The impact of a home-based computerized cognitive training intervention on fall risk measure performance in community dwelling older adults, a pilot study

Abstract

Objectives

Cognitive intervention studies have reported improvements in various domains of cognition as well as a transfer effect of improved function post training. Despite the availability of web based cognitive training programs, most intervention studies have been performed under the supervision of researchers. Therefore, the purpose of this study was to first, examine the feasibility of a six week home based computerized cognitive training (CCT) program in a group of community dwelling older adults and, second, to determine if a CCT program which focused on set shifting, attention, and visual spatial ability impacted fall risk measure performance.

Design

This pilot study used a pretest/posttest experimental design with randomization by testing site to an intervention or control group.

Participants

Community dwelling older adults (mean age = 74.6 years) participated in either the control (N=25) or the intervention group (N=19).

Intervention

Intervention group subjects participated in 6 weeks of home based CCT 3x/week for an average of 23 minutes/session, using an online CCT program.

Measurements

Comparisons of mean scores on three measures of physical function (usual gait speed, five times sit to stand, timed up and go) were completed at baseline and week 7.

Results

Following the completion of an average of 18 sessions of CCT at home with good adherence (86%) and retention (92%) rates, a statistically significant difference in gait speed was found between groups with an average improvement of 0.14m/s in the intervention group.

Conclusion

A home based CCT program is a feasible approach to targeting cognitive impairments known to influence fall risk and changes in gait in older adults.

Introduction

The largest growing segment of the population in the United States includes those 65 years and older with estimations that 72 million Americans will be over the age of 65 by the year 2030 (1). An increasing prevalence of age-associated diseases and disorders including cognitive disorders will accompany that population growth (2) which will inevitably place a strain on available health care resources (3, 4).

Recently a new area of study has emerged that has focused on the cognitive processes which influence balance and stability in older adults with impaired cognition. Cognitive impairments, specifically deficits in executive function, have been associated with an increased number of falls, (5) abnormal gait, (6) impaired performance on instrumental activities of daily living (7) and other mobility tasks (8, 9). Executive function (EF) is a higher order of cognition defined by the ability to control, integrate, organize, and maintain information when continuously presented (10). EF has been found to decline with aging due to changes in the prefrontal cortex of the brain (10). Intact EF requires appropriate coordination of various subcomponents of cognition (e.g. attention, set shifting, visual spatial ability) in order to complete common and complex tasks (11). Deficits in set shifting and processing speed have been associated with decreased walking speeds, changes in gait stability, and an increased risk of falling in older adults (12). Intact visual spatial ability is necessary for safe functional mobility (13) and has been found to decline rapidly in the preclinical period of Alzheimer’s disease (14). The impact that cognitive impairments have on function and fall risk has become more evident resulting in an increased amount of research performed that addresses interventions to prevent, maintain, and improve cognition.

Recently, cognitive intervention efforts have concentrated on the maintenance of current cognitive capacities and or the remediation of cognitive deficits with the overall goal of reducing the risk of future cognitive decline. Cognitive training interventions have been successful in addressing declines in cognition, (15, 16) including reported improvements in those with early cognitive loss (17, 18). Specificity of cognitive training interventions is believed to be the contributing factor to those improvements as a domain specific approach to training has reportedly provided the most optimal outcomes (16, 17). Cognitive training has been reported to be effective in maintaining the cognitive vitality of healthy older adults and it has been suggested that it may serve to ‘optimize’ the cognitive functioning of persons with early cognitive loss and perhaps contribute to a slowing of cognitive decline and the onset of disability (19). Computer based cognitive training interventions offer the benefit of a cost-effective, performance based and accessible approach to addressing cognitive decline in community dwelling older adults.

