Abstract
Background:
Governments are promoting the importance of maintaining cognitive health into older age to minimize risk of cognitive decline and dementia. Older people with amnestic mild cognitive impairment (aMCI) are particularly vulnerable to memory challenges in daily activities and are seeking ways to maintain independent living.
Objective:
To evaluate the effectiveness of memory groups for improving memory strategies and memory ability of older people, especially those with aMCI.
Methods:
113 healthy older adults (HOA) and 106 adults with aMCI were randomized to a six-week memory group or a waitlist control condition. Outcome was evaluated through knowledge and use of memory strategies, memory ability (self-report and neuropsychological tests), and wellbeing. Assessments included a six-month follow-up.
Results:
Using intention to treat analyses, there were intervention effects for HOA and aMCI groups in strategy knowledge (HOA: η2= 0.20; aMCI: η2= 0.06), strategy use (HOA: η2= 0.18; aMCI: η2= 0.08), and wellbeing (HOA: η2= 0.11; aMCI: η2= 0.05). There were also intervention effects in the HOA group, but not the aMCI group, in self-reported memory ability (η2= 0.06) and prospective memory tests (η2= 0.02). By six-month follow-up, gains were found on most HOA outcomes. In the aMCI group gains were found in strategy use, and by this stage, gains in prospective memory were also found.
Conclusion:
Memory groups can engage older people in techniques for maintaining cognitive health and improve memory performance, but more modest benefits are seen for older adults with aMCI.
Keywords
INTRODUCTION
In our aging societies, maintaining independent living is important for sustaining physical and mental health, thereby contributing to minimizing pressure on limited social services, and governments worldwide are promoting the importance of maintaining cognitive health into older age [1]. In this respect, many older people experience troublesome changes in memory performance [2, 3] and seek guidance regarding how best to manage memory challenges in everyday activities. This is especially the case for older people who meet criteria for the amnestic form of mild cognitive impairment (aMCI) in which, in addition to subjective memory complaints, memory impairment can be quantified through neuropsychological testing [4]. The importance of aMCI is that it carries increased risk of development of dementia due to Alzheimer’s disease (AD), occurring at a rate of 5–15% per year as distinct from 1–2% in healthy adults [5]; although a proportion of this group never progress to develop dementia [6, 7]. Nevertheless, it has been estimated that up to 21% of older people (aged between 70 and 90 years) will display features of aMCI [7]. Therefore, early cognitive interventions which moderate memory difficulties and assist in maintaining functional independence as long as possible would be useful to large numbers of older people.
Studies of memory training for older people through controlled practice on memory tasks (typically computer-based) have provided compelling results, forcing reconsideration of the potential for neuronal plasticity in older brains [8, 9]. However, the mental-exercise approach (or brain training) remains contentious, with much of the criticism relating to lack of generalization [10, 11], especially to everyday functioning [12]. This is important for aMCI, as the memory deficit is more problematic than for healthy older people and a critical issue is to reduce memory disability in daily activities rather than focusing exclusively on performance in arcane memory tests.
Following from research demonstrating the efficacy of cognitive strategies, or mnemonics, in experimental memory studies [13–15], an alternative approach is to train older adults in compensatory cognitive-behavioral memory strategies to be used in real-world situations. These approaches can be varied; for example, using mnemonic techniques such as associative strategies [16] to aid retrieval of a person’s name; or, using an external aid, such as a mobile phone reminder, to remember to pick up the grandchildren after supermarket shopping. Several meta-analyses have identified pre to post gains on memory tests in trained groups of healthy older adults [17–20], although all have cited caution in over-interpretation of findings due to a highly variable and methodologically limited literature base [21]. Nevertheless, an important study in memory strategy research is the ACTIVE trial [22], involving over 2,800 healthy older adults who were randomly allocated to one of three training groups (memory, reasoning, or speed of processing) or a no-contact control group. The memory training module involved practicing various mnemonic techniques, resulting in small gains on training-related memory tests, up to a five-year follow-up (d = 0.23) [23]. Although there was no effect on naturalistic cognitive tests, there was some evidence of transfer at 10-year follow-up in self-ratings of instrumental activities of daily living [24]. Whether similar effects can be obtained in aMCI is yet to reach a consensus, with some research finding positive results [25–28] but other research reporting no, or only limited, effects [21, 29–32].
