) and rapidly progressing or terminal illnesses before accepting a person for participation.

84 A priori power analysis (G*Power 3.1.3 Software, (Faul et al., 2007) revealed that a total of 75 participants were needed to achieve 80% power for a three- group pretest, 3-month and 6-month test design (25 participants per group). The α-level was set at 0.05 and the effect size f was set at 0.3. The randomization scheme was generated with the Web site Randomization.com (Ran- domization, 2017), applying block randomization to achieve three groups with a ratio of 1:1:1. Participants were blinded to the expected study outcome, while blinding of the investigators was not possible since they supervised and conducted training and testing sessions.

5.2.2 Training programs

Two 1-hour training sessions per week were performed in groups of five to six participants, under instruction of two trained postgraduate students. At least 1 day was implemented between sessions for recovery. Training programs were based on current recommendations for physical fitness and fall prevention for the elderly persons (Chodzko-Zajko et al., 2009; Sherrington et al., 2011; Granacher et al., 2012). The three multicomponent programs consisted of 20 minutes aerobic endurance training (video game dancing [DANCE], treadmill memory training [MEMORY], or treadmill walking [PHYS]) and complementary strength and balance exercises (20 minutes each). The training components are described in detail in the following sections, and an overview is provided in Table 1. The exercise training principles of progression and overload were applied for every training component (Ammann et al., 2014) and were adapted to each participant’s abilities in terms of treadmill speed and inclination, step frequency in DANCE, or number of sets and repetitions, in order to achieve moderate-to-vigorous exercise intensity. This level of intensity corresponds to a subjective rate of perceived exertion of 5–7 points on the 10-point Borg scale as recommended by the American College of Sports Medicine position stand on exercise with older adults (Chodzko- Zajko et al., 2009). In total, 52 sessions were conducted within 6 months (26 weeks), with some participants missing certain sessions due to personal reasons. Sessions 25–32 (4 weeks) were conducted individually according to a home exercise plan, due to the Christmas holidays and 3-month test sessions. The home exercise plan comprised the same strength and balance exercises

85 as instructed during normal training sessions, but no DANCE and MEMORY training. Compliance to the home exercise plan was assessed with a training diary.

TABLE 1 | Description of training contents according to the FITT principles

Frequency Intensity Time Type

Two sessions Moderate to 20 min Aerobic endurance with cognitive–motor coordination aspect of 1 hour/wk vigorous (RPE (DANCE, MEMORY) or without (PHYS): (26 weeks) 5–7 points on DANCE: virtual reality video game dancing. 2–3 min/ game/ ten-point Borg- song, 1–2 min rest periods only if required scale) or MEMORY: continuous flat or inclined treadmill walking with simultaneous verbal memory training or PHYS: continuous flat or inclined treadmill walking (or running) 20 min Muscular strength (complement to each intervention): Examples of lower body exercises: seated leg extensions with 2 kg ankle weights, chair rises, split leg squats, calf raises (all with or without 5–10 kg weight vest), standing toe raises; two exercises per session Examples of upper-body exercises: standing arm row, biceps curls (both with resistive rubber bands), standing wall push- ups. knee push-ups; 1–2 exercises per session Examples of trunk stabilization exercises: incline seated single- leg raises, crunches, front plank; 1–2 exercises per session One to three sets with eight to 12 repetitions, progressing from slow to fast movement speed, ∼1 min rest between sets 20 min Balance (complement to each intervention): Examples: tandem stand, two-leg stand on foam pad, walking over a skipping rope on the floor, single-leg stand on air pad, single-leg stand with eyes closed Two to four sets of four to five different exercises per session, 20–60 s per exercise, ∼30– 60 s rest

Notes: Training programs were created based on current recommendations for physical fitness and fall prevention for the elderly persons. Data from Nelson et al. (2007); Sherrington et al. (2011); Granacher et al. (2012). Abbreviations: DANCE, virtual reality video game dancing; FITT, frequency, intensity, time, and type of exercise; MEMORY, treadmill walking with simultaneous verbal memory training; PHYS, treadmill walking; RPE, rate of perceived exertion.

86 Video game dancing The program DANCE included virtual reality video game dancing as a simultaneous cognitive–motor training (Figure 1A). This training component combines an attention-demanding cognitive action with a simultaneous motor coordination aspect. We used two Impact Dance Platforms (Positive Gam- ing BV, Haarlem, the Netherlands) and created various levels of difficulty in step patterns and frequency with the StepMania Software (StepMania, 2016). Several styles of music were selected to add variety and meet preferences of participants. Participants stood on the one-by-one meter platform, which contained four pressure sensitive areas to detect steps forward, backward, to the left, and to the right, respectively. Stepping sequences were cued with arrows appearing on a large screen and had to be performed exactly when an arrow reached a highlighted area on the screen in order to achieve best scores in the game. Participants were instructed to hold on to ropes for security reasons.

FIGURE 1 | Simultaneous cognitive–motor training components.

87 Treadmill memory training The program MEMORY comprised treadmill walking with verbal memory exercise as a simultaneous cognitive–motor training (Figure 1B). Verbal memory training consisted of a computer-based serial position training that was presented on a computer screen in front of the treadmill, with a standard computer mouse as an input device. E-Prime 2.0 Professional software (Psy- chology Software Tools, Pittsburgh, PA, USA) was used to program the training. Participants were asked to memorize the correct sequence of 3–20 words lighting up one after the other for 3 seconds on the computer screen. There- after, a distraction task followed, where participants had to de ne whether three presented words had a meaning or not. Then, the initially memorized words were presented again, either in the same or a different sequence, and participants had to decide whether the sequence remained the same or not, by pressing the mouse button. The initial level for this training was set at a sequence of three words and was extended by one word as soon as the participants reached 80% of correct answers within the level.

Treadmill walking The program PHYS included aerobic treadmill walking without any additional cognitive task and acted as a reference group with exclusively physical training components. Participants were instructed to walk or run at a constant pace.

Complementary strength and balance exercises In addition to one of the three different aerobic training components described previously, muscular strength and balance exercises complemented each program (Figure 2). Four to five strength exercises for lower and upper extrem- ities and trunk stabilization were performed using own body weight, resistive rubber bands, and weight vests of maximum of 10 kg. Balance training consisted of different exercises including two- and single-leg stance variations, either on the floor or on various types of instable surfaces (e.g., foam and air pads, ropes, etc.) (de Bruin and Murer, 2007).

FIGURE 2 | Examples of complementary balance (A) and strength (B) exercises.

Notes: The participant in (A) steps from one object to the next, trying to maintain balance under assistance of an instructor (objects are soft rubber “stones”, skipping ropes, and balance pads). (B) shows one participant performing chair rises (squats) and another participant holding the “plank” position for global trunk stability, which was an exercise for the elderly with higher fitness level.

5.2.3 Measurements

Gait analysis (primary outcome) Temporal and spatial gait variables were assessed using the GAITRite electronic walkway system (CIR Systems, Havertown, PA, USA) with the Plati- num Version 4.0 software. The validity and reliability of the GAITRite system have been well established (Bilney et al., 2003; van Uden and Besser, 2004; Webster et al., 2005). Walking was initiated 2 m before the 7.3 m active area of the walkway and ended 2 m thereafter to allow for steady-state gait assessment. The testing protocol comprised four conditions, which were single-task (ST) and DT walking at the individual’s preferred and fast speed. Each test condition was repeated three times, and the mean value was used for further analysis. Trials were repeated when a participant stopped walking or performing the cognitive task. The cognitive–motor DT condition was adjusted to the participant’s cognitive abilities and included walking while counting backward in steps of seven or three from a random number between 200 and

89 250 or enumerating objects (e.g., flowers, country names, or first names). The purpose of this procedure was to quantify cognitive–motor interference while walking (Al-Yahya et al., 2011). Participants were instructed not to prioritize either task and were allowed to use assistive walking devices. Relative dual- task costs (DTCs) of walking, as the percentage of relative loss to the ST walking performance, were calculated according to the formula DTC = 100 × (dual- task score − single-task score) / single-task score.

Secondary outcomes Fall frequency was assessed retrospectively for the 6-month period prior to the intervention, after 3 months and 6 months of training, and 6 months and 12 months after the intervention. Participants who were not available for the 6-month and 1-year follow-up events were interviewed via telephone at the respective time points.

The Short Physical Performance Battery (SPPB) was used to assess lower extremity functioning with a balance test, a 3-m-walk test, and a 5-chair-rises test (Guralnik et al., 1994). We extended the standard SPPB balance test with two additional levels to increase difficulty and to avoid possible ceiling effects. The first additional level comprised a 20-second single-leg stance. One point was added when 10 seconds were reached and another point when 20 seconds were completed. In the second additional level, a single-leg stance with eyes closed was required to be maintained as long as possible. One point was added to the balance score for every 5 seconds the position was held.

Functional aerobic endurance performance was measured with the 6-minute walk test (6-MWT) following the guidelines of the American Thoracic Society (Laboratories, 2002). Participants were asked to walk as far as possible on a 30-m walking course within 6 minutes. This test was not repeated at follow- up to reduce the test time for participants. Maximum walking distance and rate of perceived exertion were recorded immediately after completion of the test. Ratings of perceived exertion were assessed by means of the ten-level Category Ratio Scale (adapted Borg scale) (Neely et al., 1992).

The Falls Efficacy Scale International (FES-I) was applied as a measure of fear of falling (Yardley et al., 2005), while symptoms of depression were recorded using the German version of the Geriatric Depression Scale (GDS) (Gauggel and Birkner, 1999).

90 5.2.4 Statistical analyses

Group differences in the baseline demographic and performance data were compared with one-way analysis of variance (ANOVA). Multiple regression analysis with planned comparisons, including orthogonal contrast and polynomial trend coding, was applied to investigate the training effects on gait variables and functional fitness for the 6-month training. We produced contrast-coding variables based on the hypotheses. The first contrast was set to compare the two combined cognitive–motor training groups with PHYS. The second contrast compared the two cognitive–motor training groups (DANCE vs MEMORY). Effect code variables were produced for each group’s individuals to account for subject effects. Repeated-measures ANOVA were used to assess differences between the 6-month test and the 1-year follow-up, as well as for the supplementary analysis of sex differences. Missing values from participants who completed the full 6-month trial, but missed single-test items due to health constraints or social obligations, were replaced by the group mean value at the respective time point of measurement. Statistical calculations were performed with IBM SPSS Statistics software for Macintosh, Ver- sion 22.0 (IBM Corporation, Armonk, NY, USA) with a significance level of �=.05. Effect sizes, represented as R2-change in the multiple regression analysis, were considered as small for R2-change=.01, medium for R2-change=.06, and large for R2-change=.14 and above. Effect size r from ANOVA was defined as small at r=.10, medium at r=.30, and large at r=.50 and above (Cohen, 1988).

5.3 RESULTS

Of the 89 participants initially enrolled, 71 participants completed the 6- month training intervention (20.2% attrition) and were included in the analysis of the outcomes derived at pretest, 3-month, and 6-month tests. Time points and reasons for dropouts are presented in Figure 3.

FIGURE 3 | Trial design and participants’ flow.

Notes: Participants were randomly assigned to one of two multicomponent simultaneous cognitive– motor training groups (DANCE and MEMORY) or an exclusively physical multicomponent training group (PHYS) and were trained over 6 months twice weekly for 1 hour. Gait variables, functional fitness, and fall frequency were assessed at pretest, 3-month and 6-month test, and at 1-year follow- up (except 6-MWT not repeated at follow-up). Fall frequency was additionally assessed 6 months after training. Abbreviations: DANCE, virtual reality video game dancing; MEMORY, treadmill walking with simultaneous verbal memory training; PHYS, treadmill walking; 6-MWT, 6-minute walk test.

Dropouts were equally distributed between groups, and therefore, the final analyses were performed only in those individuals who completed the 6- month intervention. At 1-year follow-up, 47 participants were available for gait and SPPB measurements, while falls data of 66 persons were recorded

92 and included in the analysis. The following numbers of missing values were replaced by the group mean value: 33 gait, one 3-m-walk, fourteen 6-MWT, and one GDS and FES-I. Participants’ recruitment lasted from August 2012 until the end of September 2012, when pretests were performed. The training intervention lasted from October 2012 until the end of March 2013, with follow-up test in April 2014. Table 2 shows baseline demographic characteristics and training compliance of the three intervention groups.

TABLE 2 | Baseline demographic characteristics and training compliance

Variable DANCE MEMORY PHYS p-value, two-tailed

N 24 22 25 Sex, female 14, 58.3% 16, 72.7% 16, 64.0% .602 Age, years 77.3 (6.3) 78.5 (5.1) 80.8 (4.7) .079t Height, cm 165.1 (7.7) 163.9 (8.5) 162.0 (8.9) .425 Weight, kg 75.8 (12.3) 73.6 (9.4) 69.5 (14.1) .198 BMI, kg/m2 23.0 (3.6) 22.4 (2.2) 21.4 (3.9) .259 MMSE, score 28.4 (1.4) 28.3 (1.2) 28.0 (1.7) .533 Total training compliance 84.3% (12.7%) 86.1% (9.1%) 87.1% (7.9%) .633 (52 sessions) Home-training compliance 79.9% (23.0%) 90.0% (14.8%) 83.5% (18.4%) .201 (eight sessions)

Notes: Data are means (± standard deviation in brackets) or numbers. Bold values indicate significance or trend, t p<.10 trend. Abbreviations: BMI, body mass index; MMSE, Mini Mental State Examination; DANCE, virtual reality video game dancing; MEMORY, treadmill walking with simultaneous verbal memory training; PHYS, treadmill walking.

5.3.1 Gait analysis

Performance data for all gait variables are presented in Table 3. Comparison of baseline performance shows significant differences between intervention groups for two gait variables (step length variability fast-DTC, p=.050; step time variability fast-DTC, p=.039). Baseline data were not different for the other gait variables (p-values from 0.11 to 0.91). Statistical details of the multiple regression analysis over the first three time points of measurement, including two planned comparisons or contrasts, are provided in

up SE 7.5 9.1 4% 9.5 8.9 5% 2.3 2.9 2% 2.7 2.6 3% 0.3 0.3 15% 0.3 0.4 18% -

8% 18% 2% 4% Follow mean 127.5 117.1 - 167.1 135.0 - 66.7 65.8 - 73.4 70.2 - 2.7 2.8 16% 2.6 3.1 34%

SE 4.7 6.1 3% 6.1 5.5 2% 1.6 2.0 1% 2.2 1.9 1% 0.2 0.2 16% 0.2 0.3 16%

12% 20% 2% 6% 15) Posttest mean 131.1 115.3 - 169.5 134.2 - 66.0 64.5 - 73.0 68.5 - 2.4 2.7 36% 2.7 3.4 38%

SE 5.3 6.1 3% 6.9 6.1 2% 1.7 2.0 2% 2.2 2.1 1% 0.1 0.3 14% 0.2 0.2 14% up N= up

Test

14% 24% 4% 8% mean 128.8 110.3 - 162.7 122.9 - 66.2 63.4 - 71.9 66.0 - 2.2 3.2 51% 2.5 3.1 44% Mid

SE 5.4 5.6 3% 5.5 6.3 3% 1.9 1.9 1% 1.9 1.9 1% 0.2 0.3 11% 0.2 0.2 12%

13% 24% 5% 8% PHYS (N=25, follow (N=25, PHYS Pretest mean 115.8 100.4 - 151.9 115.6 - 62.1 59.2 - 69.2 63.7 - 3.0 3.3 17% 2.8 3.8 52%

up SE 5.9 6.6 3% 8.4 7.7 2% 2.7 2.7 2% 2.8 2.9 1% 0.2 0.5 18% 0.2 0.3 10% -

1% 14% 5% Follow mean 127.5 126.3 - 166.7 143.1 - 66.4 66.2 0% 73.6 69.9 - 2.5 3.3 41% 2.6 3.3 28%

SE 5.4 6.0 2% 6.8 6.7 2% 2.3 2.6 1% 2.7 2.7 2% 0.1 0.2 9% 0.2 0.3 14% task. -

7% 17% 2% 7% up N=17) up Posttest mean 126.3 117.0 - 168.8 139.9 - 66.2 64.7 - 75.0 69.9 - 2.4 3.1 35% 2.8 3.5 31% - st, st, fast walking speed; MEMORY, treadmill walking with simultaneous verbal

SE 5.5 6.9 4% 7.3 6.9 3% 2.4 2.8 2% 2.7 2.8 2% 0.2 0.4 14% 0.2 0.3 11%

Test

13% 23% 6% 9% Mid mean 122.8 107.6 - 167.7 128.7 - 64.9 60.9 - 73.4 66.6 - 2.6 3.5 40% 2.6 3.3 34% task cost; Fa

SE 3.8 5.9 5% 5.1 6.0 3% 1.8 2.4 2% 2.3 2.4 2% 0.3 0.5 23% 0.3 0.4 13%

6% 20% 3% 7% MEMORY (N=22, follow (N=22, MEMORY Pretest mean 109.4 102.1 - 156.3 125.5 - 62.0 60.5 - 71.9 67.1 - 3.3 4.2 44% 3.2 3.9 33%

task; DTC, dual up - SE 8.6 8.3 3% 10.7 10.7 2% 3.1 2.8 2% 3.6 3.3 1% 0.2 0.1 16% 0.3 0.2 15% -

11% 3% Follow mean 130.0 129.1 0% 161.3 143.3 - 65.8 65.4 0% 70.6 68.0 - 2.5 2.9 35% 2.5 3.2 44%

SE 5.2 7.0 3% 7.1 7.2 2% 1.9 2.0 1% 2.2 2.3 1% 0.2 0.3 15% 0.2 0.3 13%

7% 19% 2% 7% Posttest mean 133.4 124.7 - 179.0 144.8 - 67.4 65.8 - 74.8 69.9 - 2.2 3.2 57% 2.8 3.4 30%

up N=15) up

SE 4.3 5.5 2% 6.7 8.6 4% 1.8 1.9 1% 2.4 2.2 1% 0.2 0.2 14% 0.2 0.3 22%

Test

4% 20% 2% 7% Mid mean 127.1 123.0 - 168.7 137.6 - 66.3 64.9 - 74.3 69.2 - 2.7 3.1 24% 2.5 3.5 64%

SE 5.3 6.0 3% 7.2 5.9 2% 2.1 1.8 2% 2.7 2.1 1% 0.2 0.2 11% 0.2 0.2 9%

8% 16% 2% 7% DANCE (N=24, follow (N=24, DANCE Pretest mean 123.0 113.8 - 157.3 130.5 - 65.3 63.5 - 72.5 67.2 - 3.0 3.4 27% 3.2 3.2 12%

ST DT DTC ST DT DTC ST DT DTC ST DT DTC ST DT DTC ST DT DTC DANCE, virtual reality video game dancing; DT, dual DT, game dancing; video reality virtual DANCE, ------

Condition Pref Pref Pref Fast Fast Fast Pref Pref Pref Fast Fast Fast Pref Pref Pref Fast Fast Fast

TABLE 3 | Gait performance data Abbreviations: single ST, error; standard ± SE, speed; walking preferred Pref, walking; treadmill PHYS, training; memory Variable Velocity (cm/s) length Step (cm) length Step variability (cm)

up SE 0.02 0.03 4% 0.01 0.02 5% 0.01 0.01 71% 0.01 0.01 57% -

Follow mean 0.54 0.58 8% 0.45 0.55 21% 0.03 0.05 169% 0.03 0.04 98%

SE 0.01 0.02 4% 0.01 0.01 3% 0.00 0.01 46% 0.00 0.00 31%

15) Posttest mean 0.51 0.58 14% 0.44 0.52 19% 0.02 0.04 145% 0.02 0.03 117%

SE 0.02 0.03 4% 0.01 0.02 4% 0.00 0.01 38% 0.00 0.01 45% up N= up

- Test

- mean 0.54 0.60 13% 0.45 0.55 23% 0.02 0.05 138% 0.02 0.04 139% Mid

SE 0.02 0.03 4% 0.01 0.01 3% 0.00 0.01 27% 0.00 0.00 11%

PHYS (N=25, follow (N=25, PHYS Pretest mean 0.56 0.62 12% 0.46 0.55 19% 0.03 0.05 78% 0.03 0.03 42%

up SE 0.01 0.01 2% 0.01 0.01 2% 0.01 0.00 20% 0.01 0.00 29% -

Follow mean 0.53 0.53 1% 0.44 0.49 12% 0.03 0.03 38% 0.03 0.02 30%

SE 0.01 0.01 2% 0.01 0.01 2% 0.00 0.01 25% 0.00 0.00 29% task. -

up N=17) up Posttest mean 0.54 0.57 6% 0.45 0.51 14% 0.02 0.03 64% 0.02 0.03 83% - st, st, fast walking speed; MEMORY, treadmill walking with simultaneous verbal

SE 0.01 0.03 3% 0.01 0.02 3% 0.00 0.01 25% 0.00 0.01 39%

Test

- Mid mean 0.54 0.60 10% 0.45 0.53 20% 0.02 0.04 92% 0.02 0.04 121% task cost; Fa

SE 0.01 0.02 4% 0.01 0.01 4% 0.00 0.01 28% 0.00 0.01 42%

MEMORY (N=22, follow (N=22, MEMORY Pretest mean 0.57 0.62 8% 0.47 0.54 17% 0.03 0.05 88% 0.02 0.04 119%

task; DTC, dual up - SE 0.01 0.02 2% 0.01 0.02 2% 0.01 0.01 21% 0.01 0.01 61% -

Follow mean 0.51 0.52 0% 0.44 0.48 8% 0.03 0.03 48% 0.02 0.04 153%

SE 0.01 0.04 6% 0.01 0.02 3% 0.00 0.01 39% 0.00 0.00 19%

Posttest mean 0.51 0.57 10% 0.42 0.49 17% 0.02 0.03 113% 0.02 0.03 70%

up N=15) up

SE 0.01 0.01 2% 0.01 0.02 2% 0.00 0.00 25% 0.00 0.00 17%

Test

Mid mean 0.52 0.54 3% 0.45 0.49 11% 0.02 0.03 64% 0.02 0.03 90%

SE 0.01 0.03 4% 0.01 0.02 2% 0.00 0.01 19% 0.01 0.00 15%

DANCE (N=24, follow (N=24, DANCE Pretest mean 0.53 0.58 7% 0.46 0.52 13% 0.03 0.04 52% 0.03 0.03 34%

| Gait performance data

ST DT DTC ST DT DTC ST DT DTC ST DT DTC DANCE, virtual reality video game dancing; DT, dual DT, game dancing; video reality virtual DANCE, ------

Condition Pref Pref Pref Fast Fast Fast Pref Pref Pref Fast Fast Fast

, continued

TABLE 3 Abbreviations: single ST, error; standard ± SE, speed; walking preferred Pref, walking; treadmill PHYS, training; memory Variable time Step (s) time Step variability (s)

95 Tables S1–S5. Linear global time effect showed significant performance improvements in all intervention groups from pretest to 6-month test in 19 of the 20 gait variables (all p<.05, R2 from 0.008 to 0.118). In the DTC gait variables, the linear global time effect did not show any significant reductions (p- values from 0.069 to 0.96).

In the first contrast of the multiple regression analysis, a significant linear interaction was found in “step time variability fast-DTC” with no change in the two cognitive–motor interventions (DANCE and MEMORY) and rising DTC in PHYS (Figure 4A; F(1, 136) = 2.95, p=.044, one-tailed, R2=.010).

The second contrast indicated three time × intervention interaction effects (Figure 4B–D). A significant linear interaction was found for the variable “step time fast-ST”, showing continuous improvement in DANCE and maintained performance for MEMORY (F(1, 136) = 7.51, p=.007, two-tailed, R2=.009). A trend for a significant linear interaction was identified for “step length variability preferred-DT”, representing unchanged performance for DANCE and declining variability for MEMORY (F(1, 136) = 3.53, trend p=.062, two-tailed, R2=.009). “Step time variability preferred-DTC” showed another trend for a significant linear interaction with an increase in DANCE and a reduction in MEMORY over the course of the 6-month training (F(1, 136) = 3.55, trend p=.062, two-tailed, R2=.011).

Performance remained unchanged from 6 months to follow-up test in 12 of the 20 gait variables, whereas performance in six variables decreased (statistical analyses available in Table S6). The variables “velocity preferred-DT” and “step-time preferred-DT” showed significant improvements (F(2, 44) = 7.10, p=.011, two-tailed, r=.37 and F(2, 44) = 9.36, p=.004, two-tailed, r=.42, respectively). DTCs of walking were maintained after 1-year follow-up in four of the ten DTC gait variables and were reduced, significantly or with a trend, in the six other DTC variables.

FIGURE 4 | Development of the gait variables that showed time × intervention contrasts from pretest to 6-month test.

Notes: (A) Depicts a significant first contrast with unchanged DTC in the two cognitive–motor interventions (DANCE and MEMORY) due to parallel improvements in ST and DT and rising DTC in PHYS due to improvement in ST and no change in DT (p=.044, one-tailed). (B–D) illustrate significant or trend to significant second contrasts (DANCE versus MEMORY): (B) in favor of DANCE (p=.007, two- tailed) and (C and D) in favor of MEMORY (trend p=.062, two-tailed and trend p=.062, two-tailed, respectively). Performance from 6 months until 1-year follow-up remained unchanged in (A, C, and D; all p>.10), whereas “step time fast-ST” increased (B; trend p=.077, two-tailed). error bars indicate ± standard error of the mean. Abbreviations: DTC, dual-task cost; DANCE, virtual reality video game dancing; MEMORY, treadmill walking with simultaneous verbal memory training; ST, single-task; DT, dual-task; PHYS, treadmill walking.

97 5.3.2 Fall frequency

Fall frequency of the three intervention groups is depicted in Figure 5. Base- line data were not different between groups (p=.23). Multiple regression analysis of the linear global time effect over the first three 6-month periods of fall observation revealed a significant reduction in fall frequency (F(1, 126) = 12.93, p<.001, one-tailed, R2=.056). No differences between interventions were found with planned comparisons (all p<.05). Statistical details for the multiple regression analysis are provided in Table S7. A trend was evident for an increase in fall frequency from the 6-month period after training to the period from 6 months to 12 months (F(1, 63) = 3.70, trend p=.059, two-tailed, r=.24).

FIGURE 5 | Development of fall frequency over 2 years.

Notes: A significant reduction over the first three 6-month periods (p<.001, one-tailed) and a subsequent trend for an increase after the fourth period was shown (trend p=.059, two-tailed). error bars indicate ± standard error of the mean. Abbreviations: DANCE, virtual reality video game dancing; MEMORY, treadmill walking with simultaneous verbal memory training; PHYS, treadmill walking.