Studies that have employed cognitive training interventions which have targeted cognitive domains believed to influence balance or gait have reported a transfer effect into improvements in function. This was first reported in a large randomized control trial, the Advanced Cognitive Training for Independent and Vital Elderly study (N=2398), where the completion of progressively complex speed tasks on a computer was associated with less difficulty with instrumental activities of daily living in subjects after completion of the training and at both the five and ten year follow up (20, 21). Further, improvements in gait speed and on measures of balance have been reported in older adults after participating in a progressively challenging community based computerized cognitive training (CCT) program (22, 23, 24). Due to the increased likelihood of having a decline in cognitive function in advanced age (25) and the influence of cognitive loss on fall risk, (5) interventions that address early cognitive loss which can be done outside of supervised clinical settings have an opportunity to improve the health and quality of life in older adults.

Studies addressing the effect of cognitive interventions on improvements in physical function are preliminary in nature, but they support the idea that a domain specific CCT program focusing on components of EF (visual spatial ability, set shifting, and attention) may improve the cognitive influence on gait, and perhaps reduce the risk of falls (13, 14, 26). Whereas previous studies have been performed in community based settings under the supervision of study examiners, it is unknown if the transfer effect of improvements in balance and gait will be present in those who have completed a home based CCT program. Feedback provided by examiners, motivational support of peers, and the emphasis of the applicability of cognitive training to daily activities by study personnel have been identified as potential factors which may influence outcomes in community based CCT programs (27). Therefore, the purpose of this study was to first, examine the feasibility of a six week home based CCT program in a group of community dwelling older adults and, second, to determine if a CCT program which focused on set shifting, attention, and visual spatial ability impacted fall risk measure performance.

Methods

A pretest-posttest experimental design was used to assess the influence of CCT on measures of fall risk in two groups of community dwelling older adults. Recruitment occurred at three different senior centers in the following order: 1. an announcement regarding a study information session was posted in the facility’s paper and/or electronic newsletter one month prior to the session, 2. a 30 minute presentation about the study was completed by 2 research team members with a question/answer session afterwards, and 3. a written description advertising the study was published in the newsletter and written advertisements were posted in the community rooms for two consecutive months after the presentation. Potential subjects contacted the research team and were scheduled for an eligibility screen. Eligibility criteria included being age 65 or older, independent with transfers and ambulation, able to communicate in English, able to independently consent to participate, were living independently in the community, had good vision with or without prescriptive lenses, and had a positive history of computer use (4 or more hours/week). Participants were excluded from the study if they had a diagnosed cognitive impairment, were currently receiving healthcare services for a mobility impairment, had a history of a cerebrovascular accident, head trauma, traumatic brain injury or had surgery on their legs six months prior to testing, were unable to meet the requirements of testing or the time commitment for the study, or if they were currently using or had a history of using any cognitive training programs nine months prior to the study. Human subject institutional review board approval was obtained from the primary investigator’s institutions.

To minimize the threat to internal validity, group assignment was based on the location of testing. Subjects from one senior center were assigned to the control group, while subjects from the other two centers were assigned to the intervention group. Outcomes were assessed at baseline and in the seventh week of the study, to allow for intervention group members to complete the CCT program.

Testing

Examiners for this study consisted of the primary investigator and two graduate student assistants. All examiners were trained on the research intake process, the administration of the tests and measures, and the use of the computerized software program during a two day training session to assure consistency between examiners. All testing occurred in a quiet room at the senior centers which lasted approximately 30 minutes. Demographic data was gathered first followed by the cognitive assessments and then the physical measures for both groups. Introduction to and training on the cognitive software occurred immediately after testing for those in the intervention group which lasted up to 45 minutes. Examiners were not blinded to experimental conditions as a result of the location of testing but were blinded to pretest assessment scores.

Demographic information was gathered through self-report and included age, race and ethnicity, and education level. Other information gathered from the subjects included medications, medical and surgical history, self-report of their height and weight, and the use of prescription eyewear. Subjects were asked about their independence with instrumental activities of daily living (IADLS), activities of daily living (ADLs), and their frequency of using a computer. History of falling was by self-report and recorded as the number of times the subject had fallen in the past six months (28). A fall was defined as an unintentional loss of balance that led to an unexpected change of position. Vision was screened with an assessment of whether the subject could correctly read printed material of two different sized fonts (12 point and 18 point font).