Translating newly acquired knowledge about memory strategies to everyday situations requires behavior change, and it is recognized from health promotion studies that behavior change in older age is difficult to achieve [33]. Readiness for using a memory strategy at the appropriate time firstly requires a knowledge base about strategies [34]. Hutchens et al. [35] demonstrated that older people with aMCI have less capacity for accessing strategy knowledge than healthy older people, and that level of strategy knowledge was related to actual memory performance. This indicates that memory interventions, especially for people with aMCI, need to include components devoted to increasing knowledge about memory strategies.
An additional issue, and consistent with theoretical models of behavior change [36, 37], is that personally-relevant skills training and promotion of self-confidence in ability to apply these skills will be more likely to encourage older people to use strategies in daily activities [38, 39]. Group-based interventions, or memory groups, provide opportunity for peer-learning through sharing of experience in strategy-use and have been identified as facilitating better outcomes than individual-based training [18, 20]. Using this information, we undertook a preliminary study of a memory group for older adults with aMCI using a multi-strategy approach and incorporating education about health behaviors that facilitate effective memory performance [27]. We were able to demonstrate gains in knowledge and use of strategies, and also a small effect on everyday memory, demonstrable on prospective memory tests. Many daily activities rely on prospective memory, or remembering to perform intended actions in the future (such as passing on a message when you next meet a friend), and prospective memory performance can provide a useful approach to measurement of everyday memory competence [12]. However, the sample size was small (n = 20 per group) and we did not include a comparison healthy older adult group to evaluate the extent of gain from intervention by the aMCI group. This indicates the need for further study, especially as Reijnders et al. [29] concluded from their systematic review of interventions for older people that whether the effects of cognitive interventions generalize to improvement in everyday activities remains unresolved.
Therefore, our objective was to evaluate older people’s responses to a cognitive-behavioral intervention, i.e., memory group training, for improving everyday memory performance. We focused on people with aMCI as this group is particularly vulnerable to memory challenges in daily activities and are seeking ways to maintain independent living. Our aim was to improve memory ability through increased knowledge of effective memory strategies, and by training and modeling the use of several strategies that can be applied in daily activities. We expected that following intervention, and at six-month follow-up, older participants, including those with aMCI, would demonstrate greater knowledge and use of memory strategies, and improved memory ability in daily life. By contrast, there was limited expectation that performance would improve on standard neuropsychological tests of memory as many strategies are difficult to use on these researcher-controlled performance measures. We did, however, expect that participants would report improved wellbeing and self-confidence in everyday memory performance.
MATERIALS AND METHODS
Participants
Participants were recruited to form two cohorts: healthy older adults (HOA) and older adults diagnosed with aMCI. Initial inclusion criteria for all participants were: (a) age >60 years; (b) community-dwelling; and (c) English language sufficient for neuropsychological assessment. Additional specific criteria were used in the selection of participants for each cohort. Participants with aMCI were diagnosed and referred from memory clinics or experienced aged care specialists, following multidisciplinary diagnostic consensus review (i.e., neurological, psychiatric, radiological, neuropsychological, and functional assessment). Eligible participants were further screened to satisfy clinical criteria [4, 5]: (a) subjective memory complaint (participant or informant had sought professional investigation due to concern about memory performance); (b) objective memory impairment –1.5 SD or more below age norms for delayed recall on at least one of four screening measures of episodic memory: Hopkins Verbal Learning Test – Revised (HVLT-R) [40], Logical Memory subtest from the Wechsler Memory Scale Third Edition [41], Verbal Paired Associates subtest from the Wechsler Memory Scale Fourth Edition [42], Complex Figure of Rey [43]; (c) absence of significant impairment in basic activities of daily life (Clinical Dementia Rating) [44] score <1, and independent or only occasional assistance on the Alzheimer’s Disease Functional Assessment and Change Scale-personal sub-scale [45]; and (d) absence of dementia using NINCDS-ADRDA criteria [46] and a Mini-Mental Status Examination (MMSE) score of ≥24 [47]. HOA participants were volunteers who responded to advertising in community centers and groups to find out more about memory and memory strategies. In addition to the initial inclusion criteria listed above, HOA criteria were: (a) had not sought professional assessment due to concern about memory; (b) performance above - 1.5 SD of age norms on all of the four screening measures of delayed memory; (c) absence of impairment in activities of daily life (see above for assessment).