98 5.3.3 SPPB and 6-MWT

Baseline performance was not different between intervention groups (p-values from 0.61 to 0.99). Multiple regression analysis of the linear global time effect showed a significant increase in all functional fitness measures from pretest to 6-month test (all p<.001, one-tailed, R2 from 0.023 to 0.141). Per- formance declined significantly from 6 months to follow-up test in SPPB total score (F(1, 44) = 5.26, p=.027, two-tailed, r=.33), while performance in the SPPB subtests 3-m-walk, 5-chair-rises, and extended balance test was maintained (all p<.10). Planned comparisons did not reveal any significant time × intervention interactions (all p<.05). Figure 6 depicts performance development for SPPB total score and 6-MWT distance. Performance data for the functional fitness variables are presented in Table 4, and details of statistical analyses are provided in Tables S8 and S9.

FIGURE 6 | Development of functional fitness.

Notes: (A) and (B) illustrate significant enhancements in two functional fitness measures from pretest to 6-month test (p<.001, one-tailed). Additionally, (A) shows attenuated performance after 1-year follow-up (p=.027, two-tailed). no differences between interventions were found. six-minute walk test was not repeated at follow-up. error bars indicate ± standard error of the mean. Abbreviations: DANCE, virtual reality video game dancing; MEMORY, treadmill walking with simultaneous verbal memory training; PHYS, treadmill walking.

99 5.3.4 Sex differences

This supplementary analysis indicated time × sex interactions in favor of the male participants in the gait variables “velocity fast-ST” (Figure 7A) and “step length fast-ST” (F(1.83, 126.05) = 3.20, p=.049, two-tailed, r=.16 and F(1.73, 119.08) = 2.68, trend p=.080, two-tailed, r=.15, respectively), as well as in the 5-chair-rises test (Figure 7B, F(1.42, 97.96) = 3.21, trend p=.061, two-tailed, r=.18). In the ANOVA for these three variables, Mauchly’s test indicated that the assumption of sphericity had been violated (χ2(2) = 6.77, 11.77, and 35.74, respectively, p<.05), therefore, Greenhouse–Geisser tests are reported (ε = 0.91, 0.86, and 0.71, respectively). Baseline performance was different between sexes in “velocity fast-ST” (trend p=.077) and “step length fast-ST” (p=.001) but not in 5-chair-rises (p=.68). Data without statistical trends or significant interactions are not presented.

FIGURE 7 | Performance developments for sexes.

Notes: (A) and (B) depict time × sex interactions from pretest to 6-month test in favor of the male participants in two measures related to muscular power (p=.049, two-tailed and trend p=.061, two- tailed, respectively). No significant interaction was found from 6-month test to 1-year follow-up. Error bars indicate ± standard error of the mean. Abbreviation: ST, single-task.

100 5.3.5 GDS and FES-I

Baseline data were not different between groups in GDS and FES-I (p=.38 and .55, respectively). Repeated measures ANOVA from pretest to 3- and 6- month tests did not reveal significant changes in both GDS (F(1.77, 120.16) = 0.045, p=.94, two-tailed) and FES-I (F(1, 136) = 1.55, p=.22, two-tailed). In the ANOVA for GDS, Mauchly’s test indicated that the assumption of sphericity had been violated, χ2(2) = 9.47, p<.05; therefore, Greenhouse–Geisser tests are reported (ε = 0.88). Similarly, from 6 months to follow-up test, scores in GDS and FES-I remained unchanged (F(1, 44) = 0.045, p=.83, two-tailed and F(1, 44) = 1.56, p=.22, two-tailed, respectively). Performance data for GDS and FES-I are presented in Table 4.

101

- SE 0.3 0.3 0.9 0.2 — — 0.6 0.7

MWT, MWT, -

Follow mean 11.5 5.4 8.0 2.9 — — 3.1 18.3

SE 0.2 0.2 0.5 0.1 21 0.4 0.7 1.0

Posttest mean 11.7 5.2 8.5 2.8 538 5.0 4.2 20.7

up N=15) up - SE 0.2 0.3 0.5 0.1 19 0.3 0.7 0.9

Test

Mid mean 11.5 5.0 8.5 3.2 514 3.9 4.5 19.6

SE 0.4 0.3 1.0 0.2 18 0.2 0.7 1.2

PHYS (N=25, follow (N=25, PHYS Pretest mean 10.5 4.5 11.4 3.3 506 3.4 4.3 21.2

- SE 0.2 0.2 0.6 0.1 — — 0.7 1.2

Follow mean 11.8 5.0 8.7 2.6 — — 3.6 20.5

SE 0.1 0.3 0.3 0.1 20 0.4 0.7 0.9

1 up N=17) up Posttest mean 11.8 5.2 8.5 2.8 530 4.3 3.3 20. -

2 SE 0. 0.2 0.7 0.1 19 0.3 0.8 0.7

Test

International; GDS, Geriatric Depression Scale; MEMORY, treadmill walking with - Mid mean 11.4 5.1 9.7 3.2 510 4.1 3.6 19.8

SE 0.3 0.4 0.7 0.1 16 0.3 0.5 0.7

MEMORY (N=22, follow (N=22, MEMORY Pretest mean 10.5 4.5 11.8 3.4 489 3.7 3.1 19.6

- SE 0.6 0.5 1.7 0.2 — — 0.8 0.9

I, Falls Efficacy Scale

Follow mean 10.7 4.9 10.0 3.0 — — 3.2 19.9

SE 0.2 0.3 0.5 0.1 21 0.4 0.6 0.7

ttest

Pos mean 11.6 5.1 8.7 2.8 560 4.4 3.1 19.2 up N=15) up - I performance data

SE 0.4 0.3 0.7 0.1 21 0.3 0.4 0.8

YS, YS, treadmill walking; RPE, rate of perceived exertion; SE, ± standard error; SPPB, Short Physical Performance Battery; 6 Test

Mid mean 11.2 5.0 9.3 3.1 522 3.5 2.7 19.6

SE 0.4 0.4 1.0 0.2 25 0.3 0.6 1.1

DANCE (N=24, follow (N=24, DANCE Pretest mean 10.5 4.8 12.0 3.3 505 3.3 3.3 20.1

DANCE, DANCE, virtual reality video game dancing; FES

rises(s) -

walk (s) walk -

chair m - -

I -

WT RPE WT minute walk test. walk minute - Variable score total SPPB balance adapted SPPB testscore 5 SPPB 3 SPPB 6MWT(m) 6M GDS FES TABLE 4 | Functional fitness, GDS, and FES Abbreviations: simultaneous verbal memory training; PH 6

102 5.4 DISCUSSION

This study aimed to compare two multicomponent simultaneous cognitive– motor training interventions with an exclusively physical exercise program on their effects on DT walking. The study comprised a 6-month training intervention and a 1-year follow-up. The two main findings were 1) that the two simultaneous cognitive–motor programs resulted in a significant advantage in DTCs of walking compared to the exclusively physical program but not in any other gait variables and 2) that the two simultaneous cognitive–motor interventions led to different training-specific adaptations in gait.

5.4.1 Are there advantages of simultaneous cognitive–motor training programs on DT walking?

This study found one indication of an advantage of the two cognitive–motor programs (DANCE and MEMORY) over PHYS. This significant first contrast was evident in the gait variable “step time variability fast-DTC”, whereby unchanged DTC in the two cognitive–motor interventions represented parallel improvements in ST and DT, and in contrast, rising DTC in PHYS resulted due to improvements in ST and no change in the DT walking condition over the 6-month training period. Baseline levels were different between interventions in this variable, but this should not affect the time × intervention interaction. Hence, this outcome probably reflects improved cognitive–motor DT abilities from simultaneous cognitive–motor training. Nevertheless, it is important to note that the analyses of the first contrast did not reach statistical significance in any other gait variable. Moreover, global linear time effects were significant in almost all gait variables and, therefore, support very similar improvements in the three intervention groups. Similarly, Forte et al. (2013) reported improvements in DT fast walking after both single strength training and multicomponent cognitive–motor training. Their and our own findings support the importance of the physical training components in simultaneous cognitive–motor programs for older adults.

Additionally, the results of this study demonstrated that the two simultaneous cognitive–motor training variations led to differential adaptations in gait reflected by the second contrast (DANCE versus MEMORY). DANCE, which

103 promotes fast, rhythmic, and accurate foot movements, improved step time (or cadence) significantly more in the fast walking condition than MEMORY. According to the five-factor model for gait domains suggested by Lord et al. (2013) step time belongs to the “rhythm” domain of gait, with the other domains being pace, variability, asymmetry, and postural control. The five factors are suggested to be associated with selected cognitive and motor characteristics under the assumption that gait is not a unitary concept (Lord et al., 2013). It was reported that gait rhythm is associated with performance in cognitive tests for information processing speed and attention (Stroop reading and Stroop color naming test, and Letter-Digit Substitution Task) (Verlinden et al., 2014). Both these cognitive domains are also part of the video game dance training and, therefore, seem to support the finding of a training-specific adaptation or a transfer to a related gait variable.

Another training-specific adaptation or transfer to gait was found after the MEMORY training. This training represents typical cognitive–motor DT training and specifically reduced variability of step length under the DT condition, as well as DTC of step time variability compared to DANCE. Thereby, in the MEMORY training, treadmill walking itself might have additionally promoted lower gait variability since participants must adapt to the constant treadmill speed. Our assumption is supported by a pilot study, including nine patients with Parkinson’s disease, which reported a trend for reduced swing time variability after 6 weeks of ST treadmill walking (Herman et al., 2007).

Interestingly, elevated stride length variability is related to lower levels of hippocampal neuronal metabolism in older persons with mild memory impairments (Zimmerman et al., 2009). The hippocampus plays a role not only in locomotion but also in memory and learning (Wicking et al., 2014), and reductions in hippocampal volume and metabolism are associated with diffi- culties in verbal memory in older adults without dementia (Zimmerman et al., 2009). These findings support our notion of a second training-specific adaptation since verbal memory was explicitly practiced in the MEMORY training. This result is of great relevance because gait variability was also suggested to be the most important intermediate gait variable in the relation between cognition and fall risk (Hausdorff et al., 2001; Verlinden et al., 2014).

To date, no other study has investigated training-specific adaptations or transfers to gait after different simultaneous cognitive–motor training

104 modalities. Nonetheless, in two previous studies, virtual reality video game dancing resulted in increased walking speed in the DT fast walking condition compared to strength and balance training alone (Pichierri et al., 2012b) and reduced DTCs of walking compared to a usual care physical training (de Bruin et al., 2011). However, interpretation of these studies’ results is limited since training volume was higher in the video game dancing groups.

5.4.2 Are performance gains in gait maintained?

Performance in six gait variables was reduced after 1 year without any further training being applied by the investigators. The reduction in gait performance was particularly prominent in the fast-ST walking condition, affecting velocity, step length, step time, and step time variability. Additionally, two gait variability measures were significantly elevated at preferred walking speed (step length variability and step time variability). In contrast, gait performance under preferred and DT walking conditions was maintained until 1-year follow-up. This finding may demonstrate retention of more basic fitness parameters, whereas fast walking is physically more demanding and therefore showed an earlier reduction due to detraining. Elevated gait variability may be an early indication of increased fall risk, since it was attributed to be the most important intermediate gait variable in association with cognition and fall risk as mentioned previously (Hausdorff et al., 2001; Verlinden et al., 2014). In fact, fall frequency tended to increase in the 6- to 12-month period after training, compared to the previous 6-month period during follow- up.

Nonetheless, maintenance of several gait and functional fitness variables over a relatively long period of 1 year is quite surprising and may be explained by two reasons: 1) the application of a progressive training protocol with sufficient volume and intensity over the 6-month period and 2) some participants continuing to exercise individually during follow-up. This outcome also demonstrates that our interventions successfully motivated and enabled the elderly to keep up training after the intervention. To our knowledge, only one other multicomponent intervention study performed 6 months of training and a 1-year follow-up with older adults. This study reported similar long-lasting training effects in functional fitness parameters, but some decline in strength measures (Gudlaugsson et al., 2012). In contrast, the few other

105 multicomponent training studies in older persons, which we are aware of, did not successfully maintain physical performance, although applying shorter follow-up periods of 6 weeks (Toraman and Ayceman, 2005) or 3 months (Car- valho et al., 2009; Seco et al., 2013), and despite performing quite long interventions of 8 months or 9 months training (Carvalho et al., 2009; Seco et al., 2013), respectively. Reasons for the lack of retention effects in the latter two studies may be the lower training intensity due to the large training groups of 15–20 persons.

5.4.3 Effects on fall frequency and functional fitness

For the development of fall frequency and functional fitness, planned comparisons did not show any significant contrasts between the simultaneous cognitive–motor and exclusively physical interventions. Nevertheless, global linear time effects indicated that each of the three training programs increased functional fitness and very effectively reduced fall frequency for ~77% from 0.79 falls per person-year at baseline to 0.18 falls per person-year during the first 6-month period after training. This is about half of the commonly reported minimal fall frequency of 0.33 falls per person-year in this age group (Tinetti et al., 1988; Hausdorff et al., 2001; Gill et al., 2005). In our study, fall frequency was doubled thereafter to ~0.39 falls per person-year within the 6–12- month period after training, demonstrating that preventive effects from training were fading out. Our interventions led to a considerably larger reduction in fall frequency compared to the 16 trials included in the meta-analysis by Gillespie et al. (2012). These multicomponent group exercise trials included totally 3’622 participants and showed reduction of fall frequency by 29% compared to control participants.

Additionally, the present results demonstrated sex-specific training adaptations in favor of the male participants in measures that are related to muscular power performance. We are not aware of any other reports about sex differences in functional outcomes after multicomponent training in elderly persons. Nonetheless, this outcome may be explained by smaller absolute muscle hypertrophic adaptations after strength training in women compared to men, as reported by Melnyk et al. (2009). These authors hypothesized that this difference might be associated with sex differences in blood androgen levels. However, other studies demonstrated similar relative hypertrophic,

106 neuromuscular, or maximal strength adaptations in men and women of younger and older age (Tracy et al., 1999; Ivey et al., 2000; Roth et al., 2001; Folland and Williams, 2007).

5.4.4 Strengths and limitations

Methodological strengths of this study were the comparably large number of participants and the long training period with follow-up measurements. How- ever, some limitations have to be considered as well. First, the conclusions and recommendations from this study are limited to physically and mentally healthy elderly persons, because such participants were included in the study. Training effects might have been even larger in a population of lower physical and mental status. This assumption is based on the exercise training principle “Initial Values”, stating that improvement in the outcome of interest will be greatest in those with lower initial values (Ammann et al., 2014). Fur- ther, we did not include a passive control group in the design of the study, which means that we could not exactly differentiate between training effects and learning effects from repeated testing. However, this was not the main focus of the present study since previous research has demonstrated training- related gains in physical functioning (Gillespie et al., 2012; Gudlaugsson et al., 2012; van het Reve and de Bruin, 2014). Based on similar results from the literature and the long intervals of 3 months between test sessions, we assume that performance improvements in our study can mostly be accounted for as training effects. Although participants were blinded to the expected study outcome, blinding of the investigators was not possible since they also supervised and conducted training and testing sessions. This is an additional limitation to this study.

5.5 CONCLUSIONS

This is the first study comparing long-term training and retention effects of two multicomponent cognitive–motor programs with an exclusively physical exercise program on DT gait in healthy older adults. Thereby, a significant advantage of the simultaneous cognitive–motor training programs became

107 evident in DTC of gait variability, which supports our first hypothesis. None- theless, each of the three multicomponent programs efficiently increased performance in most other gait variables. In accordance to our second hypothesis, the two novel simultaneous cognitive–motor training programs led to differential training-specific adaptations in the rhythm and variability domains of gait. Gait performance was partly retained over the relatively long period of 1 year after all three programs, with some attenuation in fast walking speed and gait variability. These two variables may serve as early indicators of functional fitness decline and increased fall risk in clinical settings.

To summarize, we conclude that the two novel training concepts of simultaneous cognitive–motor training and the exclusively physical exercise program displayed similarly great potential to counteract age-related decline of physical functioning in the elderly persons, while possible advantages of simultaneous cognitive–motor interventions are well worth further investigation.

Acknowledgments This work was supported by the Zürcher Kantonalbank within the framework of sponsoring of Movement Sciences, Sports and Nutritional Sciences at ETH Zurich. Zürcher Kantonalbank had no influence on the study design and the analyses presented in this paper, had no access to the data, and did not contribute to this manuscript in any way. The authors would like to thank PD Dr. med Thomas Münzer, chief physician, and the management of Geriat- rische Klinik, St.Gallen, Switzerland, for supporting the study and providing room for training and data acquisition. Furthermore, we thank our postgraduate students, Marius Angst, Fabienne Hüppin, Manuela Kobelt, Alexandra Schättin, and Sara Tomovic for instructing trainings and helping with data acquisition. We very much appreciated the support of the team of physiotherapists at Geriatrische Klinik, St.Gallen. Last but not least, we would like to thank all participants for their enthusiasm, kindness, and patience during our extensive training and testing interventions.

Author contributions PE contributed to study preparation and conception, participants’ recruitment, data acquisition, statistical analysis, data interpretation, and drafting

108 the manuscript. NT contributed to study conception, conception of serial position training, supporting statistical analysis, data interpretation, and revising manuscript. SH contributed to study preparation, training instruction, data acquisition and interpretation, and revising manuscript. VS contributed to study conception, data interpretation, and revising manuscript. EDB contributed to study conception, data interpretation, and critically revising the manuscript. All authors read and approved the final manuscript.

Disclosure The authors report no conflicts of interest in this work.

109 5.6 SUPPLEMENTARY MATERIALS

Table S1 | Multiple regression for the linear global time effect (from pretest to 6-month test, N=71) and the interaction between orthogonal contrasts and time effect for gait variable “velocity”

Dependent Predictor b 95% CI SE b β p, two- R2 - variable tailed change

Velocity Linear global time effect 7.10 5.19 9.01 0.97 .23 <.001*** .053 preferred Quadratic global time effect -1.03 -2.13 0.08 0.56 -.06 .068t .003 Linear interaction AB × C -0.26 -1.60 1.07 0.67 -.01 .695 .000 Linear interaction A × B -1.63 -4.00 0.75 1.20 -.04 .178 .002 Velocity fast Linear global time effect 8.64 6.35 10.92 1.16 .22 <.001*** .048 Quadratic global time effect -0.86 -2.18 0.47 0.67 -.04 .203 .001 Linear interaction AB × C -0.08 -1.68 1.51 0.81 .00 .919 .000 Linear interaction A × B 2.27 -0.57 5.11 1.44 .05 .116 .002 Velocity Linear global time effect 6.79 4.60 8.99 1.11 .18 <.001*** .033 preferred DT Quadratic global time effect -0.48 -1.74 0.79 0.64 -.02 .456 .000 Linear interaction AB × C -0.33 -1.85 1.20 0.77 -.01 .674 .000 Linear interaction A × B -1.00 -3.72 1.72 1.38 -.02 .468 .000 Velocity fast Linear global time effect 7.90 5.40 10.39 1.26 .20 <.001*** .039 DT Quadratic global time effect 0.67 -0.77 2.11 0.73 .03 .361 .001 Linear interaction AB × C -0.69 -2.43 1.05 0.88 -.03 .431 .001 Linear interaction A × B -0.03 -3.13 3.07 1.57 .00 .985 .000 Velocity Linear global time effect 0.00 -0.02 0.02 0.01 .00 .930 .000 preferred Quadratic global time effect 0.00 -0.01 0.01 0.01 .03 .480 .001 DTC Linear interaction AB × C 0.00 -0.01 0.01 0.01 -.01 .858 .000 Linear interaction A × B 0.00 -0.02 0.03 0.01 .02 .717 .000 Velocity fast Linear global time effect 0.01 -0.01 0.02 0.01 .04 .461 .002 DTC Quadratic global time effect 0.01 0.00 0.02 0.01 .10 .065t .011 Linear interaction AB × C -0.01 -0.02 0.01 0.01 -.06 .258 .004 Linear interaction A × B -0.01 -0.04 0.01 0.01 -.07 .213 .005

Notes: Bold values indicate significance or trend. *** p<.001. t p<.10 trend. Abbreviations: A, DANCE; B, MEMORY; C, PHYS; CI, confidence interval; DANCE, virtual reality video game dancing; DT, dual- task; DTC, dual-task cost; MEMORY, treadmill walking with simultaneous verbal memory training; PHYS, treadmill walking; se, ± standard error.

110 Table S2 | Multiple regression for the linear global time effect (from pretest to 6-month test, N=71) and the interaction between orthogonal contrasts and time effect for gait variable “step length”

Dependent Predictor b 95% CI SE b β p, two- R2 - variable tailed change

Step length Linear global time effect 1.71 1.12 2.30 0.30 .15 <.001*** .022 preferred Quadratic global time effect -0.31 -0.65 0.03 0.17 -.05 .072t .002 Linear interaction AB × C -0.12 -0.54 0.29 0.21 -.02 .551 .000 Linear interaction A × B -0.54 -1.27 0.19 0.37 -.04 .147 .001 Step length Linear global time effect 1.55 0.97 2.13 0.30 .11 <.001*** .012 fast Quadratic global time effect -0.16 -0.50 0.18 0.17 -.02 .352 .000 Linear interaction AB × C -0.19 -0.60 0.21 0.21 -.02 .346 .000 Linear interaction A × B -0.19 -0.92 0.53 0.37 -.01 .600 .000 Step length Linear global time effect 1.98 1.37 2.60 0.31 .15 <.001*** .024 preferred Quadratic global time effect -0.02 -0.37 0.34 0.18 .00 .920 .000 DT Linear interaction AB × C -0.34 -0.77 0.09 0.22 -.04 .119 .001 Linear interaction A × B -0.49 -1.25 0.28 0.39 -.03 .209 .001 Step length Linear global time effect 1.69 1.06 2.33 0.32 .13 <.001*** .016 fast DT Quadratic global time effect 0.16 -0.21 0.52 0.19 .02 .400 .000 Linear interaction AB × C -0.33 -0.78 0.11 0.22 -.04 .136 .001 Linear interaction A × B -0.03 -0.81 0.76 0.40 .00 .947 .000 Step length Linear global time effect 0.00 -0.01 0.01 0.01 .04 .482 .001 preferred Quadratic global time effect 0.00 0.00 0.01 0.00 .08 .127 .006 DTC Linear interaction AB × C 0.00 -0.01 0.00 0.00 -.06 .293 .003 Linear interaction A × B 0.00 -0.01 0.01 0.01 -.01 .858 .000 Step length Linear global time effect 0.00 -0.01 0.01 0.00 .03 .494 .001 fast DTC Quadratic global time effect 0.01 0.00 0.01 0.00 .09 .049* .009 Linear interaction AB × C 0.00 -0.01 0.00 0.00 -.05 .264 .003 Linear interaction A × B 0.00 -0.01 0.01 0.01 .00 .983 .000

Notes: Bold values indicate significance or trend. * p<.05. *** p<.001. t p<.10 trend. Abbreviations: A, DANCE; B, MEMORY; C, PHYS; CI, confidence interval; DANCE, virtual reality video game dancing; DT, dual-task; DTC, dual-task cost; MEMORY, treadmill walking with simultaneous verbal memory training; PHYS, treadmill walking; se, ± standard error.

111 Table S3 | Multiple regression for the linear global time effect (from pretest to 6-month test, N=71) and the interaction between orthogonal contrasts and time effect for gait variable “step length variability”

Dependent Predictor b 95% CI SE b β p, two- R2 - variable tailed change

Step length Linear global time effect -0.39 -0.50 -0.27 0.06 -.35 <.001*** .118 variability Quadratic global time effect 0.08 0.01 0.15 0.03 .12 .020* .015 preferred Linear interaction AB × C -0.03 -0.11 0.06 0.04 -.03 .548 .001 Linear interaction A × B 0.01 -0.13 0.16 0.07 .01 .866 .000 Step length Linear global time effect -0.15 -0.31 0.01 0.08 -.11 .059t .013 variability Quadratic global time effect 0.13 0.04 0.22 0.05 .17 .005** .029 fast Linear interaction AB × C -0.05 -0.16 0.06 0.06 -.05 .377 .003 Linear interaction A × B 0.01 -0.18 0.21 0.10 .01 .911 .000 Step length Linear global time effect -0.32 -0.50 -0.13 0.09 -.18 .001** .030 variability Quadratic global time effect 0.03 -0.07 0.14 0.05 .03 .549 .001 preferred DT Linear interaction AB × C -0.02 -0.14 0.11 0.07 -.01 .817 .000 Linear interaction A × B 0.22 -0.01 0.45 0.12 .10 .062t .009 Step length Linear global time effect -0.10 -0.30 0.09 0.10 -.06 .286 .004 variability Quadratic global time effect 0.07 -0.04 0.19 0.06 .07 .192 .005 fast DT Linear interaction AB × C 0.05 -0.09 0.18 0.07 .04 .490 .002 Linear interaction A × B 0.14 -0.10 0.38 0.12 .06 .262 .004 Step length Linear global time effect 0.07 -0.04 0.17 0.05 .08 .229 .006 variability Quadratic global time effect -0.01 -0.07 0.05 0.03 -.02 .811 .000 preferred Linear interaction AB × C -0.02 -0.09 0.06 0.04 -.03 .693 .001 DTC Linear interaction A × B 0.10 -0.04 0.23 0.07 .09 .150 .008 Step length Linear global time effect 0.00 -0.11 0.12 0.06 .00 .962 .000 variability Quadratic global time effect -0.05 -0.11 0.02 0.03 -.10 .141 .010 fast DTC Linear interaction AB × C 0.04 -0.04 0.11 0.04 .06 .375 .004 Linear interaction A × B 0.05 -0.09 0.19 0.07 .05 .468 .002

Notes: Bold values indicate significance or trend. * p<.05. ** p<.01. *** p<.001. t p<.10 trend. Abbre- viations: A, DANCE; B, MEMORY; C, PHYS; CI, confidence interval; DANCE, virtual reality video game dancing; DT, dual-task; DTC, dual-task cost; MEMORY, treadmill walking with simultaneous verbal memory training; PHYS, treadmill walking; se, ± standard error.