Two cognitive measures were used (Trail Making Test Part B (TMT-B), Montreal Cognitive Assessment tool (MoCA) [29,30]. Three valid and reliable measures of physical function were used (Five Times Sit to Stand Test, Timed Up and Go, and gait speed) (31, 32, 33). Standardized instructions were provided to all subjects and cognitive assessments were performed first followed by the physical measures during baseline and post-testing. Each subject’s posttest date was scheduled in the seventh week after baseline testing, written down on a study contact sheet, and provided to the subject during the initial testing session. Reminder phone calls of their posttest session were made during the 5th week. Additional questions about whether subjects had started a physical activity program, cognitive training program, used other CCT games or had a significant health change were asked at posttest.

The MoCA is a global measure of MCI and is comprised of items that assess the cognitive domains of short term memory/ recall, visuospatial ability, executive function, attention, concentration, working memory, orientation, and language (30). Performance on the items are added to create a total summed score with a maximum possible score of 30. A score of less than 26 indicates a positive screen for MCI (30).

The TMT-B is a measure of executive function (EF) which requires the cognitive capacities of attention, (34) visual scanning, motor speed and coordination, mental flexibility, (35) and working memory for completion (29, 34, 35). This paper and pencil test required the subjects to create a continuous line (i.e., the trail) connecting randomly positioned numbers (1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12) and letters (A-M) without lifting the pencil in ascending alternating order from number to letter (e.g., 1 to A to 2 to B to 3 etc.) until all numbers and letters were used. Subjects were given one untimed trial of the TMT-B to assure compliance with directions and familiarity with the task. During both the trial and the test, the researcher observed the creation of the trail, immediately identified any errors, and instructed the subject to correct the error (36). Following the clarification of any questions, subjects were given a different version of the TMT-B, placed upside down in front of the subject. Subjects were told to immediately begin making the trail once the paper was flipped over starting the trail on number 1. Timing began from the moment the subject placed his pencil on the starting number and ended when the subject connected the last letter in the trail. The total completion time was recorded in seconds.

Slower gait speeds have been identified as risk factors for disability, cognitive impairment, institutionalization, falls, and/or mortality (33). Gait speed, measured as distance traveled over time, has been found to be a highly reliable (31) predictor of both physical (37) and cognitive decline (38). Gait speed was measured as the amount of time required to walk 3.048 meters (10 feet) at the subject’s normal walking speed which was later converted into meters/second for analyses. To allow for acceleration and deceleration, 1.524 meters (5 feet) of space was provided on each side of the marked distance. Subjects performed this task twice and the mean time of the two trials was recorded.

The Five Times Sit to Stand (FTSTS) test is a valid and reliable tool which functions as an indicator of frailty and a predictor of fall risk. Impaired performance has been reported to be associated with increased disability, morbidity and falls (32, 39). This test measured the amount of time it took for a subject to go from sitting in a supportive chair to full standing five successive times without the use of his arms to aid in the process. Time was measured with a stopwatch from the moment the subject was told to ‘go’ to the moment he sat down in the chair following the fifth repetition. Two measures were performed with the average of the two times being recorded.

The Timed Up and Go (TUG) test is a valid, reliable, and efficient measure of physical function and fall risk in community dwelling older adults which includes both physical and cognitive components (40, 41). This timed test required subjects to go from sitting to standing, walk three meters, turn around and return back to the original chair and sit down again. One trial and one timed test was performed. Timing began from the moment the subject was told to ‘go’ to the moment he sat back down in the chair after walking the set distance (40).