Exclusion criteria for both groups were: (a) presence of significant co-morbidities liable to impact on cognition; and (b) presence of significant visual and/or auditory impairment uncorrected by aids. All potential participants were reviewed by a case selection panel including a neurologist (ES), a psychiatrist (DA), a neuropsychologist (GJK), and the research coordinator (KEP), who formed a consensus decision on eligibility.
Study design and expectations
The study was a randomized controlled cross-over trial of a memory group intervention (see Fig. 1 which illustrates the study design). The trial was approved by ethics committees at La Trobe University and participating health services, and registered with the Australian and New Zealand Clinical Trials Registry, N012605000711617. At recruitment, participants (HOA and aMCI) were offered a training program in strategies for everyday memory challenges. If willing to participate, informed consent procedures, a neuropsychological assessment, and screening assessment for eligibility were undertaken (see Table 1 for performances on selected neuropsychological tests for screening). This was followed by baseline pre-test assessment on the outcome measures (listed below), and subsequent block randomization by diagnostic group to early- or late-intervention by a member of the research team (BO) who had no contact with participants, using a computer algorithm. Participants in the early-intervention condition were invited to the next memory group scheduled to start within two weeks and participants in the late-intervention (a wait-list control group at this stage) were informed that there was a delay until they could join the next memory group. Following the six-week memory group intervention there was a two-week delay interval before both groups underwent post-assessment (early post-test). This was followed by a cross-over phase in procedure such that the participants in the late-intervention groups were provided with the intervention while the participants in the early-intervention groups crossed to a maintenance phase. On completion of the cross-over phase and a further two-week delay interval, both groups underwent a further post-assessment (late post-test), followed at six months after training completion with a final assessment (six-month follow-up).
By comparing the randomized intervention groups at early post-test (for both HOA and aMCI participants), the effect of intervention on the outcome measures was evaluated, with the expectation that the early-intervention groups would perform better than the late-intervention groups (waitlist controls). At late post-test, when all participants had received intervention, we expected that all participants would have improved as compared to pre-test performances. By six-month follow-up, we anticipated some decline in performance, but expected that all participants would have made gains in outcome performances as compared topre-test.
Intervention
The La Trobe and Caulfield Hospital (LaTCH) Memory Group program consisted of six 2-hour weekly sessions conducted by neuropsychologists or an experienced occupational therapist assisted by co-facilitators, and held in health service facilities throughout Melbourne and regional Victoria, Australia. Group size ranged from six to twelve, and consisted of a mix of HOA and aMCI participants who could bring a family member or friend if they wished. The LaTCH program was developed from our earlier experience of memory groups [27] and built upon an evidence-based literature relating to memory, aging, and memory strategies that can be implemented by persons with aMCI as well as older adults [25, 48]. The LaTCH program was manualized, facilitators were trained in the program, and there were regular training meetings throughout the study to ensure protocol adherence. The interactive sessions provided information about memory and changes in memory as a result of health and lifestyle issues, aging, or neurological disorder. We used a problem-solving approach to illustrate common everyday memory problems, and provided practice in cognitive strategies (e.g., task organization, semantic association, visual imagery, retrieval practice, goal planning). As a frequently used approach to compensation of memory impairment is the use of external aids [49], the sessions explored a variety of aids (e.g., diaries/smartphones/manual timers) for various tasks and situations. We also addressed coping strategies to develop self-efficacy and emotional balance in responding to everyday memory challenges [39]. At the end of each session, participants received handouts covering the session material, and an assignment to facilitate application and practice of the skills in everydayactivities.
Participants assigned to the late-intervention groups (the waitlist controls) did not receive experimental intervention until the cross-over stage; but the aMCI participants did receive usual care from their referring clinicians in the memory clinics and all the participants could access community-based support services.
Outcome assessments
Assessors, who were graduate psychologists, were masked to the randomized condition and also to the diagnostic group of the participant (HOA, aMCI). Reflecting the multidimensional structure of the intervention, the primary outcomes covered several domains of interest:
Strategy knowledge: The Strategy Repertoire Test [31] measured the number and quality of memory strategies that were familiar to the participant by providing eight hypothetical everyday memory scenarios (e.g., remembering to pass on a message), and required participants to list as many strategies as possible that would be useful for each situation (internal consistency – Cronbach’s alpha 0.78 [35]). Responses were scored by effectiveness and specificity: 2 points for each specific strategy; 1 point for a non-specific strategy; and, 0 points for an ineffective or non-strategy (inter-rater reliability r = 0.94 [35]).