112 Table S4 | Multiple regression for the linear global time effect (from pretest to 6-month test, N=71) and the interaction between orthogonal contrasts and time effect for gait variable “step time”

Dependent Predictor b 95% CI SE b β p, two- R2 - variable tailed change

Step time Linear global time effect -0.02 -0.024 -0.011 0.00 -.21 <.001*** .043 preferred Quadratic global time effect 0.00 -0.002 0.005 0.00 .03 .452 .001 Linear interaction AB × C 0.00 -0.002 0.007 0.00 .04 .331 .001 Linear interaction A × B 0.00 -0.004 0.012 0.00 .03 .373 .001 Step time Linear global time effect -0.01 -0.017 -0.009 0.00 -.23 <.001*** .054 fast Quadratic global time effect 0.00 -0.001 0.003 0.00 .04 .264 .001 Linear interaction AB × C 0.00 -0.003 0.002 0.00 -.01 .735 .000 Linear interaction A × B -0.01 -0.011 -0.002 0.00 -.09 .007** .009 Step time Linear global time effect -0.02 -0.027 -0.005 0.01 -.10 .006** .011 preferred DT Quadratic global time effect 0.00 -0.003 0.010 0.00 .04 .323 .001 Linear interaction AB × C 0.00 -0.006 0.009 0.00 .02 .680 .000 Linear interaction A × B 0.01 -0.003 0.024 0.01 .06 .136 .003 Step time Linear global time effect -0.02 -0.022 -0.008 0.00 -.16 <.001*** .026 fast DT Quadratic global time effect 0.00 -0.005 0.003 0.00 -.02 .662 .000 Linear interaction AB × C 0.00 -0.004 0.005 0.00 .01 .898 .000 Linear interaction A × B 0.00 -0.008 0.008 0.00 .00 .993 .000 Step time Linear global time effect 0.00 -0.016 0.024 0.01 .02 .690 .000 preferred Quadratic global time effect 0.00 -0.009 0.014 0.01 .02 .667 .000 DTC Linear interaction AB × C 0.00 -0.016 0.012 0.01 -.01 .808 .000 Linear interaction A × B 0.01 -0.012 0.038 0.01 .05 .295 .002 Step time Linear global time effect 0.00 -0.015 0.017 0.01 .01 .901 .000 fast DTC Quadratic global time effect -0.01 -0.014 0.004 0.01 -.05 .287 .002 Linear interaction AB × C 0.00 -0.009 0.013 0.01 .02 .726 .000 Linear interaction A × B 0.02 -0.001 0.038 0.01 .08 .069t .006

Notes: Bold values indicate significance or trend. ** p<.01. *** p<.001. t p<.10 trend. Abbreviations: A, DANCE; B, MEMORY; C, PHYS; CI, confidence interval; DANCE, virtual reality video game dancing; DT, dual-task; DTC, dual-task cost; MEMORY, treadmill walking with simultaneous verbal memory training; PHYS, treadmill walking; se, ± standard error.

113 Table S5 | Multiple regression for the linear global time effect (from pretest to 6-month test, N=71) and the interaction between orthogonal contrasts and time effect for gait variable “step time variability”

Dependent Predictor b 95% CI SE b β p, two- R2 - variable tailed change

Step time Linear global time effect -0.004 -0.005 -0.003 0.001 -.28 <.001*** .081 variability Quadratic global time effect 0.000 0.000 0.001 0.000 .05 .318 .003 preferred Linear interaction AB × C 0.001 0.000 0.002 0.000 .11 .034* .012 Linear interaction A × B -0.001 -0.003 0.001 0.001 -.05 .301 .003 Step time Linear global time effect -0.006 -0.009 -0.004 0.001 -.31 <.001*** .098 variability Quadratic global time effect 0.001 0.000 0.003 0.001 .12 .067t .014 fast Linear interaction AB × C 0.000 -0.002 0.002 0.001 .02 .791 .000 Linear interaction A × B -0.002 -0.005 0.001 0.002 -.08 .204 .007 Step time Linear global time effect -0.003 -0.007 0.000 0.002 -.09 .047* .008 variability Quadratic global time effect 0.000 -0.001 0.002 0.001 .02 .631 .000 preferred DT Linear interaction AB × C 0.000 -0.002 0.002 0.001 .00 .949 .000 Linear interaction A × B 0.003 -0.001 0.007 0.002 .06 .165 .004 Step time Linear global time effect -0.004 -0.007 0.000 0.002 -.12 .024* .015 variability Quadratic global time effect -0.002 -0.004 0.000 0.001 -.11 .047* .011 fast DT Linear interaction AB × C -0.001 -0.003 0.001 0.001 -.04 .438 .002 Linear interaction A × B 0.002 -0.002 0.006 0.002 .05 .339 .003 Step time Linear global time effect 0.194 -0.015 0.402 0.106 .10 .069t .010 variability Quadratic global time effect -0.027 -0.148 0.093 0.061 -.03 .657 .001 preferred Linear interaction AB × C -0.071 -0.216 0.075 0.074 -.05 .339 .003 DTC Linear interaction A × B 0.247 -0.012 0.506 0.131 .11 .062t .011 Step time Linear global time effect 0.123 -0.083 0.329 0.104 .07 .241 .005 variability Quadratic global time effect -0.131 -0.250 -0.012 0.060 -.13 .032* .016 fast DTC Linear interaction AB × C -0.125 -0.268 0.019 0.073 -.10 .088t .010 Linear interaction A × B 0.181 -0.075 0.437 0.129 .08 .165 .007

114 Table S6 | Repeated-measures ANOVA from 6 months to follow-up test for gait variables, N=47

Dependent variable Effect F(2, 44) p, two-tailed r

Velocity preferred Time 0.090 .765 .05 Time × intervention 0.207 .813 .07 Velocity fast Time 7.844 .008** .39 Time × intervention 2.097 .135 .21 Velocity preferred DT Time 7.095 .011* .37 Time × intervention 0.094 .910 .05 Velocity fast DT Time 0.073 .789 .04 Time × intervention 0.299 .743 .08 Velocity preferred DTC Time 11.667 .001** .46 Time × intervention 0.075 .928 .04 Velocity fast DTC Time 9.498 .004** .42 Time × intervention 1.718 .191 .19 Step length preferred Time 0.004 .951 .01 Time × intervention 0.179 .837 .06 Step length fast Time 10.23 .003** .43 Time × intervention 0.444 .644 .10 Step length preferred DT Time 2.413 .127 .23 Time × intervention 0.392 .678 .09 Step length fast DT Time 0.04 .842 .03 Time × intervention 0.552 .580 .11 Step length preferred DTC Time 3.522 .067t .27 Time × intervention 0.433 .652 .10 Step length fast DTC Time 5.165 .028* .32 Time × intervention 1.748 .186 .20 Step length variability preferred Time 6.57 .014* .36 Time × intervention 0.658 .523 .12 Step length variability fast Time 1.04 .314 .15 Time × intervention 0.173 .842 .06 Step length variability preferred Time 0.2 .657 .07 DT Time × intervention 0.047 .954 .03 Step length variability fast DT Time 0.889 .351 .14 Time × intervention 0.158 .855 .06 Step length variability preferred Time 0.985 .326 .15 DTC Time × intervention 0.722 .492 .13 Step length variability fast DTC Time 0.034 .855 .03 Time × intervention 0.412 .665 .10

Notes: Bold values indicate significance or trend. * p<.05. ** p<.01. t p<.10 trend. Abbreviations: ANOVA, analysis of variance; DT, dual-task; DTC, dual-task cost.

115 Table S6, continued | Repeated-measures ANOVA from 6 months to follow-up test for gait variables, N=47

Dependent variable Effect F(2, 44) p, two-tailed r

Step time preferred Time 0.079 .779 .04 Time × intervention 1.056 .357 .15 Step time fast Time 3.287 .077t .26 Time × intervention 3.312 .046* .26 Step time preferred DT Time 9.355 .004** .42 Time × intervention 0.077 .926 .04 Step time fast DT Time 0.124 .727 .05 Time × intervention 0.67 .517 .12

Step time preferred DTC Time 11.028 .002** .45 Time × intervention 0.267 .767 .08 Step time fast DTC Time 3.514 .068t .27 Time × intervention 1.975 .151 .21 Step time variability preferred Time 4.176 .047* .29 Time × intervention 0.074 .929 .04 Step time variability fast Time 7.101 .011* .37 Time × intervention 0.204 .816 .07 Step time variability preferred Time 1.378 .247 .17 DT Time × intervention 0.137 .873 .06 Step time variability fast DT Time 2.702 .107 .24 Time × intervention 0.638 .533 .12 Step time variability preferred Time 2.549 .118 .23 DTC Time × intervention 0.089 .915 .04 Step time variability fast DTC Time 0.105 .747 .05 Time × intervention 1.393 .259 .18

Notes: Bold values indicate significance or trend. * p<.05. ** p<.01. t p<.10 trend. Abbreviations: ANOVA, analysis of variance; DT, dual-task; DTC, dual-task cost.

116 Table S7 | Multiple regression for the linear global time effect (from the 6-month period before training, to training, to 6 months after training, N=66) and the interaction between orthogonal contrasts and time effect for fall frequency

Dependent Predictor b 95% CI SE b β p, two- R2 - variable tailed change

Fall Linear global time effect -0.15 -0.24 -0.07 0.04 -.24 <.001*** .056 frequency Quadratic global time effect -0.03 -0.08 0.01 0.02 -.09 .165 .009 Linear interaction AB × C -0.02 -0.08 0.04 0.03 -.04 .588 .001 Linear interaction A × B 0.02 -0.08 0.12 0.05 .03 .663 .001

Notes: Bold values indicate significance or trend. *** p<.001. Abbreviations: A, DANCE; B, MEMORY; C, PHYS; CI, confidence interval; DANCE, virtual reality video game dancing; MEMORY, treadmill walking with simultaneous verbal memory training; PHYS, treadmill walking; se, ± standard error.

Table S8 | Multiple regression for the linear global time effect (from pretest to 6-month test, N=71) and the interaction between orthogonal contrasts and time effect for functional fitness variables

Dependent Predictor b 95% CI SE b β p, two- R2 - variable tailed change

SPPB, total Linear global time effect 0.60 0.46 0.74 0.07 .34 <.001*** .115 score Linear interaction AB × C 0.01 -0.09 0.11 0.05 .01 .845 .000 Linear interaction A × B -0.05 -0.22 0.13 0.09 -.02 .595 .000 SPPB, Linear global time effect 0.27 0.13 0.41 0.07 .15 <.001*** .023 adapted Linear interaction AB × C -0.04 -0.13 0.06 0.05 -.03 .471 .001 balance test Linear interaction A × B -0.09 -0.26 0.09 0.09 -.04 .333 .002 SPPB, Linear global time effect -1.64 -1.99 -1.30 0.17 -.38 <.001*** .141 5-chair- Linear interaction AB × C -0.02 -0.26 0.22 0.12 -.01 .882 .000 rises Linear interaction A × B 0.02 -0.41 0.45 0.22 .00 .923 .000 SPPB, Linear global time effect -0.26 -0.32 -0.19 0.03 -.30 <.001*** .090 3-m-walk Linear interaction AB × C -0.02 -0.06 0.03 0.02 -.03 .503 .001 Linear interaction A × B 0.06 -0.02 0.14 0.04 .06 .152 .003 6-min-walk, Linear global time effect 21.32 15.91 26.74 2.74 .18 <.001*** .032 distance Linear interaction AB × C 2.52 -1.26 6.29 1.91 .03 .190 .001 Linear interaction A × B 3.33 -3.40 10.06 3.40 .02 .329 .001 6-min-walk, Linear global time effect 0.54 0.35 0.74 0.10 .26 <.001*** .067 RPE Linear interaction AB × C -0.12 -0.26 0.02 0.07 -.08 .083t .007 Linear interaction A × B 0.13 -0.11 0.37 0.12 .05 .294 .002

Notes: Bold values indicate significance or trend. *** p<.001. t p<.10 trend. Abbreviations: A, DANCE; B, MEMORY; C, PHYS; CI, confidence interval; DANCE, virtual reality video game dancing; MEMORY, treadmill walking with simultaneous verbal memory training; PHYS, treadmill walking; se, ± standard error; SPPB, Short Physical Performance Battery; RPE, rate of perceived exertion.

117 Table S9 | repeated-measures ANOVA from 6 months to follow-up test for SPPB variables, N=47

Dependent variable Effect F(2, 44) p, two-tailed r

SPPB, total score Time 5.260 .027* .33 Time × intervention 1.688 .197 .19 SPPB, adapted balance test Time 0.591 .446 .12 Time × intervention 0.336 .716 .09 SPPB, 5-chair-rises Time 2.465 .124 .23 Time × intervention 0.719 .493 .13 SPPB, 3-m-walk Time 0.154 .696 .06 Time × intervention 0.566 .572 .11

Notes: Bold values indicate significance or trend. * p<.05. Abbreviations: ANOVA, analysis of variance; SPPB, Short Physical Performance Battery.

118 6

Brain functional adaptations (study 3)

119 EXERGAME AND BALANCE TRAINING MODULATE PREFRONTAL BRAIN ACTIVITY DURING WALKING AND ENHANCE EXECUTIVE FUNCTION IN OLDER ADULTS

Patrick Eggenberger1, Martin Wolf2, Martina Schumann1, and Eling D. de Bruin1

1Institute of Human Movement Sciences and Sport, Department of Health Sciences and Technology, ETH Zurich, 2Biomedical Optics Research Labora- tory, Department of Neonatology, University Hospital Zurich, Switzerland

Published in Frontiers in Aging Neuroscience, 12 April 2016: 8, 66. http://dx.doi.org/10.3389/fnagi.2016.00066

120 ABSTRACT

Background: Different types of exercise training have the potential to induce structural and functional brain plasticity in the elderly. Thereby, functional brain adaptations were observed during cognitive tasks in functional magnetic resonance imaging studies that correlated with improved cognitive performance. This study aimed to investigate if exercise training induces functional brain plasticity during challenging treadmill walking and elicits associated changes in cognitive executive functions.

Methods: Forty-two elderly participants were recruited and randomly assigned to either interactive cognitive–motor video game dancing (DANCE) or balance and stretching training (BALANCE). The 8-week intervention included three sessions of 30 min per week and was completed by 33 participants (mean age 74.9 ± 6.9 years). Prefrontal cortex (PFC) activity during preferred and fast walking speed on a treadmill was assessed applying functional near infrared spectroscopy pre- and post-intervention. Additionally, executive functions comprising shifting, inhibition, and working memory were assessed.

Results: The results showed that both interventions significantly reduced left and right hemispheric PFC oxygenation during the acceleration of walking (p<.05 or trend, r=.25 – .36), while DANCE showed a larger reduction at the end of the 30-s walking task compared to BALANCE in the left PFC (F(1, 31) = 3.54, p=.035, r=.32). These exercise training-induced modulations in PFC oxygenation correlated with improved executive functions (p<.05 or trend, r=.31 – 0.50).

Conclusions: The observed reductions in PFC activity may release cognitive resources to focus attention on other processes while walking, which could be relevant to improve mobility and falls prevention in the elderly. This study provides a deeper understanding of the associations between exercise training, brain function during walking, and cognition in older adults.

Keywords: prefrontal cortex, hemispheric asymmetry, functional near-infrared spectroscopy, cognition, gait, interactive cognitive–motor video game dancing, simultaneous cognitive–motor training, elderly

121 6.1 INTRODUCTION

The human brain experiences a variety of prominent structural changes during the course of aging. The most apparent change includes a progressive brain atrophy, which is compensated for by increased ventricular spaces and cerebrospinal fluid. Shrinkage of the brain’s size accelerates in old age and is gradually manifested from anterior to posterior regions. Thereby, the volume of prefrontal regions is affected in the most pronounced manner (Deary et al., 2009; Lockhart and DeCarli, 2014). Advancing age is also associated with a reduction of cognitive functioning, which is prevalent in almost every second elderly person (Scafato et al., 2010). However, the evidence appears not to be sufficient to conclude that altered brain structures reflect the neuroanatomical substrates for the age-related decline of cognitive performance (Fjell and Walhovd, 2010; Salthouse, 2011; Bennett and Madden, 2014). An important reason for this seemingly discrepancy is that the aging brain is able to coun- terbalance structural attenuations by altering the functional recruiting patterns and, thereby, maintaining cognitive functions (Deary et al., 2009; Grady, 2012). Such processes reflect functional brain and cognitive plasticity in the aging human brain.

Intriguingly, functional and structural brain plasticity has been observed in older adults after various types of physical exercise training, including cardiovascular, strength, coordination, and balance training, and was consistently correlated with improved cognitive performance (Voss et al., 2010; Erickson et al., 2011; Voelcker-Rehage et al., 2011; Liu-Ambrose et al., 2012; Voss et al., 2013; ten Brinke et al., 2015). Nonetheless, the functional brain adaptations were evident in simple cognitive tasks that can be performed during functional magnetic resonance imaging (fMRI) administration. This leaves the important question unanswered whether functional brain adaptations after exercise training could also be observed in real-life physical functioning of older adults, such as during walking? It was demonstrated with functional near-infrared spectroscopy (fNIRS) that brain activity is elevated during walking particularly in the prefrontal cortex (PFC), the premotor cortex, and in the supplementary motor area (Harada et al., 2009; Holtzer et al., 2014; Hamacher et al., 2015). Moreover, brain activity was further increased to maintain gait in circumstances with interference from additional secondary tasks (Holtzer et al., 2011; Lu et al., 2015) or under challenging walking

122 situations (Suzuki et al., 2004; Koenraadt et al., 2014). Although, none of these studies assessed cognitive functioning, their findings of frontal and prefrontal brain areas being involved in walking tasks supports the theory that attention and executive functions, which are associated with these brain areas, play an important role in locomotion in older adults (Amboni et al., 2013). Emerging evidence indicates that aging-related decline in higher order cognitive processing, e.g., in attention and executive functioning, are associated with impaired gait in older adults (Yogev-Seligmann et al., 2008; de Bruin and Schmidt, 2010). It would, therefore, be of particular interest to under- stand brain-mediated effects of exercise training on improved mobility and cognition in order to define effective exercise programs for the elderly.

Exercise programs comprising simultaneous cognitive and motor training, such as video game dancing, complemented with conventional strength and balance training were shown to improve both, single- and dual-task gait parameters (Pichierri et al., 2012b; Eggenberger et al., 2015b) as well as higher cognitive processing as measured by standard neuropsychological tests (Fra- ser et al., 2014; Eggenberger et al., 2015a). Also other interactive cognitive– motor step training games led to improvements in cognitive functions in older adults (Schoene et al., 2015). These findings support the results of a systematic review investigating the positive effects video games may have on cognition and brain structure (Shams et al., 2015). However, the assumption that exercise training-induced alterations in frontal and prefrontal brain functioning could play a mediating role for gait and cognitive improvements warrants further investigation to be verified. First, since the literature on walking-related brain activity is sparse and mostly included small samples of young adults. Moreover, longitudinal interventions assessing exercise training-induced changes in functional brain activity during walking and concomitant changes in cognitive performance are, to the best of our knowledge, lacking.

To date, it has not been shown that a cognitive–motor training based on a video game approach can improve brain functioning in frontal and prefrontal brain areas during walking. Therefore, the aim of this study was to compare the effects of cognitive–motor video game dancing against conventional balance training on PFC activity during walking and on executive functions. We hypothesized, (1) that cognitive–motor video game dancing would elicit larger training-induced reductions of PFC oxygenation during walking than

123 conventional balance training and (2) that training-induced changes in PFC oxygenation would correlate with changes in executive functions.

6.2 MATERIALS AND METHODS

6.2.1 Study design and participants

This study was a randomized, controlled trial, including a two groups parallel 8-week training intervention. Assessments were performed pre- and posttraining. Data collection and training were performed at Geriatrische Klinik St.Gallen, Switzerland. The study protocol was approved by the local ethics committee of the canton St.Gallen, Switzerland (study-number: EKSG 13/089) and registered at Current Controlled Trials under ISRCTN 82949128. No changes were made to the planned methods after trial commencement. Our reporting adheres to the CONSORT 2010 guidelines (Moher et al., 2010).

Participants’ recruitment lasted from September until November 2013. The first 8-week training block started in mid-October, and the second 8-week training block, with participants that were recruited later, started at the beginning of December 2013. Participants were recruited through a newspaper article, a local senior organization (Pro Senectute St.Gallen), senior residence facilities, and senior sports clubs. Interested persons were invited to an information event. We included male and female participants because both sexes are similarly affected by age-related cognitive decline (Scafato et al., 2010). For eligibility, participants had to be older than 65 years of age, live independently or at senior residence facilities, and sign informed consent in accordance with the Declaration of Helsinki. Participants had to be able to walk for about 10 min on a treadmill. Residents of retirement homes classified 0, 1, or 2 within the Swiss classification system for healthcare requirements (BESA-levels, German abbreviation for: Bewohner-Einstufungs- und Abrechnungs-System) could enroll in the study. Level 0 meaning the person does not need care or treatment; Level 1 to 2 meaning, the person only needs little care or treatment. Seniors with diagnosed Alzheimer’s disease, dementia, or recent head injury were excluded. Judgment by their primary care

124 physician was required in the case of acute or instable chronic diseases (e.g., stroke, diabetes) and rapidly progressing or terminal illnesses before accepting a person for participation.

A priori power analysis was performed with the G*Power 3.1.3 Software (Faul et al., 2007) and revealed a sample size of 34 participants in order to achieve 80% power for a two-groups pre- and posttest design (17 participants per group). The �-level was set at 0.05 and the effect size f at 0.25. For the two- groups randomization scheme, a random number between zero and 100 was generated for each participant with Microsoft Excel software. Thereafter, the lower half of random numbers/participants was assigned to the video game dance group and the upper half to the balance and stretching group. Partici- pants were blinded to the expected study outcome, while blinding of the investigators was not possible since they supervised and conducted training and testing sessions.

6.2.2 Training interventions

Three 30-min training sessions per week were performed on separate days (Monday, Wednesday, Friday) for 8 weeks. Groups of four participants were instructed by two trained postgraduate students. Training programs adhered to the current recommendations for physical fitness and fall prevention for older adults (Chodzko-Zajko et al., 2009; Sherrington et al., 2011). The exercise training principles of progression and overload were applied in both interventions (Ammann et al., 2014), and were adapted to each participant’s abilities to achieve a moderate to vigorous training intensity (Chodzko-Zajko et al., 2009). Twenty-four training sessions were performed within 8 weeks, with some participants missing certain sessions due to personal reasons.

Intervention DANCE included interactive video game dancing, a so-called exergame, as a simultaneous cognitive–motor training (Figure 1). This training combines an attention demanding cognitive task with a simultaneous motor coordination aspect. Two Impact Dance Platforms (Positive Gaming BV, Haarlem, the Netherlands) were positioned side-by-side such that two participants could practice at the same time. Participants stood on the one-by-one meter platform, which contained four pressure sensitive areas to detect steps forward, backward, to the left, and to the right, respectively. Stepping

125 sequences were cued with arrows appearing on a large screen. Steps had to be performed exactly when an arrow reached a highlighted area on the screen in order to achieve best scores in the game. Participants could hold on to ropes if necessary to maintain balance. Several levels of difficulty in stepping patterns and frequency were created with the StepMania software (StepMania, 2016) and different styles of music were chosen to add variety and meet participants’ preferences. Training difficulty was adapted to each individual’s coordination ability and was increased progressively.

FIGURE 1 | Interactive cognitive–motor video game dancing (DANCE).

Notes: Two participants perform steps on a pressure sensitive platform to the rhythm of the music. Step timing and direction are cued with arrows on a screen.

Intervention BALANCE consisted of 20 min conventional balance training and 10 min stretching in each session. Balance training included two and single leg stand exercises, either on the floor or on various types of instable surfaces (e.g., foam and air pads, ropes, etc.). The stretching part of this intervention comprised four to five exercises for the major muscle groups. Stretch- ing positions were held static for about 20–30 s.

6.2.3 Assessment of prefrontal cortex activity during walking

Functional near-infrared spectroscopy (fNIRS) is a non-invasive optical imaging technique to measure blood flow changes in the brain associated with brain activity. Within the “optical window” from approximately 700–900 nm wavelength, light readily penetrates most biological tissues, including bone.

126 Thereby oxygenated hemoglobin (HbO2) and deoxygenated hemoglobin (Hb) represent the most important light absorbers or chromophores within the optical window, besides water that remains constant (Ekkekakis, 2009). A multi distance, frequency domain fNIRS instrument (Oxiplex TS Tissue Spectrom- eter, ISS Inc., Champaign, IL, USA) was used to measure prefrontal cortex (PFC) activity during walking. This instrument emits a modulated light beam at two distinct wavelengths of 690 and 830 nm to measure absolute concentrations of HbO2 and Hb, respectively. The applied wavelengths have the advantage of low cross-talk (Boas et al., 2004). Moreover, multi distance technique, consisting of eight light emitting fibers placed at four different distances from the fiber optic detector, allowed to exclude measurement bias from skin blood flow changes during the walking experiment. Two of these sensors were tightly secured on the participants’ forehead with a bandage to avoid displacement during walking and interference with extraneous light. The two sensors were placed according to the international 10/20 EEG elec- trode system, covering the areas between Fp1–F3–F7 and Fp2–F4–F8 that correspond to the left and the right PFC, respectively (Leff et al., 2008).

6.2.4 fNIRS treadmill walking protocol

A similar treadmill protocol, as reported by Suzuki et al. (2004), was applied. Our intermittent walking protocol lasted for 9 min and was repeated two times with a recovery period of 3 min in between. Each 9-min trial started with 1 min of very slow walking at 0.2 km/h for the measurement of baseline fNIRS-values. Thereafter, eight intervals of 30 s at either preferred or fast individual walking speed (four intervals at each speed) were performed, with 30 s of rest at 0.2 km/h after each interval (Figure 2). In order to minimize anticipation effects, participants were not prepared or cued by the instructors about when the walking intervals would start and if the interval was going to be at preferred or fast walking speed. For that purpose, the display of the treadmill was covered. Participants were allowed to hold on to the handrails of the treadmill and, if necessary due to insecure gait, were secured with a climbing harness that was attached to the ceiling to avoid falls. Before participants were equipped with the fNIRS sensors for the treadmill test, individual preferred walking speed was measured with a 4-meter-walk test on the ground. Individual fast walking speed was then calculated by adding 2

127 km/h. This difference between preferred and fast walking speed was defined based on gait measurements in one of our previous studies with older adults (Eggenberger et al., 2015b). Thereafter, participants were accustomed to walking on the treadmill at different speeds for about 2–3 min to assure they were capable of walking at the two predefined speeds. For several participants, preferred walking speed as measured on the ground was subjectively rated as quite fast when performed on the treadmill and it was not possible for them to increase speed by 2 km/h. If this was the case, fast walking speed was reduced until it fitted the participant’s abilities and preferred walking speed was set at 2 km/h below. The same individual walking speeds were applied for pre- and posttest measurements.