Intervention Condition

Intervention group subjects participated in 6 weeks of CCT using a web based neuropsychological software program, Lumosity (Lumos Labs, CA). An introductory training session occurred immediately after baseline testing to familiarize the subjects on how to log in and use Lumosity. This was conducted in a one on one mode by one member of the research team, herein referred to as the trainer, who was blinded to the baseline measures. Subjects were given a randomly created login name and password. Standardized instructions for playing the games were provided in both verbal and written form, followed by a demonstration of the game by the trainer. Each subject then performed a trial of each game and afterwards, questions about how to play the games were answered by the trainer. Following the initial session, subjects were asked to independently complete each of the games on a computer of their choice at home once per session, three sessions per week for 20-25 minutes/session for a total of six weeks. Subjects were instructed to refrain from completing other Lumosity games and any other forms of cognitive training during the study period. The trainer’s contact information was provided to address any problems using Lumosity throughout the duration of the study. Following the initial training session, no other communication was initiated by any member of the research team with the exception of a phone call in week 5 to remind subjects of their scheduled post testing session.

Games chosen focused on addressing specific domains of EF which have been reported to influence fall risk (13, 14, 27). The four games were: Disconnection which addressed set shifting and attention; Playing Koi which addressed visual spatial ability and attention; Birdwatching which addressed attention and set shifting; and Memory Matrix which addressed visual spatial ability and recall. Responses were recorded and tracked dynamically over time both within a session and across games with the level of challenge being optimized continuously for each user, increasing the challenge as performance improved and backing off when incorrect responses were made which created an individualized training effect. Data gathered from Lumosity included the subject’s progress in each of the games and across all games along with the frequency of sessions completed. The number of completed sessions per week and over the training program as well as the number of games completed per session were summed to determine overall compliance with the intervention. A compliance rate of 70% completion of the program (>14 sessions) was determined a priori as the cut point for inclusion or exclusion of a subject’s data in the final analyses.

Control Condition

A measurement-only control condition was used. Subjects in the control group had measures of cognition and fall risk performed at baseline and at 7 weeks. With the exception of the posttest reminder phone, control group participants were not contacted during the study period. Following study completion, all control group participants were given free access to the cognitive training software by the research team.

Analysis

Descriptive statistics including means, medians, standard deviations, and confidence intervals were performed to describe the data. Between groups differences at baseline of the demographic variables and cognitive and fall risk measures were calculated with the Chi Square statistic for categorical variables and independent t-tests for comparisons of continuous, parametric data or Mann Whitney U (‘exact’ method) for nonparametric comparisons. Differences between groups on measures of fall risk and cognition were performed at baseline and at 7 weeks using individual t-tests for data that met the assumptions for parametric statistics or the Mann Whitney U (‘exact’ method) tests for nonparametric data. To determine the effect size of the results, Cohen’s d was calculated for parametric analyses, while Pearson’s r was used for non-parametric analyses. A significance level of .05 was used for all comparisons and statistical analyses were completed with SPSS version 20 (SPSS Inc., Chicago, IL).

Results

Sixty-one older adults contacted researchers to participate, thirty from the senior center which served as the control, thirty-one from the intervention facilities. Twelve subjects were excluded due to one or more of the following: having a history of a stroke (N=2), Parkinson’s disease (N=1), or diagnosed cognitive impairment (N=3); unable to meet the time commitment for the intervention (N=2) or testing (N=2); less than 65 years of age (N=2); and/or used the computer fewer than 4 hours per week (N=3). Forty-nine met inclusion criteria, provided consent and completed baseline measures with twenty-five subjects assigned to the control group and twenty four assigned to the intervention group.

The intervention group lost five subjects within the study. Two subjects dropped out; one due to computer operating issues and the other due to self-reported noncompliance with the training protocol due to not being able to access a computer during a portion of the study. Two additional subjects did not complete post testing despite being contacted by the researchers as scheduled and, lastly, data from one subject was excluded due to poor compliance (< 50%) with the training protocol. This resulted in data from a total of nineteen intervention group subjects used in the analyses. No control group subjects were lost from the study.

Demographics at baseline for each of the groups can be found in Table 1. All subjects were independent with ADLs and IADLs and passed the vision screening. Individual t-tests or Mann Whitney U revealed no significant differences by group at baseline on demographic variables (age, gender, education), cognitive measures (TMT-B, MoCA) or on outcomes measures (FTSTS, gait speed, TUG). All subjects in both groups reported that they did not initiate any changes in their physical activity level or experience any health related declines throughout the duration of the study. All intervention group subjects reported that they did not play any of the other CCT games during the study.