Strategy use: The Multifactorial Memory Questionnaire, the 19-item Strategy subscale (MMQ-Strategy) [50] was used to assess self-reported frequency of strategy use in everyday memory tasks (Cronbach’s alpha 0.83; test-retest r = 0.88). Internal strategies are reliant on cognitive manipulation to improve memory (e.g., forming a semantic association of a name with a face), whereas external strategies recruit external aids or environmental cues (e.g., smartphone reminders). To reflect this distinction, we used factor scores for internal and external strategies [51].
Memory ability: The Multifactorial Memory Questionnaire–Ability subscale (MMQ-Ability) [50] provided a 20-item self-rating of memory ability in everyday life (Cronbach’s alpha 0.93; test-retest r = 0.86). In addition, observed memory performance was assessed through two standard neuropsychological tests: the Cambridge Assessment of Prospective Memory (CAMPROMPT) [52] total score assessed prospective memory or remembering to perform intended activities; and the California Verbal Learning Test–Second Edition (CVLT-II) [53] long delay recall score assessed episodic memory and new learning. For both of these measures, alternate forms were used and counterbalanced across the assessment phases.
Wellbeing: The Multifactorial Memory Questionnaire–Contentment subscale (MMQ-Contentment) [50] was used to assess wellbeing. Eighteen items described emotions and perceptions linked to memory performance and self-belief in the controllability of memory (Cronbach’s alpha 0.95; test-retest r = 0.93).
Statistical method and sample size estimate
All statistical analyses were undertaken with IBM SPSS v. 21. Socio-demographic and clinical characteristics between diagnostic groups (HOA, aMCI) at baseline were compared by ANOVAs for interval-scaled variables, with eta-squared (η2) given as the effect size, and chi-square (χ2) analyses were used for categorical variables, with phi-squared (φ2) given as the effect size.
Missing data analysis with SPSS (EM technique) suggested that data could be assumed to be missing at random, Little MCAR test, χ2 (1830, n = 219) = 1923, p = 0.06. Therefore, adopting an intention-to-treat approach, missing data was imputed with SPSS in which demographic and screening data was used to predict missing pre-test scores and then, together, to predict missing scores at early post-test, late post-test, and follow-up. Statistical results arising from ANOVAs and ANCOVAs were pooled from the 20 imputed datasets; namely, means were pooled by averaging, F and η2 by median, SD by formula (see [54]), and p by re-calculation from F and df.
As significant differences were expected between diagnostic groups at pre-test on the outcome measures from four domains (strategy knowledge, strategy use, memory ability, and wellbeing), the effect of randomized memory intervention (early versus late) was analyzed separately for each diagnostic group (HOA, aMCI), and assessed by comparing randomized subgroups (early- versus late-intervention) at early post-test, using ANCOVAs, with pre-test scores as covariates. Gains from any intervention effects on outcomes measures in HOA and aMCI groups were assessed by ANOVAs at late post-test and at six-month follow-up, using gain (difference) scores from pre-test.
The differential impact of memory intervention on diagnostic groups was compared at late post-test using ANOVA on gain scores (from pre-test), and at six-month follow-up using ANOVA on gain scores (from pre-test).
The sample size required to detect a medium effect of intervention between randomized groups (early- versus late-intervention) was estimated to be 45 per randomized group for a power of 0.80 at α= 0.05 [55]. Assuming a 15% drop-out rate, we planned for a minimum of 104 HOA and 104 aMCI participants (52 per randomized group).
RESULTS
Characteristics of the groups (HOA and aMCI)
From a referral pool of 378 potential participants, 138 did not meet study criteria but only 21 potential participants declined or were unavailable during the study phase. Therefore, we recruited 219 (113 HOA, 106 aMCI). A flow diagram of the progress of the sample through the phases of the study and reasons for exclusions is provided in Fig. 1. By the final six-month follow-up, there were 174 participants (i.e., 80%) remaining from the initial sample. It should be noted that attrition by six-month follow-up (which included non-adherence to the intervention, i.e., attending <4 out of the 6 sessions) was considerably lower from the HOA group than from the aMCI group (11% versus 31%); however, within each diagnostic group, there was no differential attrition between the randomized intervention groups (HOA: 10% early versus 11% late; aMCI: 28% early versus 34% late). There was no significant difference in age (HOA p = 0.116; aMCI p = 0.225), gender (HOA p = 0.506; aMCI p = 0.289), education (HOA p = 0.079; aMCI p = 0.463), or screening memory performance (HVLT – R delayed recall, HOA p = 0.137; aMCI p = 0.814) between the initial sample and the participants retained at six-month follow-up, in either diagnostic group.