FIGURE 2 | Intermittent fNIRS treadmill walking protocol.

Notes: The protocol was repeated two times with 3 min recovery in between. It started with 1 min of very slow walking at 0.2 km/h for the measurement of baseline fNIRS-values. Thereaf- ter, eight intervals of 30 s at either preferred or fast individual walking speed (3 and 5 km/h, re-

spectively, in this example), with 30 s of rest at 0.2 km/h, were performed.

6.2.5 Processing of fNIRS data fNIRS data were recorded with a sampling frequency of 1 Hz and exported to Microsoft Excel for processing and analysis. Data from the two sensors on the left and right PFC, respectively, were analyzed separately. Raw data were de- trended and transformed to concentration change values (Δ µM) by subtracting a 60-s moving average as a high-pass filter. This procedure allowed for direct comparison of PFC activity between individuals and groups. After

128 visual inspection of the variation range of the data, motion artifacts in HbO2- values were defined as >2.5 and < −2.5µM and were excluded from further analyses. Artifact cut-off for Hb-values was defined as >1.5 and < −1.5 µM. In five data sets with larger variations (two posttests in DANCE, one pretest and two posttests in BALANCE), artifact cut-off range for HbO2 and Hb was extended by ±1.5 µM, respectively. From the two times 9 min treadmill walking protocol, 16 one-min data blocks comprising 30 s preferred or fast walking intervals, followed by 30 s rest were available. Time-triggered averages were calculated for the eight 1-min walk/rest blocks containing preferred walking intervals and the eight blocks with fast walking intervals. This procedure was applied to minimize bias from Mayer waves that change HbO2 concentrations with about 0.1 Hz frequency and about 0.1 µM amplitude (Julien, 2006). From the two 1-min phases, that preceded the first walking interval of each 9-min trial, a baseline average value was calculated from the 10th to the 50th second. After visual inspection of the time courses of HbO2 within the 1-min walk/rest blocks in the pretest data, four timeframes were defined and averaged for further comparisons. The first timeframe was set at the initiation and acceleration of walking from the first to the seventh s of walking (t1–7), the second timeframe t10–25 represented “steady-state” walking, while the third timeframe t26–34 reflected the end of walking with deceleration and a drop of HbO2 back to baseline. The fourth timeframe t35–46 was chosen to analyze the drop of HbO2 below baseline during the rest phases.

6.2.6 Secondary assessments

Cognitive performance tasks Cognitive performance in executive functions, general cognitive ability, and information processing speed was assessed with five “paper-and-pencil” tasks. According to Miyake et al.’s model of executive functions (Miyake et al., 2000), these can be divided into three lower-level factors, which comprise shifting, inhibition, and working memory updating. Shifting was measured with the Trail Making Test part B (TMT-B) (Lezak et al., 2012), inhibition was assessed with the Stroop Word-Color Interference task (Oswald and Fleischmann, 1997), and working memory was assessed with the Executive Control task (Baller et al., 2006). Furthermore, general cognitive ability was recorded using the Montreal Cognitive Assessment (MoCA) (Nasreddine et

129 al., 2005) and information processing speed was measured with the Trail Making Test part A (TMT-A) (Lezak et al., 2012).

Short Physical Performance Battery The Short Physical Performance Battery (SPPB) assesses lower extremity functioning (Guralnik et al., 1994). This test battery comprises a balance test, a 4-meter-walk test, and a 5-chair-rises test. The standard SPPB balance test was extended with two additional levels of difficulty to avoid ceiling effects. The first additional level included a 20-s single-leg stance. One point was added to the SPPB balance test score when 10 s were reached and another point when 20 s were completed. For the second additional level, a single-leg stance with eyes closed was required to maintain for as long as possible. One point was added to the balance score for every 5 s the position was held.

Fear of falling and depression With the Falls Efficacy Scale International (FES-I) fear of falling was assessed (Yardley et al., 2005), while symptoms of depression were measured with the German version of the Geriatric depression scale (Gauggel and Birk- ner, 1999).

Training enjoyment Training enjoyment was recorded at posttest applying the German eight-item version of the Physical Activity Enjoyment Scale (PACES) (Mullen et al., 2011; Jekauc et al., 2013). The average score of the eight items was used for statistical analysis.

6.2.7 Statistical analysis

Baseline group differences of demographic characteristics, cognitive and physical performance, FES-I and GDS data were analyzed with unpaired student’s t-tests. Two-way repeated measures analyses of variance (ANOVA), with “intervention” as between-subject factor and “time,” or “hemisphere,” or “walking speed” as within-subject factors, were applied to analyze differences in PFC activity between pre- and posttest, or left and right hemisphere, or preferred and fast walking variables, respectively. Pearson’s correlation was calculated to identify associations of PFC activity changes from pre- to

130 posttest with cognitive performance changes from pre- to posttest. A paired student’s t-test was used to assess differences between baseline and experimental PFC activity levels at the fNIRS pretest. Statistical calculations were performed with IBM SPSS Statistics software for Macintosh, version 22.0 (IBM Corp., Armonk, NY, USA). A significance level of �=.05 was applied and effect size r was defined as small at r=.10, medium at r=.30, and large at r=.50 and above (Cohen, 1988).

131 6.3 RESULTS

Thirty-three participants completed the 8-week exercise training intervention (21.4% attrition). Participants’ flow is presented in Figure 3. We did not apply strict intention-to-treat analysis in favor of a clear declaration of the reasons for dropouts, as recommended by the CONSORT 2010 guidelines (Moher et al., 2010). Dropouts due to health issues were not associated to the intervention and dropouts due to personal reasons were equally distributed between groups. Therefore, nine participants who did not complete the trial were excluded from pre-/posttest analyses. Table 1 shows demographics and training characteristics (enjoyment, intensity, and compliance) of the two intervention groups.

FIGURE 3 | Participants’ flow and trial design.

Notes: Participants were randomly assigned to either DANCE or BAL- ANCE and were trained over 8 weeks three times per week for 30 min. fNIRS prefrontal cortex activity during treadmill walking, cognitive performance, Short Physical Perfor- mance Battery, fear of falling, and depression were assessed at pre- and posttest. DANCE, video game dance training; BALANCE, balance and stretching training.

132 TABLE 1 | Demographics and training characteristics.

Variable DANCE BALANCE p-value, two-tailed

N 19 14 Sex, female 12, 63.2% 9, 64.3% Age, years 72.8 (5.9) 77.8 (7.4) .039* Height, cm 169.4 (9.3) 167.1 (8.8) .477 Weight, kg 70.0 (14.9) 64.9 (11.5) .290 BMI, kg/m2 24.4 (5.0) 23.1 (2.9) .402 Education, years 13.4 (1.8) 13.6 (2.1) .830 Training compliance (24 sessions) 91.4% (7.1%) 94.3% (6.2%) .234 Training enjoyment, PACES score (1-7) 6.13 (0.76) 6.11 (0.82) .931 Training intensity perceived physically, RPE (1-10) 5.0 (2.0) 5.5 (2.1) .515 Training intensity perceived cognitively, RPE (1-10) 5.4 (2.0) 4.6 (2.4) .314 Preferred treadmill walking speed, km/h 3.8 (0.6) 3.7 (0.6) .818 Fast treadmill walking speed, km/h 5.8 (0.6) 5.7 (0.6) .818

Notes: Data are numbers or means (± standard deviation in brackets). Bold values indicate significance, * p<.05. Abbreviations: BMI, body mass index; PACES, Physical Activity Enjoyment Scale; RPE, rate of perceived exertion; DANCE, video game dance training; BALANCE, balance and stretching training.

133 6.3.1 fNIRS treadmill test

Comparison of pre- vs. posttest PFC activity Table 2 depicts statistical results of pre- vs. posttest PFC activity within three timeframes during preferred or fast walking, and in the left and right hemisphere. Figure 4 shows the time-triggered group averages for the left

PFC fNIRS sensor. HbO2 concentration was significantly reduced at posttest within t1–7 at preferred walking speed in the left and right PFC, whereas a trend for a reduced activity was evident at fast walking speed in the left PFC. Additionally, during t26–34 a significant time × intervention interaction reflected reduced HbO2 concentration in DANCE and increased HbO2 in BAL- ANCE.

TABLE 2 | Results of repeated measures ANOVA for pre- vs. posttest PFC activity.

HbO2 Main effect (time, pre vs. post) Interaction effect Variables (time × intervention)

Averaged Walking speed Hemisphere df F p, one- r F p, one- r time-frame tailed tailed t1–7 Preferred Left PFC (1, 31) 4.59 .020* .36 0.16 .344 .07 Right PFC (1, 29) 3.32 .040* .32 1.54 .112 .22 Fast Left PFC (1, 31) 2.03 .082t .25 0.28 .300 .09 Right PFC (1, 29) 0.60 .224 .14 0.03 .435 .03 t10–25 Preferred Left PFC (1, 31) 0.25 .311 .09 0.02 .450 .02 Right PFC (1, 29) 0.46 .251 .13 0.13 .360 .07 Fast Left PFC (1, 31) 0.48 .247 .12 0.04 .422 .04 Right PFC (1, 29) 0.18 .337 .08 0.09 .385 .05 t26–34 Preferred Left PFC (1, 31) 5.52 (.013) .39 1.32 .130 .20 Right PFC (1, 29) 0.68 .208 .15 0.76 .196 .16 Fast Left PFC (1, 31) 0.08 .393 .05 3.54 .035* .32 Right PFC (1, 29) 0.23 .318 .09 0.04 .418 .04

Notes: Bold values indicate trend or significance, t p<.10 trend, * p<.05, significant p-value in brackets does not correspond with one-sided hypothesis. Abbreviation: PFC, prefrontal cortex.

134

FIGURE 4 | Comparison of pre- vs. posttest PFC activity.

Notes: Graphs represent time-triggered group averages. Identical line color and style in Figures 4–

6 represent the same data. Within the first 7 s of walking (t1–7) HbO2 concentration was significantly reduced at posttest at preferred walking speed (graphs on the left side), whereas a trend for a reduced activity was evident at fast walking speed in the same timeframe (graphs on the right side). At the end of fast walking (t26–34, graphs on the right side) a significant time × intervention interaction reflected reduced HbO2 concentration in DANCE and increased HbO2 in BALANCE. DANCE, video game dance training; BALANCE, balance and stretching training, PFC, prefrontal cortex.

135 Comparison of left vs. right PFC activity No differences in left vs. right PFC activity during walking were present at pretest (statistical data are presented in Table 3). At posttest, within t10–25 and t26–34, at preferred and fast walking speed, left PFC activity was higher than right PFC activity. Two trends for significant hemisphere × intervention interaction effects were found at fast walking in t1–7 and t26–34, which represent equal left vs. right PFC activity in DANCE, and lower (t1–7) or higher (t26–34) left compared to right PFC activity in BALANCE. Figure 5 shows time triggered group averages at preferred walking for left vs. right PFC at posttest.

TABLE 3 | Results of repeated measures ANOVA for left vs. right PFC activity.

HbO2 Main effect (hemisphere, left versus Interaction effect Variables right) (hemisphere × intervention) Averaged Walking speed Test time df F p, one- r F p, one- r time-frame tailed tailed t1–7 Preferred Pre (1, 29) 0.06 .403 .05 0.36 .276 .11 Post (1, 31) 0.01 .456 .02 0.15 .353 .07 Fast Pre (1, 29) 1.00 .163 .18 0.33 .285 .11 Post (1, 31) 1.61 .107 .22 1.74 .099t .23 t10–25 Preferred Pre (1, 29) 0.80 .190 .16 0.57 .229 .14 Post (1, 31) 2.34 .069t .26 0.02 .444 .03 Fast Pre (1, 29) 1.05 .158 .19 0.25 .311 .09 Post (1, 31) 3.42 .037* .32 0.01 .454 .02 t26–34 Preferred Pre (1, 29) 0.01 .457 .02 0.06 .403 .05 Post (1, 31) 3.58 .034* .32 0.07 .398 .05 Fast Pre (1, 29) 0.02 .451 .02 0.30 .293 .10 Post (1, 31) 2.96 .048* .30 2.53 .061t .27

Notes: Bold values indicate trend or significance, t p<.10 trend, * p<.05; PFC, prefrontal cortex.

136 FIGURE 5 | Comparison of left vs. right PFC activity.

Notes: Graphs represent time- triggered group averages. Identical line color and style in Figures 4–6 represent the same data. Within the timeframes t10–25 and t26–34 left PFC activity was higher than right PFC activity in DANCE and BAL- ANCE. No differences in left vs. right PFC activity during walking were present at pretest (graphs not shown). DANCE, video game dance training; BALANCE, balance and stretching training, PFC, prefrontal cortex.

137 Comparison of preferred vs. fast walking PFC activity PFC activity at pretest was equal for preferred and fast walking speed (for statistical data see Table 4). At posttest, HbO2 levels were higher at fast compared to preferred walking speed within t1–7 in the right PFC, and within t10–25 in the left and right PFC. Time triggered group averages in the right PFC for preferred vs. fast walking at posttest are presented in Figure 6.

TABLE 4 | Results of repeated measures ANOVA for preferred vs. fast walking speed PFC activity.

HbO2 Main effect (walking speed, pre- Interaction effect Variables ferred versus fast) (speed × intervention) Averaged Hemisphere Test time df F p, one- r F p, one- r time-frame tailed tailed t1–7 Left PFC Pre (1, 31) 0.00 .476 .01 0.53 .236 .13 Post (1, 31) 0.01 .464 .02 1.39 .124 .21 Right PFC Pre (1, 29) 0.56 .231 .14 0.15 .352 .07 Post (1, 31) 3.00 .047* .30 0.27 .305 .09 t10–25 Left PFC Pre (1, 31) 1.41 .123 .21 0.04 .424 .03 Post (1, 31) 6.22 .009** .41 0.32 .288 .10 Right PFC Pre (1, 29) 0.10 .379 .06 0.09 .385 .05 Post (1, 31) 1.93 .088t .24 0.84 .184 .16 t26–34 Left PFC Pre (1, 31) 0.10 .376 .06 0.04 .418 .04 Post (1, 31) 2.17 (.075) .26 0.83 .185 .16 Right PFC Pre (1, 29) 0.00 .492 .00 0.58 .226 .14 Post (1, 31) 2.28 (.071) .26 1.13 .149 .19

Notes: Bold values indicate trend or significance, t p<.10 trend, * p<.05, ** p<.01, significant p-values in brackets do not correspond with one-sided hypothesis; PFC, prefrontal cortex.

138 FIGURE 6 | Comparison of preferred vs. fast walking PFC activity.

Notes: Graphs represent time-triggered group averages. Identical line color and style in Figures 4–6 represent the same data. Within the timeframes t1–7 and t10– 25 HbO2 levels were higher at fast compared to preferred walking speed in DANCE and BALANCE. PFC activity at pretest was equal for preferred and fast walking speed (graphs not shown). DANCE, video game dance training; BALANCE, balance and stretching training, PFC, prefrontal cortex.

Correlation of pre-/posttest changes in PFC activity and cognitive performance Table 5 shows correlation coefficients for pre- to posttest changes in PFC activity and cognitive performance. Reduced time (improvement) to complete

TMT-A and TMT-B correlated with reduced HbO2 activity in the left PFC at preferred and fast walking in t1–7. Increased performance in TMT-A also correlated with reduced right PFC activity at preferred walking in t10–25 and at fast walking in t26–34. A reduction in Stroop word-color interference time (improvement) correlated with elevated right PFC activity at fast walking in t1–7, while reduced activity in this HbO2 variable was associated with increased performance in the Executive Control task. Additionally, an improvement in the Executive Control task correlated also with a reduction in right PFC activity at fast walking in t10–25.

139 TABLE 5 | Results of Pearson’s correlation analysis between pre-/posttest changes in cognitive performance and PFC activity.

HbO2 TMT-B (EF, Stroop Word- Executive Con- TMT-A (pro- Variables shifting) Color trol (EF, working cessing (EF, inhibition) memory) speed) Averaged Walking Hemisphere r p, r p, r p, r p, time-frame speed two- two- two-tailed two- tailed tailed tailed

t t t1–7 Preferred Left PFC .34 .051 .16 .366 -.08 .639 .33 .062 Right PFC -.05 .789 .23 .208 -.29 .119 .04 .843 Fast Left PFC .39 .023* -.15 .391 -.06 .759 .37 .036* Right PFC -.03 .879 -.43 .016* -.33 .069t -.04 .814 t10–25 Preferred Left PFC .13 .465 -.09 .616 .08 .673 .03 .877 Right PFC -.03 .891 -.15 .417 -.18 .337 .31 .089t Fast Left PFC .11 .560 -.12 .517 -.23 .207 -.17 .356 Right PFC .01 .970 .13 .494 -.35 .056t .05 .788 t26–34 Preferred Left PFC -.13 .474 -.12 .502 .09 .636 -.18 .325 Right PFC -.05 .788 -.12 .507 .11 .547 .00 .996 Fast Left PFC -.01 .943 -.19 .296 .02 .908 -.03 .883 Right PFC .00 .988 .12 .522 -.06 .753 .50 .004**

Notes: Bold values indicate trend or significance, t p<.10 trend, * p<.05, ** p<.01; EF, executive function; TMT-A, Trailmaking Test A; TMT-B, Trailmaking Test B; PFC, prefrontal cortex.

Comparison of baseline vs. experimental PFC activity at pretest All 42 participants who performed the fNIRS pretest session where included in this comparison of HbO2 baseline vs. experimental values. Differences between the baseline HbO2-values and the experimental walking or rest interval HbO2-values were evident in almost all variables, except in t26–34 at fast walking as shown in detail in Table 6.

140 TABLE 6 | Results of paired t-test for baseline vs. experimental PFC activity at pretest.

HbO2 Variables Paired t-test Averaged time- Walking Hemisphere df t p, one-tailed r frame speed t1–7 Preferred Left PFC (1, 40) -2.27 .014* .34 Right PFC (1, 38) -1.98 .027* .31 Fast Left PFC (1, 40) -1.89 .033* .29 Right PFC (1, 38) -2.48 .009** .37 t10–25 Preferred Left PFC (1, 40) -3.20 .001** .45 Right PFC (1, 38) -1.92 .031* .30 Fast Left PFC (1, 40) -3.26 .001** .46 Right PFC (1, 38) -1.95 .029* .30 t26–34 Preferred Left PFC (1, 40) -1.58 .061t .24 Right PFC (1, 38) -1.98 .027* .31 Fast Left PFC (1, 40) -0.85 .200 .13 Right PFC (1, 38) -0.52 .302 .08 t35–46 Preferred Left PFC (1, 40) 4.77 < .001*** .60 Right PFC (1, 38) 2.52 .008** .38 Fast Left PFC (1, 40) 5.17 < .001*** .63 Right PFC (1, 38) 2.59 .007** .39

Notes: Bold values indicate trend or significance, t p<.10 trend, * p<.05, ** p<.01, *** p<.001; PFC, prefrontal cortex.

6.3.2 Secondary assessments

Baseline cognitive performance showed a significant group difference in the Stroop Word-Color Interference task (p=.035), whereas no baseline differences were present in the other tasks (TMT B p=.394, Executive Control p=.630, MoCA p=.383, TMT A p=.386). Baseline physical performance was not different between groups (SPPB total score p=.678, 4-meter-walk p=.839, 5-chair-rises p=.740, Extended Balance test p=.811). Baseline values showed a trend to a significant difference between groups in FES-I (p=.098 trend) and no significant difference in GDS (p=.104). Performance development and statistical data of two-way repeated measures ANOVA from pre- to posttest are presented in Table 7.

141 TABLE 7 | Performance and statistical data of secondary assessments.

Variable DANCE BALANCE ANOVA Time Mean SE Mean SE Effect F (1, 31) p, two- r tailed

Executive functions Trail Making B (s) Pre 90.4 7.4 99.6 7.1 t 12.27 .001** .53 Post 74.9 6.5 79.6 5.6 t×I 0.19 .665 .08 Executive Control Pre 9.79 1.29 8.93 1.09 t 1.88 .180 .24 (items) Post 10.84 1.10 10.57 0.88 t×I 0.09 .766 .05 Stroop Word-Color Pre 41.6 2.0 49.6 3.2 t 28.18 < .001*** .69 (s) Post 39.6 2.0 42.1 2.5 t×I 9.52 .004** .48

General cognition MoCA (score) Pre 25.95 0.58 26.64 0.48 t 9.43 .004** .48 Post 27.26 0.59 27.71 0.45 t×I 0.10 .755 .06

Processing speed Trail Making A (s) Pre 39.8 3.5 44.3 3.6 t 10.34 .003** .50 Post 34.3 2.6 38.5 3.0 t×I 0.01 .927 .02

Physical functioning SPPB (score) Pre 11.58 0.16 11.43 0.36 t 0.79 .382 .16 Post 11.47 0.30 11.86 0.14 t×I 2.14 .153 .25 4-meter-walk (s) Pre 3.4 0.1 3.5 0.1 t 0.03 .858 .03 Post 3.4 0.1 3.5 0.2 t×I 0.02 .892 .02 5-chair-rises (s) Pre 9.0 0.5 9.3 0.6 t 10.51 .003** .50 Post 7.8 0.4 8.3 0.5 t×I 0.16 .695 .07 Extended Balance Pre 4.95 0.31 5.07 0.43 t 2.84 .102 .29 (score) Post 4.95 0.42 5.79 0.33 t×I 2.84 .102 .29

Fear of falling and depression FES-I (score) Pre 20.84 1.21 18.57 0.54 t 2.97 .095t .30 Post 19.05 0.53 18.21 0.38 t×I 1.32 .259 .20 GDS (score) Pre 4.47 0.76 2.86 0.59 t 0.00 .979 .01 Post 4.53 0.78 2.79 0.55 t×I 0.03 .864 .03

Notes: Bold values indicate trend or significance, t p<.10 trend, * p<.05, ** p<.01, *** p<.001; t, time; t×I, time×intervention interaction; SE, ± standard error of measurement; SPPB, Short Physical Perfor- mance Battery; FES-I, Falls Efficacy Scale International; GDS, Geriatric Depression Scale; DANCE, video game dance training; BALANCE, balance and stretching training.

142 6.4 DISCUSSION

In this study, we investigated the impact of two different exercise training interventions on fNIRS brain activity during walking and on executive functions in older adults. Thereby, one intervention combined cognitive and motor training simultaneously in an exergame (interactive video game dancing, DANCE), whereas the other intervention consisted of exclusively motor training (balance and stretching exercises, BALANCE). Our results demonstrated (1) that both interventions reduced left and right hemispheric PFC oxygenation during the acceleration of walking, while DANCE showed a larger reduction at the end of the walking compared to BALANCE in the left PFC, and (2) that the exercise training-induced modulations in PFC oxygenation were associated with improved executive functions. The present study provides novel and important insight into the relation and mediation between exercise training, brain function during walking, and cognition in older adults.

6.4.1 Reduced prefrontal brain activity during walking after exercise training

The fNIRS treadmill walking experiment supports our first hypothesis that 8 weeks of DANCE or BALANCE training induce a reduction of prefrontal brain activity during walking. This was particularly evident in the acceleration phase during the first 7 s of walking in both hemispheres and in both intervention groups. To date, no other study that we are aware of has longi- tudinally investigated exercise training related alterations of PFC activity during walking. Nevertheless, our results convene with previous findings from cognitive training studies. For instance, Erickson et al. (2007b) demonstrated reduced activity in several brain regions after cognitive dual-task training in older adults. Additionally, in our study, DANCE showed an earlier reduction of left PFC activity at the end of the 30-s walking interval after the training intervention. This observation is comparable to a cross-sectional fNIRS study that reported an earlier decrease of PFC activity already during the performance of a cognitive–motor dual-task in young adults, whereas PFC activity in the older adults remained elevated about 10 s beyond the end of the task (Ohsugi et al., 2013). Thereby, brain activity in the old adults

143 resembles the untrained state of our participants at pretest, whereas brain activity in the young adults is similar to our posttest results in DANCE.

Traditionally balance training is another option to train motor skills in older adults and effect on brain properties and cognitive performance (Voelcker- Rehage et al., 2011). Training programs that contain balance components counteract risk factors for falls and restore balance control in older adults. However, the positive effects cannot be exclusively attributed to balance exercise, because all the studies taken into account used a combination of strength and balance training (Taube et al., 2008). Moreover, the authors of this review concluded that the brain adaptions to balance training are rather task-specific and hitherto little is known about the influence of balance training influencing brain plasticity. The findings of our study warrant further investigations about the relation between conventional types of balance training and the effects on brain functioning during functional movement tasks.

Generally, decreased activity of a particular brain area may represent decreased use and, therefore, increased efficiency according to the reviews by Lustig et al. (2009) and Grady (2012). Additional brain activity in older compared to younger adults was suggested to reflect a compensatory mechanism (compensation hypothesis) (Cabeza et al., 2002; Reuter-Lorenz and Cappell, 2008) to improve performance in a specific task and such over-recruitment might as well be associated with less efficient use of neural resources (Grady, 2012). This relation has not only been observed between young and old adults but also between higher fit and lower fit old adults. For instance, Voelcker- Rehage et al. (2010) reported that both high physical and high motor fitness were associated with reduced fMRI signals in frontal and other brain regions during cognitive tasks, and Harada et al. (2009) found that frontal brain activity in fast walking was higher in individuals with lower gait capacity as assessed with fNIRS. Additionally, decreased activity of prefrontal areas could also have resulted from a shift from controlled gait to more automatic gait or a shift from the indirect to the direct locomotor pathway, respectively (Hamacher et al., 2015). While the indirect locomotor pathway regulates gait via prefrontal cortex, premotor area, supplementary motor area, and basal ganglia, the direct pathway comprises primary motor cortex (M1), cerebel- lum, and spinal cord (la Fougere et al., 2010; Zwergal et al., 2012). Similar observations of reduced brain activity have also been reported in studies with highly skilled, professional pianists (Jäncke et al., 2000) or soccer players

144 (Naito and Hirose, 2014) that both showed lower recruitment of motor areas in which they have richer sensory-motor experiences (finger or foot movements, respectively) compared to amateurs. The authors explained their findings with highly efficient motor control processes due to over-years training of a motor skill, which led to a smaller number of neurons needed to perform the particular skill. Together, the effects in both interventions and particularly in DANCE might reflect an adaptation toward the function of a younger or more trained brain.