Table 1.

Demographics, Cognitive Measure Scores, and Fall History of Subjects by Group Assignment (N=44)

	Control Group N=25		Intervention Group N=19
Variable	Mean (SD)	Range	Mean (SD)	Range	p-value
Average age (years)	75.68 (6.88)	66-91	73.79 (4.36)	66-82	.30
Medications (number)	6.38 (5.21)	0-24	6.17 (3.95)	0-13	.88
Body Mass Index kg/m/m	26.93 % (5.75)	15.6-39.4	27.97% (3.77)	19.4-34.8	.51
MoCA (out of 30)	24.40 (3.22)	17-29	25.26 (2.33)	21-28	.32
TMT-B (seconds)	97.93 (36.97)	49.1-164.0	80.68 (37.99)	35.9-175.5	.13
	Total	%	Total	%
Gender					.90
Female	18	72	14	73.7
Male	7	28	5	26.3
Education					.35
High school	3	12.0	2	10.5
Some college	7	28.0	7	36.8
Associate degree	1	4.0	4	21.1
Bachelor’s degree	6	24.0	2	10.5
> Bachelor’s degree	8	32.0	4	21.1
Number of Falls in the 6 Months Prior to Testing					.22
0	18	72.0	8	42.1
1	4	16.0	8	42.1
2	2	8.0	3	15.8
3+	1	4.0	-	-

TMT-B: Trail Making Test part B, MoCA: Montreal Cognitive Assessment Tool

Two subjects in the CCT group contacted the trainer by phone during the study to seek assistance and clarification. One subject had difficulty logging in due to not being able to decipher a letter in his password, the other subject reported to the trainer that he enjoyed the CCT games and wanted to know if he could try other games, to which he was told to refrain from attempting. Besides the post testing reminder phone call in the fifth week, no other contact was made between members of the research team and subjects in either group.

Members of the intervention group completed an average of 18.83 sessions (range = 13-25, SD = 2.85) of CCT for approximately 23 minutes per session, resulting in a mean training dosage of 433.09 minutes (7.22 hours) over six weeks. Statistically significant differences between groups at post-testing were not detected for the cognitive measures; TMT-B t(42) = 1.21, Cohen’s d = 0.37, p = 0.23; MoCA t(42) = -0.25, Cohen’s d = 0.07, p = 0.80.

Of the five intervention group subjects who did not complete post testing, the average number of CCT sessions was 5.2 with a range of 1-11. Comparisons of those who completed the CCT (N=19) and those who did not (N=5) revealed no significant differences on the demographic variables (age, gender, education), cognitive measures (TMT-B, MoCA) or on outcomes measures (FTSTS, gait speed, TUG).

FTSTS

Comparisons between groups indicated that FTSTS performance did not significantly differ t(42) = 1.45, p = 0.15 between measures at baseline and 7 weeks. Mean completion times on the FTSTS improved more for control group subjects (µ = -0.96) than for intervention group subjects (µ = -0.46) (see Table 2). The magnitude of the effect size between the groups was moderate (Cohen’s d = 0.49).

Table 2.

Differences in Physical Performance Measures Between Control and Treatment Arm (N=44)

	Control Group N=25				Intervention Group N=19
	Baseline		7 Weeks		Baseline		7 Weeks
Measure	Mean	SD	Mean	SD	Mean	SD	Mean	SD	Effect Size Cohen’s d	t-value	P-value
FTSTS (seconds)	15.17	4.49	14.21	3.43	13.24	2.46	12.77	2.96	0.49	1.28	0.14
Gait Speed (meters/ second)	0.92	0.26	1.02	0.22	1.06	0.32	1.20	0.30	0.58	-2.13	0.04
TUG (seconds)	11.17	4.59	10.38	2.83	9.47	2.22	9.66	2.40	0.29	0.89	0.38

FTSTS: Five Times Sit to Stand test, TUG: Timed Up and Go test

Gait speed

Results of posttest comparisons indicated that a statistically significant difference was found between groups in gait speed t(42) = -2.13, p = 0.03. Mean gait speeds improved to a greater extent for intervention group subjects (µ = 0.14) than for control group subjects (µ = 0.09) between baseline and 7 weeks (see Table 2). A moderate effect size was found which reflects the differences between the two groups (Cohen’s d = 0.58).