Table 1 shows that the randomized HOA and aMCI groups were similar in distribution of years of education and predicted IQ. However, the mean age of the aMCI group (76.08 y) was several years older than the HOA group (72.27 y) and the percentage of females was slightly lower in the aMCI group than the HOA group. As expected, the diagnostic groups were significantly different in general cognition (MMSE) and the screening measures of memory. At baseline (pre-test) the aMCI group provided significantly lower scores than the HOA group on all of the outcome measures apart from reporting equivalent use of external strategies (MMQ Strategy - external).
Effect of randomized intervention
The effect of randomized memory intervention (early versus late) was assessed separately for HOA and aMCI groups by ANCOVAs at early post-test, using the pre-test scores of the outcome measures as covariates (see Table 2).
Significant improvements in strategy knowledge (Strategy Repertoire) were demonstrated with a large intervention effect for the HOA group (η2 = 0.20) and a moderate effect for the aMCI group (η2 = 0.06). Significant improvements in self-reported strategy use were found through large effects in the HOA group for internal strategies (η2 = 0.18) and a moderate effect for external strategies (η2 = 0.07). There was also a moderate effect reported by the aMCI group for internal strategies (η2 = 0.08); however, not for external strategies (η2 = <0.01).
With respect to self-rated memory ability in everyday tasks (MMQ ability), improvement in response to intervention was reported in the HOA group with a moderate effect (η2 = 0.06), but not in the aMCI group (η2 < 0.01). Correspondingly, observed memory performance on the neuropsychological measure of everyday prospective memory (CAMPROMPT total score) showed a small but non-significant effect in the HOA group (η2 = 0.02), but not in the aMCI group (η2 < 0.01). On the neuropsychological measure of new learning (CVLT II delayed recall), neither the HOA group nor the aMCI group showed a significant improvement following intervention.
As can be seen in Table 2, estimate of wellbeing and self-efficacy as measured through the MMQ-Contentment scale, showed improvement following intervention for both HOA and aMCI groups with a moderate effect for the HOA group (η2 = 0.11), and a small effect for the aMCI group (η2 = 0.05).
Gains in outcome measures at late post-late and six-month follow-up
By late post-test (see Table 3) when all participants had received intervention, both diagnostic groups (HOA, aMCI) showed significant gains in strategy knowledge; but only the HOA group maintained an effect at six-month follow-up (M = 7.35, 95% CI [5.67, 9.04]). Both diagnostic groups achieved significant gains in internal strategies through memory intervention at post-late (HOA M = 4.92, 95% CI [3.92, 5.92]; aMCI M = 3.19, 95% CI [2.00, 4.38]) and also at six-month follow-up (HOA M = 3.86, 95% CI [2.91, 4.81]; aMCI M = 2.29, 95% CI [1.32, 3.26]). Similarly, there were significant gains for both diagnostic groups in external strategies at post-late (HOA M = 3.02, 95% CI [2.18, 3.86]; aMCI M = 1.38, 95% CI [0.38, 2.38]) and maintained at six-month follow-up (HOA M = 2.27, 95% CI [1.34, 3.20]; aMCI M = 0.95, 95% CI [0.01, 1.88]).
There were gains in self-reported memory performance (Memory Ability) for the HOA group at late post-test (HOA M = 2.14, 95% CI [0.71, 3.57]) but not for the aMCI group (M = 1.66, 95% CI [–0.81, 3.39]). However, neither group maintained gains by six-month follow-up (HOA M = 1.05, 95% CI [–0.37, 2.47]; aMCI M = 0.79, 95% CI [–0.94, 2.53]). In relation to neuropsychological test performances, both diagnostic groups showed significant gains on the CAMPROMPT at late post-test (HOA M = 5.38, 95% CI [4.14, 6.61]; aMCI M = 4.77, 95% CI [3.35, 6.19]) and maintained at six-month follow-up (HOA M = 5.05, 95% CI [3.78, 6.31]; aMCI M = 4.09, 95% CI [2.56, 5.62]). On the CVLT long delay, the HOA group showed significant gains at late post-test and at six-month follow-up (late post-test - HOA M = 0.76, 95% CI [0.22, 1.28] and follow-up HOA M = 0.59, 95% CI [0.14, 1.04]); whereas aMCI only showed significant gain at late post-test (M = 0.64, 95% CI [0.05, 1.22]) but not at six-month follow-up (M = 0.02, 95% CI [–0.53, 0.57]).