6.4.2 Hemispheric asymmetry in prefrontal brain activity after exercise training

The comparison of brain activity in the left vs. the right PFC showed that hemispheric differences only became obvious at the end of both 8-week training interventions at posttest. At this time point of measurement, left PFC activity was larger than right PFC activity at preferred and fast walking speed during the “steady state” walking phase (t10– 25) and also at the end of the walking interval (t26–34). To the best of our knowledge and according to the review by Ekkekakis (2009), prefrontal hemispheric asymmetry has not yet been investigated with fNIRS in relation to physical exercise training. Nonetheless, our observation is in line with a cognitive training fMRI study that found increased posttest prefrontal hemispheric asymmetry during dual- task conditions (Erickson et al., 2007b). In particular, the authors reported an elevation of left ventrolateral PFC activity and a reduction of right ventrolateral PFC activity in the old adults after training and a concomitant reduction in age differences compared to young adults. Similarly, Bergerbest et al. (2009) demonstrated in their fMRI study on implicit memory (repetition priming) an initial bilateral PFC activity in older adults and left lateralized activity in younger adults. These initial activity patterns were followed by repetition related activity reductions, which were smaller for the older compared to the younger adults in the left PFC but larger in the right PFC. They assumed that the initial right PFC activity in the elderly would compensate for a lower aging-related left hemispheric PFC activity. These and our own findings correspond with the complementary hypothesis (Colcombe et al., 2005), proposing that in older adults bilateral brain activity or a reduction of asymmetric brain activity is not generally related to better performance in a

145 certain task, as it is suggested by the compensation hypothesis (Cabeza et al., 2002; Reuter-Lorenz and Cappell, 2008). Moreover, Lustig et al. (2009) hypothesized in their comprehensive review that compensatory effects seen in older adults, such as the aforementioned activation of additional contrala- teral brain regions for better performance, would typically occur in single- session studies, whereas after several sessions of training, brain activity patterns would become more similar to young adults. Based on this framework, our finding of exercise training-induced increased hemispheric asymmetry during treadmill walking, could be explained with less need for compensation through right prefrontal activity and might represent another adaption toward the function of a younger adults’ brain.

6.4.3 Gait speed related differences in prefrontal brain activity after exercise training

Gait speed related differences in PFC activity were not found at pretest but only at posttest, which converges with the previously discussed hemispheric differences. Posttest differences were evident in the right PFC during acceleration (t1–7) and in both PFC hemispheres during “steady state” walking (t10–25). Our pretest result is consistent with a cross-sectional study in old adults (mean age 63 ± 4 years) that also did not find differences in PFC oxygenation related to gait speeds comparable with those in our study (50 and 30% intensity of individual heart rate response, respectively) (Harada et al., 2009). Similarly, two fNIRS studies with young adults did not find changes in PFC activity when preferred walking speed was increased by either 20% (Meester et al., 2014) or from 3 to 5 km/h (Suzuki et al., 2004). However, the latter study found significantly elevated PFC activity when the young participants were running at 9 km/h, and in the aforementioned study by Harada et al. (2009), PFC activity increased in the older adults when they walked at 70% intensity. It can, therefore, be concluded that changes in PFC activity in young and old adults only occur when the difference in locomotion speed between conditions is large enough.

The question, however, remains what mechanism could have led to gait speed related differences in prefrontal oxygenation at posttest in our study? Several cross-sectional studies with young and old adults compared brain activity of

146 normal vs. challenging walking tasks, including different dual-task walking conditions, walking in dim lighting, negotiating obstacles, etc. (Holtzer et al., 2011; Clark et al., 2014; Meester et al., 2014; Mirelman et al., 2014). They consistently demonstrated increased PFC activity during challenging walking compared to normal walking. Noteworthy, goal-directed locomotion, such as more complex walking tasks and dual-task walking, are associated with increased activation of the indirect locomotor pathway via prefrontal areas (Hamacher et al., 2015). Interestingly, it was found that PFC activity in complex walking tasks was elevated more in older adults with better gait quality (Clark et al., 2014) or in young adults compared to old adults (Holtzer et al., 2011), referring to under-recruitment in the older or less fit old adults, respectively. In correspondence with our study these findings may reflect the transition from an untrained to a trained state. In contrast, the study by Harada et al. (2009) observed that PFC activity during fast walking (at 70% intensity) was elevated in the elderly with lower gait capacity which refers to over-recruitment. At first sight this looks like a discrepancy, however, might be explained by recent findings from Kennedy et al. (2015) and earlier studies (Mattay et al., 2006; Reuter-Lorenz and Cappell, 2008). These authors demonstrated that compensatory brain activity (over-recruitment) is effective in relation to lower task difficulty, while with increasing task complexity older adults reach a resource ceiling leading to less activity compared to the younger adults (under-recruitment). Moreover, Clark et al. (2014) and Holtzer et al. (2011) proposed that the ability to increase PFC activity in challenging walking situations would be an important mechanism to optimize performance.

These conceptions are in line with the frontal lobe hypothesis of aging (West, 1996) and with the cognitive reserve theory which assumes that younger adults increase brain activity by a larger degree to cope with elevated cognitive task difficulty (Stern, 2009). Notably, the age-related neural modulation patterns of functional over- and under-recruitment are emerging especially within the transition from middle-aged to old adults (Kennedy et al., 2015). Moreover, these functional effects of aging are mirroring age-related structural effects proposed by the “last-in-first-out” hypothesis where late matur- ing brain regions decline first in later life (Raz and Kennedy, 2009; Tamnes et al., 2013; Bender et al., 2015) and explain gait disturbances (Scherder et al., 2011). We conclude, therefore, that in our study the ability to differentiate

147 PFC activity related to walking speed was either attenuated due to reduced availability or underutilization of PFC resources at pretest when the older participants were in an untrained state. The walking speed related differences in PFC activity that we found at posttest may, therefore, again reflect a mechanism of how the older adults’ brain adapted its function toward a young adult-like brain.

6.4.4 Correlated changes in prefrontal brain activity and cognitive performance

Correlation analysis is supporting our second hypothesis and revealed associated exercise training-induced changes in prefrontal brain activity and improvements in different behavioral measures of executive function and processing speed. This association could be explained by the fact that the PFC represents a brain area that is involved in both locomotion (Suzuki et al., 2004) and in cognitive executive functioning (Shibuya-Tayoshi et al., 2007). The results of our correlation analysis extend recent findings from several exercise training studies that employed fMRI to assess brain functional or structural adaptations alongside with cognitive performance measures (Voss et al., 2010; Erickson et al., 2011; Voelcker-Rehage et al., 2011; Liu-Ambrose et al., 2012; Voss et al., 2013; ten Brinke et al., 2015). For instance, Voelcker- Rehage et al. (2011) found training-specific brain functional adaptations during the performance of an executive function task (flanker task) after both aerobic and coordination training. These adaptations were accompanied by increased cognitive executive performance. Similarly, Liu-Ambrose et al. (2012) reported correlating brain functional and cognitive plasticity after a 12-month strength training intervention. Together, it seems that different types of exercise training are able to induce cognitive improvements that are mediated by brain functional and structural adaptations in elderly persons. Nonetheless, previous studies focused on brain functional adaptations during the performance of cognitive tasks that can be employed with fMRI measurements, whereas our study provides first insight to exercise training-induced brain functional adaptations during a challenging treadmill walking task. These adaptations were correlated with cognitive improvements in the elderly participants.

148 In the present study, improved cognitive performance was mainly (in eight of nine instances) related to a reduction in PFC activity during walking. This finding is consistent with Erickson et al. (2007a) who performed a cognitive training intervention including five 1-h sessions. They reported a training- induced reduction of brain activity in most regions that displayed activity during cognitive dual-tasking. One exemption was a training-related increase in activity of the bilateral dorsolateral PFC that was correlated with better performance. The authors explained this phenomenon with a switched strategy to perform dual-task-related activities resulting in the recruitment of other brain areas. A similar mechanism could have become evident in our own study where only one task of executive function (the Stroop Word-Color Interference task) was associated with a training-induced increase in (right PFC) activity. Similarly, an fNIRS study with older adults by Hyodo et al. (2012) demonstrated enhanced right frontopolar activity during the Stroop task when performed after an acute bout of 10 min ergometer cycling at moderate intensity. Enhanced right frontopolar brain activity was also associated with improved Stroop Interference results. Furthermore, another recent fNIRS study by Dupuy et al. (2015) found that young and old women with higher aerobic fitness showed increased brain activity in the right inferior frontal gyrus during the Stroop task.

Interestingly, in the three tests assessing different components of executive function (TMT-B, Stroop Word-Color, and Executive Control tasks) correlation was evident exclusively with either left or right hemispheric PFC activity changes, whereas processing speed (TMT-A) was correlated with bilateral PFC activity reductions. Measures of processing speed were shown to be associated with age-related global white matter deterioration (fractional anisot- ropy), but not to specific brain regions (Kuznetsova et al., 2015). Moreover, processing speed was associated with myelin integrity, particularly in the prefrontal lobe and the genu of corpus callosum which represent late-mye- linating regions in brain development and are, therefore, highly vulnerable to breakdown in normal aging (Lu et al., 2011). These studies may explain why in the present study improvements in TMT-A were not specifically correlated with either left or right hemispheric PFC activity reductions, but rather to bilateral adaptations.

149 6.4.5 Strengths and limitations

Strengths of the present study were the experimental design of the fNIRS treadmill protocol where we included and averaged several repetitions of the two walking conditions in order to minimize bias from potential learning or accommodation effects. Additionally, we used a multi distance fNIRS instrument, which eliminates measurement bias from skin blood flow changes (Jung et al., 2015). Likewise, some researchers questioned the outcome of previous studies that applied other fNIRS methods and concluded that in fact not PFC oxygenation was measured but rather superficial blood flow in skin and muscle caused by cardiovascular activity during exercise (Miyazawa et al., 2013; Jung et al., 2015). The following limitations have to be acknowl- edged. We focused fNIRS assessments on left and right PFC activity during walking due to its association with cognitive executive functions. However, for future investigations it would be interesting to obtain data from other brain areas related to walking and their functional adaptations to exercise training. Furthermore, our correlation analyses of changes in brain function during walking and cognitive behavioral outcomes do not necessarily imply causality. Nonetheless, the outcome of this correlation still provides valuable information since the changes in the variables were observed between two distinct time points before and after a defined exercise training program, whereas in cross-sectional studies it is unknown at what time in the past the changes occurred (Salthouse, 2011). What can be seen as another possible limitation of our study is the difference in age between the two intervention groups. However, it is well-known that cognitive decline cannot be interpreted based on chronological aging per se but is rather affected by several biological and physical health domains as well as environmental and genetic factors (MacDonald et al., 2011; DeCarlo et al., 2014; Walhovd et al., 2014). Therefore, and considering that the physical and cognitive baseline values were not different, with the exception of one cognitive task, we argue that the age difference between the two intervention groups did not have an impact on our results. Finally, a passive control group might have helped to control bias from repeated testing. However, due to the aforementioned fNIRS test procedure that included many averaged repetitions of the same test condition at pre- and posttest assessment, we could reduce this sort of bias to a minimum.

150 6.5 CONCLUSIONS

The present study demonstrated three mechanisms of exercise training-induced functional brain plasticity during treadmill walking in elderly participants who underwent 8 weeks of interactive cognitive–motor video game dancing or conventional balance training. These mechanisms comprise (1) a bilateral reduction in prefrontal brain activity at preferred and fast locomotion speed (with larger effects in the video game dance group), (2) an increase in hemispheric PFC activity asymmetry, and (3) an increased differentiation in PFC activity related to walking speed. The adaptations resemble more trained or young adult-like brain functions as observed in previous cognitive training interventions and cross-sectional fMRI and fNIRS studies on brain activity in cognitive and walking tasks, respectively. The prefrontal adaptations were correlated with improved performance in executive functions and processing speed. These novel findings imply that exercise training is able to reduce the need of prefrontal resources of executive function and attention involved in challenging treadmill walking. We speculate that the elderly might benefit from these additional cognitive resources to focus their attention on other processes while walking. This would be of practical importance in attention demanding real-life situations such as crossing streets or walking while talking and could potentially reduce the risk of falling. Future investigations are warranted that should focus on additional brain areas involved in locomotion and that should include other types of exercise training and challenging walking conditions in order to substantiate or refute the presented findings.

Author contributions PE: study preparation and conception, participants’ recruitment, data acquisition, statistical analysis, data interpretation, drafting manuscript. MW: study conception, technical advice on fNIRS, data interpretation, critically revising manuscript. MS: study preparation and conception, participants’ recruitment, training instruction, data acquisition, critically revising manuscript. EDB: study conception, data interpretation, critically revising manuscript. All authors read and approved the final manuscript.

151 Acknowledgments This work was supported by the Zürcher Kantonalbank within the framework of sponsoring of movement sciences, sports and nutritional sciences at ETH Zurich. Zürcher Kantonalbank had no influence on the study design and the analyses presented in this paper, had no access to the data, and did not contribute to this manuscript in any way. The authors would like to thank PD Dr. med. Thomas Münzer, chief physician, and the management of Geriat- rische Klinik St.Gallen, Switzerland, for supporting the study and providing room for training and data acquisition. Dr. Vera Schumacher, Department of Psychology and Gerontology, University of Zurich, Switzerland, provided valuable advice on cognitive test selection. We thank our postgraduate students Nadine Jenni and Jana Bucher for instructing trainings and helping with data acquisition. Furthermore, we appreciated the kind support of the team of physiotherapists at Geriatrische Klinik St.Gallen. Last but not least, we would like to thank all participants for their enthusiasm, kindness, and patience during the training and testing sessions.

Conflict of interest statement The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright notice This work was initially published by Frontiers and licensed under Creative Commons Attribution 4.0 International (CC BY 4.0) https://creativecom mons.org/licenses/by/4.0/, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. The following changes were applied to the original manuscript: the term “cognitive–physical” was replaced by the term “cognitive–motor” to use consistent terminology throughout the manuscript of this doctoral thesis.

152 7

Thesis conclusions

153 7.1 MAIN FINDINGS

This thesis provides novel and valuable insight into the effects of interactive simultaneous cognitive–motor training as a means to improve cognitive performance and brain function, physical performance, gait, and prevent falls in older adults. The main findings from the three studies are in accordance with the initial hypotheses and include the following:

Study 1: About every third (35.6%) older person at the age of 70–79 years and almost three-quarters (73.8%) of persons ≥80 years cannot walk faster than 1.2 m/s, which is required to cross streets safely within the green–yellow phase of pedestrian lights, under cognitively challenging conditions.

Study 2a: First, the two interactive simultaneous cognitive–motor programs were partially advantageous to boost performance in two measures of executive function (switching attention and working memory), thereby video game dancing resulted in transfer to an untrained cognitive domain (working memory); and second, cognitive performance, including executive functions, long-term visual memory (episodic memory), and processing speed, was maintained until 1-year follow-up.

Study 2b: First, the two interactive simultaneous cognitive–motor programs resulted in a significant advantage in dual-task costs of walking compared to the exclusively physical program but not in any other gait variables; second, the two simultaneous cognitive–motor interventions led to different training- specific adaptations in the rhythm and variability domains of gait; and third, the three training programs very effectively reduced fall frequency for ~77%.

Study 3: First, both the video game dancing and the balance interventions reduced left and right hemispheric prefrontal cortex (PFC) oxygenation during the acceleration of walking, while video game dancing showed a larger reduction at the end of the walking compared to balance training in the left PFC; and second, the exercise training-induced modulations in PFC oxygenation were associated with improved executive functions.

154 7.2 MAIN CONCLUSIONS

From study 1 it can be concluded that the fitness status of many older adults is not appropriate to safely encounter the requirements for pedestrians in urban areas, which reinforces the need for regular cognitive and physical training in the older population (Kuh et al., 2014) to keep up with the demands of daily life in the community. It is proposed that fast speed dual-task walking is more reflective of the time pressure and cognitive challenge at real pedestrian lights than the previously applied assessments of single- or dual-task walking at preferred speed. This novel approach led to a more conservative estimation of the proportion of older adults that are unable to walk as fast as 1.2 m/s compared to a previous investigation assessing dual-task walking at preferred speed (Donoghue et al., 2016). Nonetheless, this proportion is still alarmingly high, given the fact that slowed walking speed is related to various adverse health outcomes (Abellan van Kan et al., 2009; Montero-Odasso et al., 2012; Gonzales et al., 2016; Liu et al., 2016).

Study 2a leads to the conclusion that multicomponent simultaneous cognitive–motor training programs have the potential to boost particular executive functions (including shifting attention and working memory) in healthy older adults more, compared to an exclusively physical multicomponent program. Nonetheless, broad cognitive performance improvements were achieved through both, combined cognitive–motor and exclusively physical training. Notably, performance levels in executive functions, long-term visual memory (episodic memory), and processing speed were maintained over 1 year after all three programs. The novel training concepts of simultaneous cognitive– motor activity tended to be enjoyed more by the older adults than traditional training and led to training specific as well as to transfer adaptations in cognition. These findings are important since executive functions, episodic memory, and processing speed are particularly affected by aging-related decline (Deary et al., 2009). Therefore, it is suggested that simultaneous cognitive–motor training should be integrated in training programs aiming to improve cognition in the elderly. Such programs may potentially counteract the large prevalence of cognitive impairments and decline in the elderly, inherently leading to more independence and a better quality of life.

The findings from Study 2b allow to conclude that an advantage of the simultaneous cognitive–motor training programs became evident in dual-task costs

155 of gait variability. Nonetheless, each of the three multicomponent programs efficiently increased performance in most other gait variables. Gait performance was partly retained over the relatively long period of 1 year after all three programs, with some attenuation in fast walking speed and gait variability. Therefore, it is suggested that these two variables may serve as early indicators of functional fitness decline and increased fall risk in clinical settings. In general, the two novel training concepts of simultaneous cognitive– motor training and the exclusively physical exercise program displayed similarly great potential to counteract age-related decline of physical functioning in older persons, while possible advantages of simultaneous cognitive–motor interventions are well worth further investigation.

Based on the results of study 3 it can be concluded that three mechanisms of exercise training-induced functional brain plasticity during treadmill walking exist. These mechanisms comprise first, a bilateral reduction in prefrontal brain activity at preferred and fast locomotion speed (with larger effects in the video game dance group), second, an increase in hemispheric PFC activity asymmetry, and third, an increased differentiation in PFC activity related to walking speed. Interestingly, the adaptations resemble more trained or young adult-like brain functions as observed in previous cognitive training interventions and cross-sectional functional magnetic resonance imaging (fMRI) and functional near-infrared spectroscopy (fNIRS) studies on brain activity in cognitive and walking tasks, respectively. The prefrontal adaptations were correlated with improved performance in executive functions and processing speed. These novel findings imply that exercise training is able to reduce the need of prefrontal resources of executive function and attention involved in challenging treadmill walking. It can be speculated that the elderly might benefit from these additional cognitive resources to focus their attention on other processes while walking. This would be of practical importance in attention demanding real-life situations such as crossing streets or walking while talking and could potentially reduce the risk of falling.

156 7.3 IMPLICATIONS FOR FUTURE RESEARCH

The promising findings presented in this thesis, illuminating the effects of interactive simultaneous cognitive–motor training on cognition and brain function in older adults, warrant further investigations. Particularly, the seminal findings on training-induced brain plasticity, measured with fNIRS under challenging walking (study 3), raise further questions:

How would the function of other brain regions and networks be modulated during challenging walking after similar training interventions?

How would other cognitive–motor training modalities affect brain functional outcomes during real-life activities?

Would a transfer to other real-life activities, e.g. crossing a (virtual) street, be evident as well?

Would the observed effects also be evident in other populations of older adults, especially in those with mild cognitive impairments (MCI) that are at risk for dementia?

The problem that needs to be solved in order to be able to address these questions is the technical feasibility and reliability of measuring the activity of larger brain areas during locomotion. As outlined in chapter 6 (section 6.2.3), fNIRS methodology provides an interesting means to assess brain activity during locomotion. fNIRS is a non-invasive optical neuroimaging technique that has been discovered in 1992 (Ferrari and Quaresima, 2012) and is based on the theory of neurovascular coupling. The latter is the result of neural and glial activity enhancing blood flow in the respective brain area to provide en- ergy for the neuronal tissue (Huneau et al., 2015). Hence, fNIRS technology enables an indirect assessment of brain activity based on hemodynamic changes in the cortical tissue, similarly to fMRI (Herold et al., 2017). The neurovascular coupling includes a temporal delay of the hemodynamic response of ∼2 to 5 seconds (Herold et al., 2017), reaching a peak typically about 6 seconds after stimulus onset (Cui et al., 2010). fNIRS has several advantages over other neuroimaging techniques, e.g. electroencephalography (EEG) and functional magnetic resonance imaging (fMRI): it is portable, potentially wearable, easy to apply, and the costs for

157 purchase and operation are relatively low (Scholkmann et al., 2014). Com- pared to fMRI, fNIRS has a higher temporal resolution (10 Hz vs. 0.5–1 Hz) and it measures concentration changes in both HbO2 and Hb simultaneously, which was shown to be useful to remove motion artefacts (Cui et al., 2010). A recent review by Herold et al. (2017) recommended to apply multi-distance measurement technique in order to improve accuracy of fNIRS; this technique has been used in the study presented in this thesis but not in the 55 other studies that were reviewed in the aforementioned review.

The limitations of fNIRS technology include its limited penetration depth (1– 2 cm) which only allows to assess changes in superficial cortical areas, as well as the limited spatial resolution of ~1 cm (Tong and Frederick, 2010). Addi- tionally, strenuous physical tasks may affect fNIRS signals due to systemic vascular changes (Perrey, 2008). Furthermore, it was pointed out that cur- rently no standardized fNIRS procedures and signal processing techniques exist to measure cortical activity during locomotion (Herold et al., 2017; Vitorio et al., 2017).

The possibility of combining different neuroimaging techniques enables to gather more detailed information, for instance about the interplay of the elec- trophysiological and the hemodynamic or metabolic signals (Scholkmann et al., 2014). Thereby, recent advances in simultaneous EEG and fNIRS measurement systems allow to cover additional brain areas to assess effects in large-scale functional brain network connectivity, while providing greater spatial and temporal resolution. EEG detects very brief processes in the range of 100 ms but suffers related to spatial resolution. fNIRS provides good local- ization but the comparably slow vascular response limits temporal resolution. Hence, when both methods are applied simultaneously they provide complementary information about neuronal and hemodynamic aspects of brain activation. Due to their tolerance to participants’ motion artefacts and portabil- ity, experiments can be performed under real-life conditions. Temporal resolution of both methods is much greater than in fMRI assessments (Wallois et al., 2012; Zama and Shimada, 2015). However, these measurement systems have not yet been applied in training studies with older adults due to techno- logical limitations. Such measurement systems could expand the presented findings, that were limited to the prefrontal cortex, to other brain areas and networks involved in cognition and locomotion, such as the default mode, the frontal executive, and the fronto-parietal network, as well as the primary

158 motor cortex (Colcombe et al., 2004; Voss et al., 2010; Huang et al., 2016; Hsu et al., 2017). Moreover, the concomitant assessment of cognitive performance changes might allow to investigate mediating effects of functional brain network adaptations to link physical exercise-induced effects with cognition (Stillman et al., 2016). Understanding these mechanisms is crucial in order to develop the most effective training interventions to counteract cognitive decline, to prevent dementia, and to reduce fall risk.

159

160 8

References

161 Abellan van Kan, G., Rolland, Y., Andrieu, S., Bauer, J., Beauchet, O., Bonnefoy, M., et al. (2009). Gait speed at usual pace as a predictor of adverse outcomes in community-dwelling older people an International Academy on Nutrition and Aging (IANA) Task Force. J Nutr Health Aging 13, 881-889. Al-Yahya, E., Dawes, H., Smith, L., Dennis, A., Howells, K., and Cockburn, J. (2011). Cognitive motor interference while walking: a systematic review and meta-analysis. Neurosci Biobehav Rev 35, 715-728. doi: 10.1016/j.neubiorev.2010. 08.008. Albinet, C.T., Abou-Dest, A., Andre, N., and Audiffren, M. (2016). Executive functions improvement following a 5-month aquaerobics program in older adults: Role of cardiac vagal control in inhibition performance. Biol Psychol 115, 69-77. doi: 10.1016/ j.biopsycho.2016.01.010. Alexander, N.B., and Hausdorff, J.M. (2008). Guest editorial: linking thinking, walking, and falling. J Gerontol A Biol Sci Med Sci 63, 1325-1328. Amboni, M., Barone, P., and Hausdorff, J.M. (2013). Cognitive contributions to gait and falls: evidence and implications. Mov Disord 28, 1520-1533. doi: 10.1002/mds.25674. Ambrose, A.F., Paul, G., and Hausdorff, J.M. (2013). Risk factors for falls among older adults: a review of the literature. Maturitas 75, 51-61. doi: 10.1016/j.maturitas.2013.02.009. Ammann, B.C., Knols, R.H., Baschung, P., de Bie, R.A., and de Bruin, E.D. (2014). Application of principles of exercise training in sub-acute and chronic stroke survivors: a systematic review. BMC Neurol 14, 167. doi: 10.1186/s12883-014-0167-2. Amosun, S.L., Burgess, T., Groeneveldt, L., and Hodgson, T. (2007). Are elderly pedestrians allowed enough time at pedestrian crossings in Cape Town, South Africa? Physiother Theory Pract 23, 325-332. doi: 10.1080/09593980701593755. Armstrong, R.A. (2014). When to use the Bonferroni correction. Ophthalmic Physiol Opt 34, 502- 508. doi: 10.1111/opo.12131. Asher, L., Aresu, M., Falaschetti, E., and Mindell, J. (2012). Most older pedestrians are unable to cross the road in time: a cross-sectional study. Age Ageing 41, 690-694. doi: 10.1093/ageing/afs076. Austin, M.P., Mitchell, P., and Goodwin, G.M. (2001). Cognitive deficits in depression: possible implications for functional neuropathology. Br J Psychiatry 178, 200-206. Baller, G., Brand, M., Kalbe, E., and Kessler, J. (2006). Inventar zur Gedächtnisdiagnostik (IGD), Manual. Goettingen: Hogrefe. Bangen, K.J., Clark, A.L., Edmonds, E.C., Evangelista, N.D., Werhane, M.L., Thomas, K.R., et al. (2017). Cerebral Blood Flow and Amyloid-beta Interact to Affect Memory Performance in Cognitively Normal Older Adults. Front Aging Neurosci 9, 181. doi: 10.3389/fnagi.2017.00181. Baumgart, M., Snyder, H.M., Carrillo, M.C., Fazio, S., Kim, H., and Johns, H. (2015). Summary of the evidence on modifiable risk factors for cognitive decline and dementia: A population-based perspective. Alzheimers Dement 11, 718-726. doi: 10.1016/j.jalz. 2015.05.016. Bender, A.R., Volkle, M.C., and Raz, N. (2015). Differential aging of cerebral white matter in middle-aged and older adults: A seven-year follow-up. Neuroimage 125, 74-83. doi: 10.1016/j.neuroimage.2015.10.030.