TUG

Results of posttest comparisons indicated that a statistically significant difference between groups was not found for performance on the TUG t(42) = 0.89, p = 0.38. Average TUG completion times between baseline and 7 weeks increased (worsened) for intervention group subjects (µ = 0.18), while mean TUG times decreased (improved) for control group subjects (µ = -0.77) (see Table 2). A small effect size was found between the two groups (Cohen’s d = 0.29).

Discussion

This pilot study found that participating in a six week progressively challenging home based CCT program was a feasible training method to facilitate a significant improvement in gait speed with a moderate effect size. The feasibility of a home based targeted CCT program to address cognitive deficits that influence fall risk is worthy of discussion as previous studies which have reported improved function after CCT have been in supervised settings. This study adds to the existing literature in that a home based cognitive training intervention, targeting the cognitive domains utilized during walking, may improve or slow the decline of walking speeds in community dwelling older adults.

To address the feasibility of performing CCT in a home based setting, we begin with a discussion of our recruitment (42). We intended to recruit thirty subjects from two senior centers which were geographically distant from one another to limit cross contamination between groups, however this did not occur as planned. For the control group, we recruited 30 subjects, but five were not eligible. Likewise, for the intervention group, only 15 of the 20 who expressed an interest were eligible to participate, which resulted in adding another senior center which was equidistant from the other facilities, however we were able to recruit 11 people, of which 9 were eligible. Following the recruitment strategy initially established in the design of the study (newsletter announcement then presentation, followed by passive information in the newsletter), our recruitment time period ended prior to enrolling other subjects. The combination of a newsletter announcement placed one month in advance describing the study’s information session followed by the presentation resulted in the greatest recruitment effort. The newsletter listing likely resulted in senior’s increased knowledge of the study, but did not translate into increased enrollment. We suspect that the recruitment success between different facilities might be attributed to other methods of recruitment (word of mouth, facility advocate) which may have encouraged participation. Nevertheless, we cannot disregard how struggles we encountered in the recruitment phase may have significantly limited the statistical power of our study which, in turn, may have influenced the robustness of our findings. Future studies which aim to employ a home based CCT intervention should consider having multiple face to face information sessions to improve recruitment.

Community based CCT programs offer the benefit of expert supervision of programs, however performing training in that type of setting may actually limit participation or compliance. In a study which employed a community based CCT program (2x/week over 10 weeks), a 36% non-adherence rate was reported which was attributed to the limited staffing available for the training sessions (24). Intervention adherence (86%) rates and retention (92%) rates of this study suggest that completing CCT in the privacy of one’s home may facilitate better compliance with less administrative demand on the study team. As the mean age, education, and gender, did not differ greatly than those in the study above (24), it is possible that other variables may influence adherence rates in a home based CCT program, such as not being exposed to the potential social stigma of completing a cognitive training program in a public forum or the benefit of using one’s own computer in which they are both familiar and comfortable, however this should be examined further in future CCT studies.

The intervention dose of 18.83 CCT sessions over six weeks for a total of 7.22 hours of training was sufficient to significantly improve gait speed. Other studies which have reported improvements in physical mobility after community based CCT interventions have used training frequencies ranging from once per week to 5x/wk with a length of time per session ranging from 15 to 45 minutes over a period of 3-14 weeks (23, 43, 44, 45). However, changes in balance and gait have been reported after 5 to 6 hours of training (24, 46). Given that significant changes in gait speed were found over a shorter duration of CCT with good adherence and low attrition supports the viability of performing home based cognitive interventions to improve gait efficiency. Due to the limitations in power and the possibility of a Type 1 error, future studies which use the same CCT training parameters should include 128 subjects (59 in each group which is adjusted for adherence and retention rates) to find a similar effect (Cohen’s d =.58) in an adequately (80%) powered study.