162 Bennett, I.J., and Madden, D.J. (2014). Disconnected aging: cerebral white matter integrity and age-related differences in cognition. Neuroscience 276, 187-205. doi: 10.1016/ j.neuroscience.2013.11.026. Bergerbest, D., Gabrieli, J.D., Whitfield-Gabrieli, S., Kim, H., Stebbins, G.T., Bennett, D.A., et al. (2009). Age-associated reduction of asymmetry in prefrontal function and preservation of conceptual repetition priming. Neuroimage 45, 237-246. doi: 10.1016/ j.neuroimage.2008.10.019. Bergmann, O., Spalding, K.L., and Frisen, J. (2015). Adult Neurogenesis in Humans. Cold Spring Harb Perspect Biol 7, a018994. doi: 10.1101/cshperspect. a018994. Bilney, B., Morris, M., and Webster, K. (2003). Concurrent related validity of the GAITRite walkway system for quantification of the spatial and temporal parameters of gait. Gait Posture 17, 68-74. Boas, D.A., Dale, A.M., and Franceschini, M.A. (2004). Diffuse optical imaging of brain activation: approaches to optimizing image sensitivity, resolution, and accuracy. Neuroimage 23 Suppl 1, S275-288. doi: 10.1016/j.neuroimage. 2004.07.011. Bohannon, R.W., and Williams Andrews, A. (2011). Normal walking speed: a descriptive meta- analysis. Physiotherapy 97, 182-189. doi: 10.1016/j.physio. 2010.12.004. Bollard, E., and Fleming, H. (2013). A study to investigate the walking speed of elderly adults with relation to pedestrian crossings. Physiother Theory Pract 29, 142-149. doi: 10.3109/09593985.2012.703760. Bridle, C., Spanjers, K., Patel, S., Atherton, N.M., and Lamb, S.E. (2012). Effect of exercise on depression severity in older people: systematic review and meta-analysis of randomised controlled trials. Br J Psychiatry 201, 180-185. doi: 10.1192/bjp.bp.111. 095174. Brown, K.C., Hanson, H.M., Firmani, F., Liu, D., McAllister, M.M., Merali, K., et al. (2015). Gait Speed and Variability for Usual Pace and Pedestrian Crossing Conditions in Older Adults Using the GAITRite Walkway. Gerontol Geriatr Med 1, 2333721415618858. doi: 10.1177/2333721415618858. Busch, T.D., Duarte, Y.A., Pires Nunes, D., Lebrao, M.L., Satya Naslavsky, M., dos Santos Rodrigues, A., et al. (2015). Factors associated with lower gait speed among the elderly living in a developing country: a cross-sectional population-based study. BMC Geriatr 15, 35. doi: 10.1186/s12877-015-0031-2. Butler, A.A., Lord, S.R., and Fitzpatrick, R.C. (2016). Perceptions of Speed and Risk: Experimental Studies of Road Crossing by Older People. PLoS One 11, e0152617. doi: 10.1371/journal.pone.0152617. Byberg, L., Melhus, H., Gedeborg, R., Sundstrom, J., Ahlbom, A., Zethelius, B., et al. (2009). Total mortality after changes in leisure time physical activity in 50 year old men: 35 year follow-up of population based cohort. Br J Sports Med 43, 482. Cabeza, R., Anderson, N.D., Locantore, J.K., and McIntosh, A.R. (2002). Aging gracefully: compensatory brain activity in high-performing older adults. Neuroimage 17, 1394- 1402. Cai, L., Chan, J.S., Yan, J.H., and Peng, K. (2014). Brain plasticity and motor practice in cognitive aging. Front Aging Neurosci 6, 31. doi: 10.3389/fnagi.2014.00031.

163 Campbell, A.J., and Robertson, M.C. (2007). Rethinking individual and community fall prevention strategies: a meta-regression comparing single and multifactorial interventions. Age Ageing 36, 656-662. doi: 10.1093/ageing/afm122. Carvalho, M.J., Marques, E., and Mota, J. (2009). Training and detraining effects on functional fitness after a multicomponent training in older women. Gerontology 55, 41-48. doi: 10.1159/000140681. Cassilhas, R.C., Viana, V.A., Grassmann, V., Santos, R.T., Santos, R.F., Tufik, S., et al. (2007). The impact of resistance exercise on the cognitive function of the elderly. Med Sci Sports Exerc 39, 1401-1407. doi: 10.1249/mss.0b013e3180 60111f. Chang, J.T., Morton, S.C., Rubenstein, L.Z., Mojica, W.A., Maglione, M., Suttorp, M.J., et al. (2004). Interventions for the prevention of falls in older adults: systematic review and meta-analysis of randomised clinical trials. BMJ 328, 680. doi: 10.1136/bmj.328. 7441.680. Chang, Y.K., Pan, C.Y., Chen, F.T., Tsai, C.L., and Huang, C.C. (2012). Effect of resistance- exercise training on cognitive function in healthy older adults: a review. J Aging Phys Act 20, 497-517. Chen, P.J., Lin, M.H., Peng, L.N., Liu, C.L., Chang, C.W., Lin, Y.T., et al. (2012). Predicting cause-specific mortality of older men living in the Veterans home by handgrip strength and walking speed: a 3-year, prospective cohort study in Taiwan. J Am Med Dir Assoc 13, 517-521. doi: 10.1016/j.jamda.2012.02. 002. Chodzko-Zajko, W.J., Proctor, D.N., Fiatarone Singh, M.A., Minson, C.T., Nigg, C.R., Salem, G.J., et al. (2009). American College of Sports Medicine. American College of Sports Medicine position stand. Exercise and physical activity for older adults. Med Sci Sports Exerc 41, 1510-1530. doi: 10.1249/MSS.0b013e 3181a0c95c. Chodzko-Zajko, W.J., Schuler, P., Solomon, J., Heinl, B., and Ellis, N.R. (1992). The influence of physical fitness on automatic and effortful memory changes in aging. Int J Aging Hum Dev 35, 265-285. Chu, Y.H., Tang, P.F., Peng, Y.C., and Chen, H.Y. (2013). Meta-analysis of type and complexity of a secondary task during walking on the prediction of elderly falls. Geriatr Gerontol Int 13, 289-297. doi: 10.1111/j.1447-0594.2012. 00893.x. Clark, D.J., Rose, D.K., Ring, S.A., and Porges, E.C. (2014). Utilization of central nervous system resources for preparation and performance of complex walking tasks in older adults. Front Aging Neurosci 6, 217. doi: 10.3389/fnagi.2014.00217. Cohen, J. (1988). Statistical power analysis for the behavioral sciences. Hillsdale, New Jersey: Lawrence Erlbaum. Colcombe, S., and Kramer, A.F. (2003). Fitness effects on the cognitive function of older adults: a meta-analytic study. Psychol Sci 14, 125-130. Colcombe, S.J., Kramer, A.F., Erickson, K.I., and Scalf, P. (2005). The implications of cortical recruitment and brain morphology for individual differences in inhibitory function in aging humans. Psychol Aging 20, 363-375. doi: 10.1037/0882-7974.20.3.363. Colcombe, S.J., Kramer, A.F., Erickson, K.I., Scalf, P., McAuley, E., Cohen, N.J., et al. (2004). Cardiovascular fitness, cortical plasticity, and aging. Proc Natl Acad Sci U S A 101, 3316-3321. doi: 10.1073/pnas.0400266101.

164 Corrigan, R., and McBurney, H. (2008). Community ambulation: environmental impacts and assessment inadequacies. Disabil Rehabil 30, 1411-1419. doi: 10.1080/09638280701 654542. Cotman, C.W., Berchtold, N.C., and Christie, L.A. (2007). Exercise builds brain health: key roles of growth factor cascades and inflammation. Trends Neurosci 30, 464-472. doi: 10.1016/j.tins.2007.06.011. Coubard, O.A., Duretz, S., Lefebvre, V., Lapalus, P., and Ferrufino, L. (2011). Practice of contemporary dance improves cognitive flexibility in aging. Front Aging Neurosci 3, 13. doi: 10.3389/fnagi.2011.00013. Cowansage, K.K., LeDoux, J.E., and Monfils, M.H. (2010). Brain-derived neurotrophic factor: a dynamic gatekeeper of neural plasticity. Curr Mol Pharmacol 3, 12-29. Cui, X., Bray, S., and Reiss, A.L. (2010). Speeded near infrared spectroscopy (NIRS) response detection. PLoS One 5, e15474. doi: 10.1371/journal.pone.0015474. CURAVIVA (2014). Statistik Alters- und Pflegeinstitutionen [homepage on the internet]. CURAVIVA Schweiz. Available from: https://www.curaviva.ch/Verlag/PDgwT/?id= 4F1159DA-C917-18D4-9E2E6B94512ABB65&method=objectdata.detail&p=3&caller id=&c=DE609894-EC42-78FB-ABEE19D5B63B1482 [Accessed July 28, 2017]. Davis, J.C., Robertson, M.C., Ashe, M.C., Liu-Ambrose, T., Khan, K.M., and Marra, C.A. (2010). International comparison of cost of falls in older adults living in the community: a systematic review. Osteoporos Int 21, 1295-1306. doi: 10.1007/s00198-009-1162-0. de Bruin, E.D., and Murer, K. (2007). Effect of additional functional exercises on balance in elderly people. Clin Rehabil 21, 112-121. doi: 10.1177/0269215506070144. de Bruin, E.D., Reith, A., Dörflinger, M., and Murer, K. (2011). Feasibility of Strength-Balance Training Extended with Computer Game Dancing in Older People; Does it Affect Dual Task Costs of Walking? J Nov Physiother 1, 104. de Bruin, E.D., and Schmidt, A. (2010). Walking behaviour of healthy elderly: attention should be paid. Behav Brain Funct 6, 59. doi: 10.1186/1744-9081-6-59. Deary, I.J., Corley, J., Gow, A.J., Harris, S.E., Houlihan, L.M., Marioni, R.E., et al. (2009). Age- associated cognitive decline. Br Med Bull 92, 135-152. doi: 10.1093/bmb/ldp033. DeCarlo, C.A., Tuokko, H.A., Williams, D., Dixon, R.A., and MacDonald, S.W. (2014). BioAge: toward a multi-determined, mechanistic account of cognitive aging. Ageing Res Rev 18, 95-105. doi: 10.1016/j.arr.2014.09.003. Ding, Q., Vaynman, S., Akhavan, M., Ying, Z., and Gomez-Pinilla, F. (2006). Insulin-like growth factor I interfaces with brain-derived neurotrophic factor-mediated synaptic plasticity to modulate aspects of exercise-induced cognitive function. Neuroscience 140, 823-833. doi: 10.1016/j.neuroscience.2006.02.084. Dommes, A., Cavallo, V., and Oxley, J. (2013). Functional declines as predictors of risky street- crossing decisions in older pedestrians. Accid Anal Prev 59, 135-143. doi: 10.1016/ j.aap.2013.05.017. Donoghue, O.A., Dooley, C., and Kenny, R.A. (2016). Usual and Dual-Task Walking Speed: Implications for Pedestrians Crossing the Road. J Aging Health 28, 850-862. doi: 10.1177/0898264315614004.

165 Draganski, B., Gaser, C., Busch, V., Schuierer, G., Bogdahn, U., and May, A. (2004). Neuroplasticity: changes in grey matter induced by training. Nature 427, 311-312. doi: 10.1038/427311a. Dumurgier, J., Elbaz, A., Ducimetiere, P., Tavernier, B., Alperovitch, A., and Tzourio, C. (2009). Slow walking speed and cardiovascular death in well functioning older adults: prospective cohort study. BMJ 339, b4460. doi: 10.1136/bmj.b4460. Dupuy, O., Gauthier, C.J., Fraser, S.A., Desjardins-Crepeau, L., Desjardins, M., Mekary, S., et al. (2015). Higher levels of cardiovascular fitness are associated with better executive function and prefrontal oxygenation in younger and older women. Front Hum Neurosci 9, 66. doi: 10.3389/fnhum.2015.00066. Duzel, E., van Praag, H., and Sendtner, M. (2016). Can physical exercise in old age improve memory and hippocampal function? Brain 139, 662-673. doi: 10.1093/brain/awv407. Eggenberger, P., Schumacher, V., Angst, M., Theill, N., and de Bruin, E.D. (2015a). Does multicomponent physical exercise with simultaneous cognitive training boost cognitive performance in older adults? A 6-month randomized controlled trial with a 1-year follow-up. Clin Interv Aging 10, 1335-1349. doi: 10.2147/CIA.S87732. Eggenberger, P., Theill, N., Holenstein, S., Schumacher, V., and de Bruin, E.D. (2015b). Multicomponent physical exercise with simultaneous cognitive training to enhance dual-task walking of older adults: a secondary analysis of a 6-month randomized controlled trial with 1-year follow-up. Clin Interv Aging 10, 1711-1732. doi: 10.2147/ CIA.S91997. Eggenberger, P., Wolf, M., Schumann, M., and de Bruin, E.D. (2016). Exergame and Balance Training Modulate Prefrontal Brain Activity during Walking and Enhance Executive Function in Older Adults. Front Aging Neurosci 8, 66. doi: 10.3389/fnagi.2016.00066. Ekkekakis, P. (2009). Illuminating the black box: investigating prefrontal cortical hemodynamics during exercise with near-infrared spectroscopy. J Sport Exerc Psychol 31, 505-553. Erickson, K.I., Colcombe, S.J., Wadhwa, R., Bherer, L., Peterson, M.S., Scalf, P.E., et al. (2007a). Training-induced functional activation changes in dual-task processing: an FMRI study. Cereb Cortex 17, 192-204. doi: 10.1093/cercor/bhj137. Erickson, K.I., Colcombe, S.J., Wadhwa, R., Bherer, L., Peterson, M.S., Scalf, P.E., et al. (2007b). Training-induced plasticity in older adults: effects of training on hemispheric asymmetry. Neurobiol Aging 28, 272-283. doi: 10.1016/j.neurobiolaging.2005.12.012. Erickson, K.I., Leckie, R.L., and Weinstein, A.M. (2014). Physical activity, fitness, and gray matter volume. Neurobiol Aging 35 Suppl 2, S20-28. doi: 10.1016/j.neurobiolaging. 2014.03.034. Erickson, K.I., Prakash, R.S., Voss, M.W., Chaddock, L., Hu, L., Morris, K.S., et al. (2009). Aerobic fitness is associated with hippocampal volume in elderly humans. Hippocampus 19, 1030-1039. doi: 10.1002/hipo.20547. Erickson, K.I., Voss, M.W., Prakash, R.S., Basak, C., Szabo, A., Chaddock, L., et al. (2011). Exercise training increases size of hippocampus and improves memory. Proc Natl Acad Sci U S A 108, 3017-3022. doi: 10.1073/pnas.1015950108. Erickson, K.I., Weinstein, A.M., and Lopez, O.L. (2012). Physical activity, brain plasticity, and Alzheimer's disease. Arch Med Res 43, 615-621. doi: 10.1016/j.arcmed.2012.09.008.

166 Ernst, A., Alkass, K., Bernard, S., Salehpour, M., Perl, S., Tisdale, J., et al. (2014). Neurogenesis in the striatum of the adult human brain. Cell 156, 1072-1083. doi: 10.1016/j.cell.2014.01.044. Etnier, J.L., Nowell, P.M., Landers, D.M., and Sibley, B.A. (2006). A meta-regression to examine the relationship between aerobic fitness and cognitive performance. Brain Research Reviews 52, 119-130. doi: Doi 10.1016/J.Brainresrev.2006.01.002. Fabel, K., Fabel, K., Tam, B., Kaufer, D., Baiker, A., Simmons, N., et al. (2003). VEGF is necessary for exercise-induced adult hippocampal neurogenesis. Eur J Neurosci 18, 2803-2812. Fabre, C., Chamari, K., Mucci, P., Masse-Biron, J., and Prefaut, C. (2002). Improvement of cognitive function by mental and/or individualized aerobic training in healthy elderly subjects. Int J Sports Med 23, 415-421. doi: 10.1055/s-2002-33735. Faul, F., Erdfelder, E., Lang, A.G., and Buchner, A. (2007). G*Power 3: a flexible statistical power analysis program for the social, behavioral, and biomedical sciences. Behav Res Methods 39, 175-191. Ferrari, M., and Quaresima, V. (2012). A brief review on the history of human functional near- infrared spectroscopy (fNIRS) development and fields of application. Neuroimage 63, 921-935. doi: 10.1016/j.neuroimage.2012.03.049. Fissler, P., Muller, H.P., Kuster, O.C., Laptinskaya, D., Thurm, F., Woll, A., et al. (2017). No Evidence That Short-Term Cognitive or Physical Training Programs or Lifestyles Are Related to Changes in White Matter Integrity in Older Adults at Risk of Dementia. Front Hum Neurosci 11, 110. doi: 10.3389/fnhum.2017.00110. Fjell, A.M., and Walhovd, K.B. (2010). Structural brain changes in aging: courses, causes and cognitive consequences. Rev Neurosci 21, 187-221. Fletcher, M.A., Low, K.A., Boyd, R., Zimmerman, B., Gordon, B.A., Tan, C.H., et al. (2016). Comparing Aging and Fitness Effects on Brain Anatomy. Front Hum Neurosci 10, 286. doi: 10.3389/fnhum.2016.00286. Folland, J.P., and Williams, A.G. (2007). The adaptations to strength training : morphological and neurological contributions to increased strength. Sports Med 37, 145-168. Folstein, M.F., Folstein, S.E., and McHugh, P.R. (1975). "Mini-mental state". A practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res 12, 189-198. Forte, R., Boreham, C.A., De Vito, G., and Pesce, C. (2015). Health and Quality of Life Perception in Older Adults: The Joint Role of Cognitive Efficiency and Functional Mobility. Int J Environ Res Public Health 12, 11328-11344. doi: 10.3390/ijerph 120911328. Forte, R., Boreham, C.A., Leite, J.C., De Vito, G., Brennan, L., Gibney, E.R., et al. (2013). Enhancing cognitive functioning in the elderly: multicomponent vs resistance training. Clin Interv Aging 8, 19-27. doi: 10.2147/CIA.S36514. Franklin, B.A., Brinks, J., Sacks, R., Trivax, J., and Friedman, H. (2015). Reduced walking speed and distance as harbingers of the approaching grim reaper. Am J Cardiol 116, 313- 317. doi: 10.1016/j.amjcard.2015.04.024. Fraser, S.A., Elliot, V., de Bruin, E.D., Bherer, L., and Dumoulin, C. (2014). The effects of combining video game dancing and pelvic floor training to improve dual-task gait and cognition in women with mixed-urinary incontinence. Games Health J 3, 172-178.

167 Fraser, S.A., Li, K.Z., Berryman, N., Desjardins-Crepeau, L., Lussier, M., Vadaga, K., et al. (2017). Does Combined Physical and Cognitive Training Improve Dual-Task Balance and Gait Outcomes in Sedentary Older Adults? Front Hum Neurosci 10, 688. doi: 10.3389/fnhum.2016.00688. Fried, L.P., Tangen, C.M., Walston, J., Newman, A.B., Hirsch, C., Gottdiener, J., et al. (2001). Frailty in older adults: evidence for a phenotype. J Gerontol A Biol Sci Med Sci 56, M146-156. Fritz, S., and Lusardi, M. (2009). White paper: "walking speed: the sixth vital sign". J Geriatr Phys Ther 32, 46-49. Fuster, J.M. (2003). Cortex and Mind: Unifying Cognition. New York: Oxford University Press. Gage, F.H. (2000). Mammalian neural stem cells. Science 287, 1433-1438. Gardner, M.M., Robertson, M.C., and Campbell, A.J. (2000). Exercise in preventing falls and fall related injuries in older people: a review of randomised controlled trials. Br J Sports Med 34, 7-17. Gatterer, G. (2008). Der Alters-Konzentrations-Test. Goettingen: Hogrefe. Gauggel, S., and Birkner, B. (1999). Validität und Reliabilität einer deutschen Version der Geriatrischen Depressionsskala (GDS). Zeitschrift für Klinische Psychologie 28, 18-27. Gazzaley, A., and D’Esposito, M. (2006). Neural Networks: an empirical neuroscience approach toward understanding cognition. Cortex 42, 1037-1040. Geraghty, J., Holland, C., and Rochelle, K. (2016). Examining links between cognitive markers, movement initiation and change, and pedestrian safety in older adults. Accid Anal Prev 89, 151-159. doi: 10.1016/j.aap.2015.12.019. Gesundheit-und-Alter (2017). Geriatrische Klinik St.Gallen [homepage on the internet]. Available from: http://www.gesundheitundalter.ch/Home/GeriatrischeKlinik/Portrait/ tabid/106/Default.aspx [Accessed July 28, 2017]. Gill, T., Taylor, A.W., and Pengelly, A. (2005). A population-based survey of factors relating to the prevalence of falls in older people. Gerontology 51, 340-345. doi: 10.1159/ 00008 6372. Gillespie, L.D., Robertson, M.C., Gillespie, W.J., Sherrington, C., Gates, S., Clemson, L.M., et al. (2012). Interventions for preventing falls in older people living in the community. Cochrane Database Syst Rev 9, CD007146. doi: 10.1002/14651858.CD007146.pub3. Goldberg, A., and Schepens, S. (2011). Measurement error and minimum detectable change in 4- meter gait speed in older adults. Aging Clin Exp Res 23, 406-412. Gomez-Pinilla, F., and Hillman, C. (2013). The influence of exercise on cognitive abilities. Compr Physiol 3, 403-428. doi: 10.1002/cphy.c110063. Gonzales, J.U., James, C.R., Yang, H.S., Jensen, D., Atkins, L., Thompson, B.J., et al. (2016). Different cognitive functions discriminate gait performance in younger and older women: A pilot study. Gait Posture 50, 89-95. doi: 10.1016/j.gaitpost.2016.08.021. Grady, C. (2012). The cognitive neuroscience of ageing. Nat Rev Neurosci 13, 491-505. doi: 10.1038/nrn3256.

168 Granacher, U., Muehlbauer, T., and Gruber, M. (2012). A qualitative review of balance and strength performance in healthy older adults: impact for testing and training. J Aging Res 2012, 708905. doi: 10.1155/2012/708905. Graves, L.E., Ridgers, N.D., Williams, K., Stratton, G., Atkinson, G., and Cable, N.T. (2010). The physiological cost and enjoyment of Wii Fit in adolescents, young adults, and older adults. J Phys Act Health 7, 393-401. Green, H.F., and Nolan, Y.M. (2014). Inflammation and the developing brain: consequences for hippocampal neurogenesis and behavior. Neurosci Biobehav Rev 40, 20-34. doi: 10.1016/j.neubiorev.2014.01.004. Gregory, M.A., Gill, D.P., and Petrella, R.J. (2013). Brain health and exercise in older adults. Curr Sports Med Rep 12, 256-271. doi: 10.1249/JSR.0b013e31829a74fd. Gudlaugsson, J., Gudnason, V., Aspelund, T., Siggeirsdottir, K., Olafsdottir, A.S., Jonsson, P.V., et al. (2012). Effects of a 6-month multimodal training intervention on retention of functional fitness in older adults: a randomized-controlled cross-over design. Int J Behav Nutr Phys Act 9, 107. doi: 10.1186/1479-5868-9-107. Guralnik, J.M., Simonsick, E.M., Ferrucci, L., Glynn, R.J., Berkman, L.F., Blazer, D.G., et al. (1994). A short physical performance battery assessing lower extremity function: association with self-reported disability and prediction of mortality and nursing home admission. J Gerontol 49, M85-94. Hamacher, D., Herold, F., Wiegel, P., Hamacher, D., and Schega, L. (2015). Brain activity during walking: A systematic review. Neurosci Biobehav Rev 57, 310-327. doi: 10.1016/j.neubiorev.2015.08.002. Harada, T., Miyai, I., Suzuki, M., and Kubota, K. (2009). Gait capacity affects cortical activation patterns related to speed control in the elderly. Exp Brain Res 193, 445-454. doi: 10.1007/s00221-008-1643-y. Härting, C., Markowitsch, H.J., Neufeld, H., Calabrese, P., Deisinger, K., and Kessler, J. (2000). WMS-R - Wechsler Gedächtnistest - revidierte Fassung. Bern: Huber. Hausdorff, J.M., Rios, D.A., and Edelberg, H.K. (2001). Gait variability and fall risk in community-living older adults: a 1-year prospective study. Arch Phys Med Rehabil 82, 1050-1056. doi: 10.1053/apmr.2001.24893. Herman, T., Giladi, N., Gruendlinger, L., and Hausdorff, J.M. (2007). Six weeks of intensive treadmill training improves gait and quality of life in patients with Parkinson's disease: a pilot study. Arch Phys Med Rehabil 88, 1154-1158. doi: 10.1016/j.apmr. 2007.05.015. Herman, T., Mirelman, A., Giladi, N., Schweiger, A., and Hausdorff, J.M. (2010). Executive control deficits as a prodrome to falls in healthy older adults: a prospective study linking thinking, walking, and falling. J Gerontol A Biol Sci Med Sci 65, 1086-1092. doi: 10.1093/gerona/glq077. Herold, F., Wiegel, P., Scholkmann, F., Thiers, A., Hamacher, D., and Schega, L. (2017). Functional near-infrared spectroscopy in movement science: a systematic review on cortical activity in postural and walking tasks. Neurophotonics 4, 041403. doi: 10.1117/1.NPh.4.4.041403.