Previous studies have reported improvements in gait speed following CCT in community dwelling older adults who were classified as slow walkers (22). Verghese and colleagues reported a statistically significant improvement (0.04m/ sec) following the completion of an 8 week supervised CCT program (3x/wk for 45-60 minutes/session) in a group of community dwelling older adults (N=10) whose initial gait speeds were less than 1.0 m/sec (22). Walking speeds of less than 1.0m/sec are believed to be significant predictors of fall risk and physical decline in older adults (33). To our knowledge our study is the first to detect a significant improvement in gait speed in older adults who walked, on average, greater than 1.0 m/sec (µ= 1.06, SD 0.32) following CCT and that the improvement of 0.14 m/sec exceeded the both the minimal detectable change in gait speed as well as other reported gait speed improvements post CCT (22). This finding supports the clinical utility of a home based CCT intervention as an adjunctive way to improve physical mobility. However, because this is a novel finding, further prospective studies should be performed to assess the effects of a home based CCT program focused on EF in those with gait speeds of less than 1.0m/sec as well as those with faster walking speeds.

Some of our results were in contrast to what we would have expected based on previous research (24). TUG scores worsened after CCT whereas FTSTS scores improved in the intervention group, but to a lesser extent than the control group. We postulate that these unexpected findings may have to do with the variability of scores on these measures in both groups. At baseline, TUG scores in the control group had a range which was twice that of the intervention group. Similarly, FTSTS scores in the control group were nearly three times greater than those of the intervention group at baseline and two times greater at post-test measurements. Perhaps using a larger sample (N = 67 per group) with an impairment based cutoff score on these measures might allow for mean scores to regress towards the mean and potentially enhance the ability to detect a moderate (> 0.49) yet significant effect after training.

Some limitations of the study should be acknowledged. Although a moderate effect size in gait speed was found after CCT in the intervention group, this study size was small and underpowered. The study design was flawed in that group assignment was based on location and not on cognitive or physical performance. A convenience sample of mostly white, highly educated older adults from one geographic location participated which limits the generalizability. Close to half (48%) of the subjects in the intervention group and nearly two-thirds (72%) in the control group did not report a history of falling. Future studies should utilize groups of older adults who have not only a history of falling, but also poorer performance on physical measures to determine the influence of cognitive training interventions on physical mobility and fall risk. Future studies should include a greater number of subjects with random assignment to the intervention or control group to provide more insight as to the influence of CCT interventions on measures of physical function and fall risk.

Due to the nature of a home based CCT intervention, we assumed that the subjects did not have assistance when completing the training, however this was not controlled for in the design and thus a limitation. In addition, self-report of activity level may not be a great way to measure changes in activity and future researchers should consider the use of accelerometers to detect any changes in physical activity throughout the study period. The study was a single, not double blinded study and fall risk was measured by proxy through the performance of subjects on measures of gait speed, the TUG, and the FTSTS. Lastly, the presence of cognitive impairments within subjects may have influenced the outcomes in both groups as mean MoCA scores were below the cutoff for MCI (30) which was not controlled for in the analyses.

Conclusion

Interventions that address cognitive deficits associated with impaired balance or falls have been provided in structured community settings with reported improvements on measures of balance and walking speed. Based on our results, home based cognitive training interventions which target specific domains reported to influence mobility and falls may offer another feasible treatment option to decrease the risk of falls associated with cognitive decline.

Conflicts of interest: Dr. Blackwood, Fogarty and Chase each declare no conflict of interest. Dr. Shubert reports receiving grants from Reflexion Health, outside the submitted work.

Ethical standard: The authors declare that the study procedures comply with the current ethical standards for investigation involving human participants in the United States.