169 Hindin, S.B., and Zelinski, E.M. (2012). Extended practice and aerobic exercise interventions benefit untrained cognitive outcomes in older adults: a meta-analysis. J Am Geriatr Soc 60, 136-141. doi: 10.1111/j.1532-5415.2011.03761.x. Holtzer, R., Epstein, N., Mahoney, J.R., Izzetoglu, M., and Blumen, H.M. (2014). Neuroimaging of mobility in aging: a targeted review. J Gerontol A Biol Sci Med Sci 69, 1375-1388. doi: 10.1093/gerona/glu052. Holtzer, R., Friedman, R., Lipton, R.B., Katz, M., Xue, X., and Verghese, J. (2007). The relationship between specific cognitive functions and falls in aging. Neuropsychology 21, 540-548. doi: 10.1037/0894-4105.21.5.540. Holtzer, R., Mahoney, J.R., Izzetoglu, M., Izzetoglu, K., Onaral, B., and Verghese, J. (2011). fNIRS study of walking and walking while talking in young and old individuals. J Gerontol A Biol Sci Med Sci 66, 879-887. doi: 10.1093/gerona/glr068. Hortobagyi, T., Lesinski, M., Gabler, M., VanSwearingen, J.M., Malatesta, D., and Granacher, U. (2015). Effects of Three Types of Exercise Interventions on Healthy Old Adults' Gait Speed: A Systematic Review and Meta-Analysis. Sports Med 45, 1627-1643. doi: 10.1007/s40279-015-0371-2. Hotting, K., and Roder, B. (2013). Beneficial effects of physical exercise on neuroplasticity and cognition. Neurosci Biobehav Rev 37, 2243-2257. doi: 10.1016/j.neubiorev.2013.04. 005. Hotting, K., Schauenburg, G., and Roder, B. (2012). Long-term effects of physical exercise on verbal learning and memory in middle-aged adults: results of a one-year follow-up study. Brain Sci 2, 332-346. doi: 10.3390/brainsci2030332. Hsu, C.L., Best, J.R., Wang, S., Voss, M.W., Hsiung, R.G.Y., Munkacsy, M., et al. (2017). The Impact of Aerobic Exercise on Fronto-Parietal Network Connectivity and Its Relation to Mobility: An Exploratory Analysis of a 6-Month Randomized Controlled Trial. Front Hum Neurosci 11, 344. doi: 10.3389/fnhum.2017.00344. Huang, P., Fang, R., Li, B.Y., and Chen, S.D. (2016). Exercise-Related Changes of Networks in Aging and Mild Cognitive Impairment Brain. Front Aging Neurosci 8, 47. doi: 10.3389/fnagi.2016.00047. Huneau, C., Benali, H., and Chabriat, H. (2015). Investigating Human Neurovascular Coupling Using Functional Neuroimaging: A Critical Review of Dynamic Models. Front Neurosci 9, 467. doi: 10.3389/fnins.2015.00467. Hyodo, K., Dan, I., Suwabe, K., Kyutoku, Y., Yamada, Y., Akahori, M., et al. (2012). Acute moderate exercise enhances compensatory brain activation in older adults. Neurobiol Aging 33, 2621-2632. doi: 10.1016/j.neurobiolaging.2011.12.022. Ivey, F.M., Tracy, B.L., Lemmer, J.T., NessAiver, M., Metter, E.J., Fozard, J.L., et al. (2000). Effects of strength training and detraining on muscle quality: age and gender comparisons. J Gerontol A Biol Sci Med Sci 55, B152-157; discussion B158-159. Jäncke, L., Shah, N.J., and Peters, M. (2000). Cortical activations in primary and secondary motor areas for complex bimanual movements in professional pianists. Brain Res Cogn Brain Res 10, 177-183. Jekauc, D., Voelkle, M., Wagner, M.O., Mewes, N., and Woll, A. (2013). Reliability, validity, and measurement invariance of the German version of the physical activity enjoyment scale. J Pediatr Psychol 38, 104-115. doi: 10.1093/jpepsy/jss088.

170 Johnson, N.F., Kim, C., Clasey, J.L., Bailey, A., and Gold, B.T. (2012). Cardiorespiratory fitness is positively correlated with cerebral white matter integrity in healthy seniors. Neuroimage 59, 1514-1523. doi: 10.1016/j.neuroimage.2011.08.032. Julien, C. (2006). The enigma of Mayer waves: Facts and models. Cardiovasc Res 70, 12-21. doi: 10.1016/j.cardiores.2005.11.008. Jung, R., Moser, M., Baucsek, S., Dern, S., and Schneider, S. (2015). Activation patterns of different brain areas during incremental exercise measured by near-infrared spectroscopy. Exp Brain Res 233, 1175-1180. doi: 10.1007/s00221-015-4201-4. Kandola, A., Hendrikse, J., Lucassen, P.J., and Yucel, M. (2016). Aerobic Exercise as a Tool to Improve Hippocampal Plasticity and Function in Humans: Practical Implications for Mental Health Treatment. Front Hum Neurosci 10, 373. doi: 10.3389/fnhum.2016. 00373. Kanton-SG (2017). Amt für Sport Kanton St.Gallen [homepage on the internet]. Available from: http://www.sg.ch/home/bildung/sport/Erwachsene.html [Accessed July 28, 2017]. Kempermann, G., Fabel, K., Ehninger, D., Babu, H., Leal-Galicia, P., Garthe, A., et al. (2010). Why and how physical activity promotes experience-induced brain plasticity. Front Neurosci 4, 189. doi: 10.3389/fnins.2010.00189. Kempermann, G., Jessberger, S., Steiner, B., and Kronenberg, G. (2004). Milestones of neuronal development in the adult hippocampus. Trends Neurosci 27, 447-452. doi: 10.1016/j.tins.2004.05.013. Kennedy, K.M., Rodrigue, K.M., Bischof, G.N., Hebrank, A.C., Reuter-Lorenz, P.A., and Park, D.C. (2015). Age trajectories of functional activation under conditions of low and high processing demands: an adult lifespan fMRI study of the aging brain. Neuroimage 104, 21-34. doi: 10.1016/j.neuroimage.2014.09.056. Kim, H.J., Park, I., Lee, H.J., and Lee, O. (2016). The reliability and validity of gait speed with different walking pace and distances against general health, physical function, and chronic disease in aged adults. J Exerc Nutrition Biochem 20, 46-50. doi: 10.20463/ jenb.2016.09.20.3.7. Koenraadt, K.L., Roelofsen, E.G., Duysens, J., and Keijsers, N.L. (2014). Cortical control of normal gait and precision stepping: an fNIRS study. Neuroimage 85 Pt 1, 415-422. doi: 10.1016/j.neuroimage.2013.04.070. Kraft, E. (2012). Cognitive function, physical activity, and aging: possible biological links and implications for multimodal interventions. Neuropsychol Dev Cogn B Aging Neuropsychol Cogn 19, 248-263. doi: 10.1080/13825585.2011.645010. Kuh, D., Karunananthan, S., Bergman, H., and Cooper, R. (2014). A life-course approach to healthy ageing: maintaining physical capability. Proc Nutr Soc 73, 237-248. doi: 10.1017/S0029665113003923. Kuznetsova, K.A., Maniega, S.M., Ritchie, S.J., Cox, S.R., Storkey, A.J., Starr, J.M., et al. (2015). Brain white matter structure and information processing speed in healthy older age. Brain Struct Funct. doi: 10.1007/s00429-015-1097-5. Kwon, S., Perera, S., Pahor, M., Katula, J.A., King, A.C., Groessl, E.J., et al. (2009). What is a meaningful change in physical performance? Findings from a clinical trial in older adults (the LIFE-P study). J Nutr Health Aging 13, 538-544.

171 la Fougere, C., Zwergal, A., Rominger, A., Forster, S., Fesl, G., Dieterich, M., et al. (2010). Real versus imagined locomotion: a [18F]-FDG PET-fMRI comparison. Neuroimage 50, 1589-1598. doi: 10.1016/j.neuroimage.2009.12.060. Laboratories, A.T.S.C.o.P.S.f.C.P.F. (2002). ATS statement: guidelines for the six-minute walk test. Am J Respir Crit Care Med 166, 111-117. doi: 10.1164/ajrccm.166.1.at1102. Lambiase, M.J., Gabriel, K.P., Kuller, L.H., and Matthews, K.A. (2014). Sleep and executive function in older women: the moderating effect of physical activity. J Gerontol A Biol Sci Med Sci 69, 1170-1176. doi: 10.1093/gerona/glu038. Langlois, J.A., Keyl, P.M., Guralnik, J.M., Foley, D.J., Marottoli, R.A., and Wallace, R.B. (1997). Characteristics of older pedestrians who have difficulty crossing the street. Am J Public Health 87, 393-397. Lautenschlager, N.T., Cox, K., and Cyarto, E.V. (2012). The influence of exercise on brain aging and dementia. Biochim Biophys Acta 1822, 474-481. doi: 10.1016/j.bbadis.2011. 07.010. Law, L.L., Barnett, F., Yau, M.K., and Gray, M.A. (2014). Effects of combined cognitive and exercise interventions on cognition in older adults with and without cognitive impairment: a systematic review. Ageing Res Rev 15, 61-75. doi: 10.1016/j.arr.2014. 02.008. Leff, D.R., Elwell, C.E., Orihuela-Espina, F., Atallah, L., Delpy, D.T., Darzi, A.W., et al. (2008). Changes in prefrontal cortical behaviour depend upon familiarity on a bimanual coordination task: an fNIRS study. Neuroimage 39, 805-813. doi: 10.1016/j.neuroimage. 2007.09.032. Legault, C., Jennings, J.M., Katula, J.A., Dagenbach, D., Gaussoin, S.A., Sink, K.M., et al. (2011). Designing clinical trials for assessing the effects of cognitive training and physical activity interventions on cognitive outcomes: the Seniors Health and Activity Research Program Pilot (SHARP-P) study, a randomized controlled trial. BMC Geriatr 11, 27. doi: 10.1186/1471-2318-11-27. Leisman, G., Moustafa, A.A., and Shafir, T. (2016). Thinking, Walking, Talking: Integratory Motor and Cognitive Brain Function. Front Public Health 4, 94. doi: 10.3389/fpubh.2016.00094. Lezak, M.D., Howiesen, D.B., Bigler, E.D., and Tranel, D. (2012). Neuropsychological Assessment. Oxford: Oxford University Press. Lichtenberg, P.A., Ross, T., Millis, S.R., and Manning, C.A. (1995). The relationship between depression and cognition in older adults: a cross-validation study. J Gerontol B Psychol Sci Soc Sci 50, P25-P32. Liu, B., Hu, X., Zhang, Q., Fan, Y., Li, J., Zou, R., et al. (2016). Usual walking speed and all-cause mortality risk in older people: A systematic review and meta-analysis. Gait Posture 44, 172-177. doi: 10.1016/j.gaitpost.2015.12.008. Liu-Ambrose, T., and Donaldson, M.G. (2009). Exercise and cognition in older adults: is there a role for resistance training programmes? Br J Sports Med 43, 25-27. doi: 10.1136/ bjsm.2008.055616. Liu-Ambrose, T., Nagamatsu, L.S., Graf, P., Beattie, B.L., Ashe, M.C., and Handy, T.C. (2010). Resistance training and executive functions: a 12-month randomized controlled trial. Arch Intern Med 170, 170-178. doi: 10.1001/archinternmed.2009.494.

172 Liu-Ambrose, T., Nagamatsu, L.S., Voss, M.W., Khan, K.M., and Handy, T.C. (2012). Resistance training and functional plasticity of the aging brain: a 12-month randomized controlled trial. Neurobiol Aging 33, 1690-1698. doi: 10.1016/ j.neurobiolaging.2011.05.010. Lockhart, S.N., and DeCarli, C. (2014). Structural imaging measures of brain aging. Neuropsychol Rev 24, 271-289. doi: 10.1007/s11065-014-9268-3. Lord, S., Galna, B., Verghese, J., Coleman, S., Burn, D., and Rochester, L. (2013). Independent domains of gait in older adults and associated motor and nonmotor attributes: validation of a factor analysis approach. J Gerontol A Biol Sci Med Sci 68, 820-827. doi: 10.1093/gerona/gls255. Lu, C.F., Liu, Y.C., Yang, Y.R., Wu, Y.T., and Wang, R.Y. (2015). Maintaining Gait Performance by Cortical Activation during Dual-Task Interference: A Functional Near-Infrared Spectroscopy Study. PLoS One 10, e0129390. doi: 10.1371/journal. pone.0129390. Lu, P.H., Lee, G.J., Raven, E.P., Tingus, K., Khoo, T., Thompson, P.M., et al. (2011). Age-related slowing in cognitive processing speed is associated with myelin integrity in a very healthy elderly sample. J Clin Exp Neuropsychol 33, 1059-1068. doi: 10.1080/ 13803395.2011.595397. Lustig, C., Shah, P., Seidler, R., and Reuter-Lorenz, P.A. (2009). Aging, training, and the brain: a review and future directions. Neuropsychol Rev 19, 504-522. doi: 10.1007/s11065-009- 9119-9. Ma, C.L., Ma, X.T., Wang, J.J., Liu, H., Chen, Y.F., and Yang, Y. (2017). Physical exercise induces hippocampal neurogenesis and prevents cognitive decline. Behav Brain Res 317, 332- 339. doi: 10.1016/j.bbr.2016.09.067. Maass, A., Duzel, S., Goerke, M., Becke, A., Sobieray, U., Neumann, K., et al. (2015). Vascular hippocampal plasticity after aerobic exercise in older adults. Mol Psychiatry 20, 585- 593. doi: 10.1038/mp.2014.114. MacDonald, S.W., DeCarlo, C.A., and Dixon, R.A. (2011). Linking biological and cognitive aging: toward improving characterizations of developmental time. J Gerontol B Psychol Sci Soc Sci 66 Suppl 1, i59-70. doi: 10.1093/geronb/gbr039. MacKenzie, D.M., Copp, P., Shaw, R.J., and Goodwin, G.M. (1996). Brief cognitive screening of the elderly: a comparison of the Mini-Mental State Examination (MMSE), Abbreviated Mental Test (AMT) and Mental Status Questionnaire (MSQ). Psychol Med 26, 427-430. Marks, B.L., Katz, L.M., Styner, M., and Smith, J.K. (2011). Aerobic fitness and obesity: relationship to cerebral white matter integrity in the brain of active and sedentary older adults. Br J Sports Med 45, 1208-1215. doi: 10.1136/bjsm.2009.068114. Marks, B.L., Madden, D.J., Bucur, B., Provenzale, J.M., White, L.E., Cabeza, R., et al. (2007). Role of aerobic fitness and aging on cerebral white matter integrity. Ann N Y Acad Sci 1097, 171-174. doi: 10.1196/annals.1379.022. Marmeleira, J. (2013). An examination of the mechanisms underlying the effects of physical activity on brain and cognition. European Review of Aging and Physical Activity 10, 83- 94. doi: Doi 10.1007/S11556-012-0105-5. Mattay, V.S., Fera, F., Tessitore, A., Hariri, A.R., Berman, K.F., Das, S., et al. (2006). Neurophysiological correlates of age-related changes in working memory capacity. Neurosci Lett 392, 32-37. doi: 10.1016/j.neulet.2005.09.025.

173 Meester, D., Al-Yahya, E., Dawes, H., Martin-Fagg, P., and Pinon, C. (2014). Associations between prefrontal cortex activation and H-reflex modulation during dual task gait. Front Hum Neurosci 8, 78. doi: 10.3389/fnhum.2014.00078. Melnyk, J.A., Rogers, M.A., and Hurley, B.F. (2009). Effects of strength training and detraining on regional muscle in young and older men and women. Eur J Appl Physiol 105, 929- 938. doi: 10.1007/s00421-008-0979-0. Menant, J.C., Schoene, D., Sarofim, M., and Lord, S.R. (2014). Single and dual task tests of gait speed are equivalent in the prediction of falls in older people: a systematic review and meta-analysis. Ageing Res Rev 16, 83-104. doi: 10.1016/j.arr.2014.06.001. Middleton, A., Fritz, S.L., and Lusardi, M. (2015). Walking speed: the functional vital sign. J Aging Phys Act 23, 314-322. doi: 10.1123/japa.2013-0236. Mirelman, A., Herman, T., Brozgol, M., Dorfman, M., Sprecher, E., Schweiger, A., et al. (2012). Executive function and falls in older adults: new findings from a five-year prospective study link fall risk to cognition. PLoS One 7, e40297. doi: 10.1371/journal.pone. 0040297. Mirelman, A., Maidan, I., Bernad-Elazari, H., Nieuwhof, F., Reelick, M., Giladi, N., et al. (2014). Increased frontal brain activation during walking while dual tasking: an fNIRS study in healthy young adults. J Neuroeng Rehabil 11, 85. doi: 10.1186/1743-0003-11-85. Miyake, A., Friedman, N.P., Emerson, M.J., Witzki, A.H., Howerter, A., and Wager, T.D. (2000). The unity and diversity of executive functions and their contributions to complex "Frontal Lobe" tasks: a latent variable analysis. Cogn Psychol 41, 49-100. doi: 10.1006/cogp.1999.0734. Miyazawa, T., Horiuchi, M., Komine, H., Sugawara, J., Fadel, P.J., and Ogoh, S. (2013). Skin blood flow influences cerebral oxygenation measured by near-infrared spectroscopy during dynamic exercise. Eur J Appl Physiol 113, 2841-2848. doi: 10.1007/s00421-013- 2723-7. Moher, D., Hopewell, S., Schulz, K.F., Montori, V., Gotzsche, P.C., Devereaux, P.J., et al. (2010). CONSORT 2010 explanation and elaboration: updated guidelines for reporting parallel group randomised trials. BMJ 340, c869. doi: 10.1136/bmj.c869. Montero-Odasso, M., Verghese, J., Beauchet, O., and Hausdorff, J.M. (2012). Gait and cognition: a complementary approach to understanding brain function and the risk of falling. J Am Geriatr Soc 60, 2127-2136. doi: 10.1111/j.1532-5415.2012.04209.x. Muir, S.W., Gopaul, K., and Montero Odasso, M.M. (2012). The role of cognitive impairment in fall risk among older adults: a systematic review and meta-analysis. Age Ageing 41, 299-308. doi: 10.1093/ageing/afs012. Muir-Hunter, S.W., and Wittwer, J.E. (2016). Dual-task testing to predict falls in community- dwelling older adults: a systematic review. Physiotherapy 102, 29-40. doi: 10.1016/j.physio.2015.04.011. Mullen, S.P., Olson, E.A., Phillips, S.M., Szabo, A.N., Wojcicki, T.R., Mailey, E.L., et al. (2011). Measuring enjoyment of physical activity in older adults: invariance of the physical activity enjoyment scale (paces) across groups and time. Int J Behav Nutr Phys Act 8, 103. doi: 10.1186/1479-5868-8-103.

174 Nagamatsu, L.S., Voss, M., Neider, M.B., Gaspar, J.G., Handy, T.C., Kramer, A.F., et al. (2011). Increased cognitive load leads to impaired mobility decisions in seniors at risk for falls. Psychol Aging 26, 253-259. doi: 10.1037/a0022929. Naito, E., and Hirose, S. (2014). Efficient foot motor control by Neymar's brain. Front Hum Neurosci 8, 594. doi: 10.3389/fnhum.2014.00594. Nasreddine, Z.S., Phillips, N.A., Bedirian, V., Charbonneau, S., Whitehead, V., Collin, I., et al. (2005). The Montreal Cognitive Assessment, MoCA: a brief screening tool for mild cognitive impairment. J Am Geriatr Soc 53, 695-699. doi: 10.1111/j.1532-5415.2005. 53221.x. Neely, G., Ljunggren, G., Sylven, C., and Borg, G. (1992). Comparison between the Visual Analogue Scale (VAS) and the Category Ratio Scale (CR-10) for the evaluation of leg exertion. Int J Sports Med 13, 133-136. doi: 10.1055/s-2007-1021244. Neider, M.B., Gaspar, J.G., McCarley, J.S., Crowell, J.A., Kaczmarski, H., and Kramer, A.F. (2011). Walking and talking: dual-task effects on street crossing behavior in older adults. Psychol Aging 26, 260-268. doi: 10.1037/a0021566. Nelson, M.E., Rejeski, W.J., Blair, S.N., Duncan, P.W., Judge, J.O., King, A.C., et al. (2007). Physical activity and public health in older adults: recommendation from the American College of Sports Medicine and the American Heart Association. Med Sci Sports Exerc 39, 1435-1445. doi: 10.1249/mss.0b013e3180616aa2. Ng, S.S., Ng, P.C., Lee, C.Y., Ng, E.S., Tong, M.H., Fong, S.S., et al. (2013). Assessing the walking speed of older adults: the influence of walkway length. Am J Phys Med Rehabil 92, 776- 780. doi: 10.1097/PHM.0b013e31828769d0. Niemann, C., Godde, B., Staudinger, U.M., and Voelcker-Rehage, C. (2014). Exercise-induced changes in basal ganglia volume and cognition in older adults. Neuroscience 281, 147- 163. doi: 10.1016/j.neuroscience.2014.09.033. Noack, H., Lovden, M., Schmiedek, F., and Lindenberger, U. (2009). Cognitive plasticity in adulthood and old age: gauging the generality of cognitive intervention effects. Restor Neurol Neurosci 27, 435-453. doi: 10.3233/RNN-2009-0496. O'Dwyer, S. (2009). The effect of exercise training, and combined exercise and cognitive training, on cognitive and physical function in older adults: A randomized controlled trial and qualitative evaluation. Dissertation, University of Queensland. Oberlin, L.E., Verstynen, T.D., Burzynska, A.Z., Voss, M.W., Prakash, R.S., Chaddock-Heyman, L., et al. (2016). White matter microstructure mediates the relationship between cardiorespiratory fitness and spatial working memory in older adults. Neuroimage 131, 91-101. doi: 10.1016/j.neuroimage.2015.09.053. Oei, A.C., and Patterson, M.D. (2014). Are videogame training gains specific or general? Front Syst Neurosci 8, 54. doi: 10.3389/fnsys.2014.00054. Ohsugi, H., Ohgi, S., Shigemori, K., and Schneider, E.B. (2013). Differences in dual-task performance and prefrontal cortex activation between younger and older adults. BMC Neurosci 14, 10. doi: 10.1186/1471-2202-14-10. Ortega-Perez, I., Murray, K., and Lledo, P.M. (2007). The how and why of adult neurogenesis. J Mol Histol 38, 555-562. doi: 10.1007/s10735-007-9114-5.

175 Oswald, W.D., and Fleischmann, U.M. (1997). Das Nürnberger-Alters-Inventar (NAI). Göttingen: Hogrefe. Oswald, W.D., Gunzelmann, T., Rupprecht, R., and Hagen, B. (2006). Differential effects of single versus combined cognitive and physical training with older adults: the SimA study in a 5-year perspective. Eur J Ageing 3, 179-192. Oswald, W.D., Rupprecht, R., Gunzelmann, T., and Tritt, K. (1996). The SIMA-project: effects of 1 year cognitive and psychomotor training on cognitive abilities of the elderly. Behav Brain Res 78, 67-72. Panel on Prevention of Falls in Older Persons, A.G.S., and British Geriatrics, S. (2011). Summary of the Updated American Geriatrics Society/British Geriatrics Society clinical practice guideline for prevention of falls in older persons. J Am Geriatr Soc 59, 148-157. doi: 10.1111/j.1532-5415.2010.03234.x. Papp, K.V., Walsh, S.J., and Snyder, P.J. (2009). Immediate and delayed effects of cognitive interventions in healthy elderly: a review of current literature and future directions. Alzheimers Dement 5, 50-60. doi: 10.1016/j.jalz.2008.10.008. Peel, N.M., Kuys, S.S., and Klein, K. (2013). Gait speed as a measure in geriatric assessment in clinical settings: a systematic review. J Gerontol A Biol Sci Med Sci 68, 39-46. doi: 10.1093/gerona/gls174. Perera, S., Mody, S.H., Woodman, R.C., and Studenski, S.A. (2006). Meaningful change and responsiveness in common physical performance measures in older adults. J Am Geriatr Soc 54, 743-749. doi: 10.1111/j.1532-5415.2006.00701.x. Perneger, T.V. (1998). What's wrong with Bonferroni adjustments. BMJ 316, 1236-1238. Perrey, S. (2008). Non-invasive NIR spectroscopy of human brain function during exercise. Methods 45, 289-299. doi: 10.1016/j.ymeth.2008.04.005. Pichierri, G., Coppe, A., Lorenzetti, S., Murer, K., and de Bruin, E.D. (2012a). The effect of a cognitive-motor intervention on voluntary step execution under single and dual task conditions in older adults: a randomized controlled pilot study. Clin Interv Aging 7, 175-184. doi: 10.2147/CIA.S32558. Pichierri, G., Murer, K., and de Bruin, E.D. (2012b). A cognitive-motor intervention using a dance video game to enhance foot placement accuracy and gait under dual task conditions in older adults: a randomized controlled trial. BMC Geriatr 12, 74. doi: 10.1186/1471- 2318-12-74. Plummer, P., Zukowski, L.A., Giuliani, C., Hall, A.M., and Zurakowski, D. (2015). Effects of Physical Exercise Interventions on Gait-Related Dual-Task Interference in Older Adults: A Systematic Review and Meta-Analysis. Gerontology. doi: 10.1159/000371 577. Pro-Senectute (2017). Pro Senectute St.Gallen [homepage on the internet]. Available from: http://www.sg.pro-senectute.ch/ [Accessed July 28, 2017]. Randomization (2017). Randomization.com [homepage on the internet]. Available from: http://randomization.com/ [Accessed July 28, 2017]. Raz, N., and Kennedy, K.M. (2009). "A systems approach to the aging brain: Neuroanatomic changes, their modifiers, and cognitive correlates," in Imaging the Aging Brain, eds. W. Jagust & M. D'Esposito. 1 ed (New York, NY: Oxford University Press), 43-70.

176 Reuter-Lorenz, P.A., and Cappell, K.A. (2008). Neurocognitive aging and the compensation hypothesis. Current Directions in Psychological Science 17, 177-182. doi: DOI 10.1111/j.1467-8721.2008.00570.x. Robitaille, A., Muniz, G., Lindwall, M., Piccinin, A.M., Hoffman, L., Johansson, B., et al. (2014). Physical activity and cognitive functioning in the oldest old: Within- and between- person cognitive activity and psychosocial mediators. Eur J Ageing 11, 333-347. doi: 10.1007/s10433-014-0314-z. Roig, M., Nordbrandt, S., Geertsen, S.S., and Nielsen, J.B. (2013). The effects of cardiovascular exercise on human memory: a review with meta-analysis. Neurosci Biobehav Rev 37, 1645-1666. doi: 10.1016/j.neubiorev.2013.06.012. Romero-Ortuno, R., Cogan, L., Cunningham, C.U., and Kenny, R.A. (2010). Do older pedestrians have enough time to cross roads in Dublin? A critique of the Traffic Management Guidelines based on clinical research findings. Age Ageing 39, 80-86. doi: 10.1093/ageing/afp206. Roth, S.M., Ivey, F.M., Martel, G.F., Lemmer, J.T., Hurlbut, D.E., Siegel, E.L., et al. (2001). Muscle size responses to strength training in young and older men and women. J Am Geriatr Soc 49, 1428-1433. Ryan, S.M., and Nolan, Y.M. (2016). Neuroinflammation negatively affects adult hippocampal neurogenesis and cognition: can exercise compensate? Neurosci Biobehav Rev 61, 121- 131. doi: 10.1016/j.neubiorev.2015.12.004. Salbach, N.M., O'Brien, K., Brooks, D., Irvin, E., Martino, R., Takhar, P., et al. (2014). Speed and distance requirements for community ambulation: a systematic review. Arch Phys Med Rehabil 95, 117-128 e111. doi: 10.1016/j.apmr.2013.06.017. Salthouse, T.A. (2011). Neuroanatomical substrates of age-related cognitive decline. Psychol Bull 137, 753-784. doi: 10.1037/a0023262. Sartor, J., Bettecken, K., Bernhard, F.P., Hofmann, M., Gladow, T., Lindig, T., et al. (2017). White Matter Changes-Related Gait and Executive Function Deficits: Associations with Age and Parkinson's Disease. Front Aging Neurosci 9, 213. doi: 10.3389/fnagi. 2017.00213. Scafato, E., Gandin, C., Galluzzo, L., Ghirini, S., Cacciatore, F., Capurso, A., et al. (2010). Prevalence of aging-associated cognitive decline in an Italian elderly population: results from cross-sectional phase of Italian PRoject on Epidemiology of Alzheimer's disease (IPREA). Aging Clin Exp Res 22, 440-449. doi: 10.3275/6970. Schaefer, S., and Schumacher, V. (2011). The interplay between cognitive and motor functioning in healthy older adults: findings from dual-task studies and suggestions for intervention. Gerontology 57, 239-246. doi: 10.1159/000322197. Schättin, A., Arner, R., Gennaro, F., and de Bruin, E.D. (2016). Adaptations of Prefrontal Brain Activity, Executive Functions, and Gait in Healthy Elderly Following Exergame and Balance Training: A Randomized-Controlled Study. Front Aging Neurosci 8, 278. doi: 10.3389/fnagi.2016.00278. Scherder, E., Eggermont, L., Visscher, C., Scheltens, P., and Swaab, D. (2011). Understanding higher level gait disturbances in mild dementia in order to improve rehabilitation: 'last in-first out'. Neurosci Biobehav Rev 35, 699-714. doi: 10.1016/j.neubiorev. 2010.08.009.

177 Schoene, D., Valenzuela, T., Toson, B., Delbaere, K., Severino, C., Garcia, J., et al. (2015). Interactive Cognitive-Motor Step Training Improves Cognitive Risk Factors of Falling in Older Adults - A Randomized Controlled Trial. PLoS One 10, e0145161. doi: 10.1371/journal.pone.0145161. Scholkmann, F., Kleiser, S., Metz, A.J., Zimmermann, R., Mata Pavia, J., Wolf, U., et al. (2014). A review on continuous wave functional near-infrared spectroscopy and imaging instrumentation and methodology. Neuroimage 85 Pt 1, 6-27. doi: 10.1016/j.neuroimage.2013.05.004. Scholz, J., Klein, M.C., Behrens, T.E., and Johansen-Berg, H. (2009). Training induces changes in white-matter architecture. Nat Neurosci 12, 1370-1371. doi: 10.1038/nn.2412. Seco, J., Abecia, L.C., Echevarria, E., Barbero, I., Torres-Unda, J., Rodriguez, V., et al. (2013). A long-term physical activity training program increases strength and flexibility, and improves balance in older adults. Rehabil Nurs 38, 37-47. doi: 10.1002/rnj.64. Shams, T.A., Foussias, G., Zawadzki, J.A., Marshe, V.S., Siddiqui, I., Muller, D.J., et al. (2015). The Effects of Video Games on Cognition and Brain Structure: Potential Implications for Neuropsychiatric Disorders. Curr Psychiatry Rep 17, 71. doi: 10.1007/s11920-015- 0609-6. Sherrington, C., Tiedemann, A., Fairhall, N., Close, J.C., and Lord, S.R. (2011). Exercise to prevent falls in older adults: an updated meta-analysis and best practice recommendations. N S W Public Health Bull 22, 78-83. doi: 10.1071/NB10056. Shibuya-Tayoshi, S., Sumitani, S., Kikuchi, K., Tanaka, T., Tayoshi, S., Ueno, S., et al. (2007). Activation of the prefrontal cortex during the Trail-Making Test detected with multichannel near-infrared spectroscopy. Psychiatry Clin Neurosci 61, 616-621. doi: 10.1111/j.1440-1819.2007.01727.x. Smith, E., Cusack, T., and Blake, C. (2016). The effect of a dual task on gait speed in community dwelling older adults: A systematic review and meta-analysis. Gait Posture 44, 250- 258. doi: 10.1016/j.gaitpost.2015.12.017. Smith, P.J., Blumenthal, J.A., Hoffman, B.M., Cooper, H., Strauman, T.A., Welsh-Bohmer, K., et al. (2010). Aerobic exercise and neurocognitive performance: a meta-analytic review of randomized controlled trials. Psychosom Med 72, 239-252. doi: 10.1097/ PSY.0b013e3181d14633. Stadt-SG (2016). Location, figures and facts [homepage on the internet]. City of St.Gallen, Fachstelle für Statistik Kanton St.Gallen. Available from: http://www.stadt.sg.ch/ home/welcome/st-gallen/geografische-lage-zahlen-fakten.html [Accessed July 28, 2017]. Stel, V.S., Smit, J.H., Pluijm, S.M., and Lips, P. (2004). Consequences of falling in older men and women and risk factors for health service use and functional decline. Age Ageing 33, 58-65. Stenhagen, M., Ekstrom, H., Nordell, E., and Elmstahl, S. (2013). Falls in the general elderly population: a 3- and 6- year prospective study of risk factors using data from the longitudinal population study 'Good ageing in Skane'. BMC Geriatr 13, 81. doi: 10.1186/1471-2318-13-81.

178 StepMania (2016). Stepmania.com [homepage on the internet]. Available from: http://www. stepmania.com/ [Accessed July 28, 2017]. Stern, Y. (2009). Cognitive reserve. Neuropsychologia 47, 2015-2028. doi: 10.1016/j.neuro psychologia.2009.03.004. Stillman, C.M., Cohen, J., Lehman, M.E., and Erickson, K.I. (2016). Mediators of Physical Activity on Neurocognitive Function: A Review at Multiple Levels of Analysis. Front Hum Neurosci 10, 626. doi: 10.3389/fnhum.2016.00626. Streiner, D.L., and Norman, G.R. (2011). Correction for multiple testing: is there a resolution? Chest 140, 16-18. doi: 10.1378/chest.11-0523. Studenski, S. (2009). Bradypedia: is gait speed ready for clinical use? J Nutr Health Aging 13, 878-880. Studenski, S., Perera, S., Wallace, D., Chandler, J.M., Duncan, P.W., Rooney, E., et al. (2003). Physical performance measures in the clinical setting. J Am Geriatr Soc 51, 314-322. Sun, Q., Townsend, M.K., Okereke, O.I., Franco, O.H., Hu, F.B., and Grodstein, F. (2010). Physical activity at midlife in relation to successful survival in women at age 70 years or older. Arch Intern Med 170, 194-201. doi: 10.1001/archinternmed.2009.503. Suzuki, M., Miyai, I., Ono, T., Oda, I., Konishi, I., Kochiyama, T., et al. (2004). Prefrontal and premotor cortices are involved in adapting walking and running speed on the treadmill: an optical imaging study. Neuroimage 23, 1020-1026. doi: 10.1016/j.neuroimage.2004.07.002. Tamnes, C.K., Walhovd, K.B., Dale, A.M., Ostby, Y., Grydeland, H., Richardson, G., et al. (2013). Brain development and aging: overlapping and unique patterns of change. Neuroimage 68, 63-74. doi: 10.1016/j.neuroimage.2012.11.039. Taube, W., Gruber, M., and Gollhofer, A. (2008). Spinal and supraspinal adaptations associated with balance training and their functional relevance. Acta Physiol (Oxf) 193, 101-116. doi: 10.1111/j.1748-1716.2008.01850.x. ten Brinke, L.F., Bolandzadeh, N., Nagamatsu, L.S., Hsu, C.L., Davis, J.C., Miran-Khan, K., et al. (2015). Aerobic exercise increases hippocampal volume in older women with probable mild cognitive impairment: a 6-month randomised controlled trial. Br J Sports Med 49, 248-254. doi: 10.1136/bjsports-2013-093184. Theill, N., Schumacher, V., Adelsberger, R., Martin, M., and Jancke, L. (2013). Effects of simultaneously performed cognitive and physical training in older adults. BMC Neurosci 14, 103. doi: 10.1186/1471-2202-14-103. Thom, J.M., and Clare, L. (2011). Rationale for combined exercise and cognition-focused interventions to improve functional independence in people with dementia. Gerontology 57, 265-275. doi: 10.1159/000322198. Thomas, A.G., Dennis, A., Bandettini, P.A., and Johansen-Berg, H. (2012). The effects of aerobic activity on brain structure. Front Psychol 3, 86. doi: 10.3389/fpsyg. 2012.00086. Tinetti, M.E., and Kumar, C. (2010). The patient who falls: "It's always a trade-off". JAMA 303, 258-266. doi: 10.1001/jama.2009.2024.

179 Tinetti, M.E., Speechley, M., and Ginter, S.F. (1988). Risk factors for falls among elderly persons living in the community. N Engl J Med 319, 1701-1707. doi: 10.1056/NEJM 198812293192604. Tong, Y., and Frederick, B.D. (2010). Time lag dependent multimodal processing of concurrent fMRI and near-infrared spectroscopy (NIRS) data suggests a global circulatory origin for low-frequency oscillation signals in human brain. Neuroimage 53, 553-564. doi: 10.1016/j.neuroimage.2010.06.049. Toraman, N.F., and Ayceman, N. (2005). Effects of six weeks of detraining on retention of functional fitness of old people after nine weeks of multicomponent training. Br J Sports Med 39, 565-568; discussion 568. doi: 10.1136/bjsm.2004.015586. Tournier, I., Dommes, A., and Cavallo, V. (2016). Review of safety and mobility issues among older pedestrians. Accid Anal Prev 91, 24-35. doi: 10.1016/j.aap.2016.02.031. Tracy, B.L., Ivey, F.M., Hurlbut, D., Martel, G.F., Lemmer, J.T., Siegel, E.L., et al. (1999). Muscle quality. II. Effects Of strength training in 65- to 75-yr-old men and women. J Appl Physiol (1985) 86, 195-201. Turner, G., Clegg, A., British Geriatrics, S., Age, U.K., and Royal College of General, P. (2014). Best practice guidelines for the management of frailty: a British Geriatrics Society, Age UK and Royal College of General Practitioners report. Age Ageing 43, 744-747. doi: 10.1093/ageing/afu138. Valenzuela, M., and Sachdev, P. (2009). Can cognitive exercise prevent the onset of dementia? Systematic review of randomized clinical trials with longitudinal follow-up. Am J Geriatr Psychiatry 17, 179-187. doi: 10.1097/JGP.0b013e3181953b57. Van Abbema, R., De Greef, M., Craje, C., Krijnen, W., Hobbelen, H., and Van Der Schans, C. (2015). What type, or combination of exercise can improve preferred gait speed in older adults? A meta-analysis. BMC Geriatr 15, 72. doi: 10.1186/s12877-015-0061-9. van het Reve, E., and de Bruin, E.D. (2014). Strength-balance supplemented with computerized cognitive training to improve dual task gait and divided attention in older adults: a multicenter randomized-controlled trial. BMC Geriatr 14, 134. doi: 10.1186/1471-2318- 14-134. van Praag, H. (2008). Neurogenesis and exercise: past and future directions. Neuromolecular Med 10, 128-140. doi: 10.1007/s12017-008-8028-z. van Praag, H., Christie, B.R., Sejnowski, T.J., and Gage, F.H. (1999). Running enhances neurogenesis, learning, and long-term potentiation in mice. Proc Natl Acad Sci U S A 96, 13427-13431. van Praag, H., Shubert, T., Zhao, C., and Gage, F.H. (2005). Exercise enhances learning and hippocampal neurogenesis in aged mice. J Neurosci 25, 8680-8685. doi: 10.1523/ JNEUROSCI.1731-05.2005. van Uden, C.J., and Besser, M.P. (2004). Test-retest reliability of temporal and spatial gait characteristics measured with an instrumented walkway system (GAITRite). BMC Musculoskelet Disord 5, 13. doi: 10.1186/1471-2474-5-13. Vaynman, S., Ying, Z., and Gomez-Pinilla, F. (2004). Hippocampal BDNF mediates the efficacy of exercise on synaptic plasticity and cognition. Eur J Neurosci 20, 2580-2590. doi: 10.1111/j.1460-9568.2004.03720.x.

180 Verlinden, V.J., van der Geest, J.N., Hofman, A., and Ikram, M.A. (2014). Cognition and gait show a distinct pattern of association in the general population. Alzheimers Dement 10, 328-335. doi: 10.1016/j.jalz.2013.03.009. Verstynen, T.D., Lynch, B., Miller, D.L., Voss, M.W., Prakash, R.S., Chaddock, L., et al. (2012). Caudate Nucleus Volume Mediates the Link between Cardiorespiratory Fitness and Cognitive Flexibility in Older Adults. J Aging Res 2012, 939285. doi: 10.1155/2012/ 939285. Vitorio, R., Stuart, S., Rochester, L., Alcock, L., and Pantall, A. (2017). fNIRS response during walking - Artefact or cortical activity? A systematic review. Neurosci Biobehav Rev 83, 160-172. doi: 10.1016/j.neubiorev.2017.10.002. Vivar, C., Potter, M.C., and van Praag, H. (2013). All about running: synaptic plasticity, growth factors and adult hippocampal neurogenesis. Curr Top Behav Neurosci 15, 189-210. doi: 10.1007/7854_2012_220. Voelcker-Rehage, C., Godde, B., and Staudinger, U.M. (2010). Physical and motor fitness are both related to cognition in old age. Eur J Neurosci 31, 167-176. doi: 10.1111/j.1460- 9568.2009.07014.x. Voelcker-Rehage, C., Godde, B., and Staudinger, U.M. (2011). Cardiovascular and coordination training differentially improve cognitive performance and neural processing in older adults. Front Hum Neurosci 5, 26. doi: 10.3389/fnhum.2011.00026. Voelcker-Rehage, C., and Niemann, C. (2013). Structural and functional brain changes related to different types of physical activity across the life span. Neurosci Biobehav Rev 37, 2268-2295. doi: 10.1016/j.neubiorev.2013.01.028. von Elm, E., Altman, D.G., Egger, M., Pocock, S.J., Gotzsche, P.C., Vandenbroucke, J.P., et al. (2007). The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies. PLoS Med 4, e296. doi: 10.1371/journal.pmed.0040296. Voss, M.W., Heo, S., Prakash, R.S., Erickson, K.I., Alves, H., Chaddock, L., et al. (2013). The influence of aerobic fitness on cerebral white matter integrity and cognitive function in older adults: results of a one-year exercise intervention. Hum Brain Mapp 34, 2972- 2985. doi: 10.1002/hbm.22119. Voss, M.W., Nagamatsu, L.S., Liu-Ambrose, T., and Kramer, A.F. (2011). Exercise, brain, and cognition across the life span. J Appl Physiol (1985) 111, 1505-1513. doi: 10.1152/ japplphysiol.00210.2011. Voss, M.W., Prakash, R.S., Erickson, K.I., Basak, C., Chaddock, L., Kim, J.S., et al. (2010). Plasticity of brain networks in a randomized intervention trial of exercise training in older adults. Front Aging Neurosci 2. doi: 10.3389/fnagi.2010.00032. VSS (1992). Lichtsignalanlagen: Übergangszeiten und Mindestzeiten [homepage on the internet]. Vereinigung Schweizerischer Strassenfachleute. Available from: http://www.vss.ch/de/webviewer/viewdocument/4096/dHash/e93858adbefa1adbf71b28 99ba382451a7133eb2/ [Accessed July 28, 2017]. Walhovd, K.B., Fjell, A.M., and Espeseth, T. (2014). Cognitive decline and brain pathology in aging - need for a dimensional, lifespan and systems vulnerability view. Scand J Psychol 55, 244-254. doi: 10.1111/sjop.12120.

181 Wallois, F., Mahmoudzadeh, M., Patil, A., and Grebe, R. (2012). Usefulness of simultaneous EEG-NIRS recording in language studies. Brain Lang 121, 110-123. doi: 10.1016/ j.bandl.2011.03.010. Webster, K.E., Wittwer, J.E., and Feller, J.A. (2005). Validity of the GAITRite walkway system for the measurement of averaged and individual step parameters of gait. Gait Posture 22, 317-321. doi: 10.1016/j.gaitpost.2004.10.005. Wechsler, D. (1981). Manual for the Wechsler Adult Intelligence Scale-Revised. New York: Psychological Corporation. Wechsler, D. (1987). Wechsler Memory Scale - Revised (WMS-R) Manual. San Antonio: The Psychological Corporation, Harcourt Brace Jovanovich. Weinstein, A.M., Voss, M.W., Prakash, R.S., Chaddock, L., Szabo, A., White, S.M., et al. (2012). The association between aerobic fitness and executive function is mediated by prefrontal cortex volume. Brain Behav Immun 26, 811-819. doi: 10.1016/j.bbi.2011. 11.008. West, R.L. (1996). An application of prefrontal cortex function theory to cognitive aging. Psychol Bull 120, 272-292. WHO (2016). Falls, fact sheet [homepage on the internet]. World Health Organization. Available from: http://www.who.int/mediacentre/factsheets/fs344/en/ [Accessed July 28, 2017]. Wicking, M., Nees, F., and Steiger, F. (2014). Neuropsychological measures of hippocampal function. Front Neurol Neurosci 34, 60-70. doi: 10.1159/000356425. Wilckens, K.A., Erickson, K.I., and Wheeler, M.E. (2016). Physical Activity and Cognition: A Mediating Role of Efficient Sleep. Behav Sleep Med, 1-18. doi: 10.1080/15402002. 2016.1253013. Williams, K.N., and Kemper, S. (2010). Interventions to reduce cognitive decline in aging. J Psychosoc Nurs Ment Health Serv 48, 42-51. doi: 10.3928/02793695-20100331-03. Williamson, J.D., Espeland, M., Kritchevsky, S.B., Newman, A.B., King, A.C., Pahor, M., et al. (2009). Changes in cognitive function in a randomized trial of physical activity: results of the lifestyle interventions and independence for elders pilot study. J Gerontol A Biol Sci Med Sci 64, 688-694. doi: 10.1093/gerona/glp014. Wilson, B.A., Evans, J.J., Emslie, H., Alderman, N., and Burgess, P. (1998). The development of an ecologically valid test for assessing patients with dysexecutive syndrome. Neuropsychol Rehabil 8, 213-228. Woollacott, M., and Shumway-Cook, A. (2002). Attention and the control of posture and gait: a review of an emerging area of research. Gait Posture 16, 1-14. Yardley, L., Beyer, N., Hauer, K., Kempen, G., Piot-Ziegler, C., and Todd, C. (2005). Development and initial validation of the Falls Efficacy Scale-International (FES-I). Age Ageing 34, 614-619. doi: 10.1093/ageing/afi196. Yogev-Seligmann, G., Hausdorff, J.M., and Giladi, N. (2008). The role of executive function and attention in gait. Mov Disord 23, 329-342; quiz 472. doi: 10.1002/mds.21720. Zama, T., and Shimada, S. (2015). Simultaneous measurement of electroencephalography and near-infrared spectroscopy during voluntary motor preparation. Sci Rep 5, 16438. doi: 10.1038/srep16438.

182 Zelinski, E.M. (2009). Far transfer in cognitive training of older adults. Restor Neurol Neurosci 27, 455-471. doi: 10.3233/RNN-2009-0495. Zimmerman, M.E., Lipton, R.B., Pan, J.W., Hetherington, H.P., and Verghese, J. (2009). MRI- and MRS-derived hippocampal correlates of quantitative locomotor function in older adults. Brain Res 1291, 73-81. doi: 10.1016/j.brainres.2009.07.043. Zito, G.A., Cazzoli, D., Scheffler, L., Jager, M., Muri, R.M., Mosimann, U.P., et al. (2015). Street crossing behavior in younger and older pedestrians: an eye- and head-tracking study. BMC Geriatr 15, 176. doi: 10.1186/s12877-015-0175-0. Zornetzer, S.F. (1985). Catecholamine system involvement in age-related memory dysfunction. Ann N Y Acad Sci 444, 242-254. Zwergal, A., Linn, J., Xiong, G., Brandt, T., Strupp, M., and Jahn, K. (2012). Aging of human supraspinal locomotor and postural control in fMRI. Neurobiol Aging 33, 1073-1084. doi: 10.1016/j.neurobiolaging.2010.09.022.

183

184 9

Acknowledgments

185 It was my great pleasure having had Prof. Dr. Eling D. de Bruin as a committed supervisor of my Ph.D. thesis. Thank you for sharing your vast knowledge about sound scientific working processes and your passion for research with older adults. I appreciated your valuable assistance for setting up meaningful research questions and study designs, while also providing the opportunity to introduce and prove own ideas. I was glad of your support for participating in many conferences and presenting my work to the scientific community, which included traveling to Barcelona, Amsterdam, Köln, and Bologna. Your feedback about my work was always well-considered, construc- tive, and guided me to improve myself. Our promising results definitely warrant further research and therefore I’m looking forward to pursue common research projects also in the future. It was a privilege having had you as my supervisor – thank you!

Prof. Dr. Nicole Wenderoth, head of the Institute of Human Movement Sciences and Sport, ETH Zurich and Prof. Dr. Jorunn L. Helbostad, head of the Department of Neuromedicine and Movement Science, NTNU Trond- heim, Norway, friendly accepted to act as co-examiners for my Ph.D. thesis. I appreciated your support and your interest in my work.

Sincere thanks to Prof. Dr. em. Kurt Murer, as well as to his successor Prof. Dr. Katrien De Bock, for providing me the opportunity to pursue a Ph.D. at the Institute of Human Movement Sciences and Sport.

This work was supported by the Zürcher Kantonalbank within the framework of sponsoring of movement sciences, sports and nutritional sciences at ETH Zurich. Zürcher Kantonalbank had no influence on the study design and the analyses presented in this thesis, had no access to the data, and did not contribute to this thesis manuscript in any way. I’m thankful for this major funding contribution that enabled my Ph.D. position.

I would like to thank the following persons and their related institutions: PD Dr. med Thomas Münzer, chief physician, and the management of Geriat- rische Klinik St.Gallen who supported my studies and provided room for training and data acquisition. I particularly appreciated the kind support of their teams of physiotherapists, including Carmen Fürer and Carole Scheidegger who approached us first with the idea of performing a study at their institution, as well as the technical services and secretariats. Pro

186 Senectute St.Gallen and the Department of Sports of the canton St.Gallen were very helpful with the search for the many participants.

Many thanks belong to my external co-authors of the publications presented in this thesis: Dr. Vera Schumacher (Department of Psychology and Ger- ontology, University of Zurich), Dr. Nathan Theill (Division of Psychiatry Research, University of Zurich), Prof. Dr. Martin Wolf (Biomedical Optics Research Laboratory, Department of Neonatology, University Hospital Zur- ich), and PD Dr. med Thomas Münzer (Geriatrische Klinik St.Gallen) provided critical and encouraging feedback on my studies, manifold scientific advice, sophisticated statistical ideas, and valuable methodological inputs.

Very importantly, I owe much to my postgraduate students, Marius Angst, Anna Balmelli, Jana Bucher, Stefan Holenstein, Fabienne Hüppin, Nadine Jenni, Manuela Kobelt, David Niedermann, Natalie Müller, Alexandra Schättin, Martina Schumann, and Sara Tomovic, for their tremendous and committed contribution to the presented studies by sharing ideas, instructing numerous training and testing sessions, as well as helping with data processing and participant recruitment.

I could not omit thanking every participant – 162 in total – for her/his enthusiasm, kindness, patience, and perseverance during our extensive training and testing sessions. Me and my postgraduate students were happy and privileged to work with you. Without your participation, this doctoral thesis and the related M.Sc. theses would not have been possible.

To the members of the study administration, study coordination, IT support, and teaching of the Department of Health Sciences and Technology, Regula Blaser, Yvonne Meier, Dr. Roland Müller, Simon Polomski, and Sarah Frédérickx, I am grateful for their kind assistance in many different mat- ters. The postdocs Dr. Rolf van de Langenberg, Dr. Laura Tomatis, and Dr. Brigitte Wirth were as well a valuable source for scientific advice. It was a particular pleasure to attend Rolf’s lectures on statistics and Matlab pro- gramming. Furthermore, I appreciated the sporadic chats and WOKA lunches with the staff members from the teaching diploma and didactics group Oliver Graf, Andreas Krebs, and Samuel Maurer.

Last but not least, it was always delighting to gather with my Ph.D. office mates: thank you Giusi for having generated numerous StepMania dances I

187 could use for my studies and for having shared your experience on “how-to do a Ph.D.”; thank you Andrea for having given me helpful advice when I was a rookie Ph.D. student; thank you Eva for having been a reliable assistance for when GaitRite was not operating properly; thank you Seline for the friendly SG vs. LU competition and thanks to both of you, Eva and Seline, for having been part of our fun “school trip” to the Mobex meeting in Köln together with Eling; thank you Rahel for having been my all-knowing statistics expert; thank you Alexandra for first having been a helpful postgraduate intern within my study in St.Gallen and then having enriched our group as a dedicated Ph.D. student; thank you Federico for many inspiring discussions about EEG measurements and best journals to get published (e.g. Vanity Fair Italia), and for bringing some Italian temper to our group; thank you Ma- nuela for your enthusiasm about neuro-cognitive research that you shared with me. You have all been a great inspiration!

And finally, I’d like to thank my Mom for sharing her experience as an expert instructor in sports for older adults; my Dad for fostering my scientific aware- ness since I was a kid and for asking questions about my studies that I would not have come up with; my sister Márcia and my brother-in-law Urs for their advice on how a scientific poster should look like; Ursi and Jürg for being very kind host parents having provided me an accommodation close to ETH; as well as the management and the team of physical education teachers at Cantonal School Trogen for the support and for keeping me involved in practical school sports activities.

188