Finally, in relation to wellbeing and self-efficacy, both diagnostic groups (HOA, aMCI) showed significant gains at late post-test on the MMQ-Contentment scale(HOA M = 4.24, 95% CI [2.67, 5.81]; aMCI M = 3.33, 95% CI [1.46, 5.20]), but only the HOA group maintained an effect at six-month follow-up (M = 2.43, 95% CI [0.66, 4.19]).
Differences in extent of gains on outcome measures between diagnostic groups
Pre-test differences between diagnostic groups (HOA, aMCI) were found, as expected, on most outcome measures, using ANCOVAs at pre-test with age as the covariate (see Table 1). Of greater interest was whether age-adjusted gains were different between diagnostic groups at late post-test and six-month follow-up (see HOA versus aMCI column, Table 3).
A differential gain between HOA and aMCI groups in Strategy Repertoire was significant at late post-test (η2 = 0.07) and at six-month follow-up (η2 = 0.09), favoring the HOA group. For internal strategies, gains were significantly different between HOA and aMCI groups at late post-test (η2 = 0.02) and six-month follow-up (η2 = 0.03), favoring the HOA group. In addition, the HOA group gained more in external strategies use at post-test than the aMCI group (η2 = 0.03), and at six-month follow-up (η2 = 0.02).
There were no significant differences in gains in self-reported Memory Ability between diagnostic groups at post-test and follow-up. There were also no diagnostic group differences in gains at post-test and follow-up in relation to neuropsychological test performances (CAMPROMPT and CVLT-II).
Finally, in relation to wellbeing estimates, there were no diagnostic group differences in gains at post-test and follow-up on the MMQ-Contentment.
DISCUSSION
We explored two main questions: (i) does a six-week memory group improve knowledge and use of strategies in older people, especially in those with aMCI? and (ii) do memory ability and wellbeing improve following a memory group? We used a cognitive-behavioral intervention (the LaTCH Memory Group program), which allowed the training and modeling of compensatory strategies that can be readily applied in daily activities. We found that in older age participants, including those with aMCI, knowledge about memory strategies improved following intervention, and that there was a behavioral change with greater use of strategies in daily activities, while wellbeing and self-confidence in memory also improved. These effects were translated into improved memory ability (in daily activities and prospective memory test performance) in the HOA group, but not for the aMCI group. By six-month follow-up, gains persisted on most outcomes for the HOA group; however, for the aMCI group gains were only maintained in strategy use in daily activities, although it was also noted that by this stage there were gains in prospective memory test performance. The trial findings are encouraging as they indicate benefits from memory groups for older age participants, albeit with generally larger and more varied effects for the HOA group than the aMCI group.
In relation to the first question, strategy knowledge is a necessary prelude to using an appropriate strategy in real life activities [35]; by increasing an understanding of the range of strategies that can be considered for specific tasks, an older person will be more prepared for unexpected situations when a ‘go to’ strategy fails [17]. For example, following training one of the participants reported being able to successfully use goal planning (implementation intention) [56] to remember an appointment later in the day after he mislaid his mobile phone (and reminder function). Nevertheless, despite immediate gains following intervention, and in contrast to the HOA group, the aMCI group was unable to demonstrate long-term gains in strategy knowledge. In earlier small sample studies with aMCI participants [27, 31], it was reported that strategy knowledge gains could be maintained up to three or four months post-training, but there was no further follow-up assessment. In contrast, our six-month follow-up provides a more challenging test of sustainable change. It was also apparent that even before the intervention our aMCI group was less able than the HOA group in identifying appropriate strategies for everyday memory scenarios, suggesting deterioration in pre-existing semantic knowledge [57]. The loss of semantic memory in aMCI adds a further constraint in learning and maintaining new skills, and highlights the potential benefit of extending the number of sessions in the primary memory group or adding booster sessions for consolidating newly acquired information in people confronted by significant memory difficulties [58]. However, our decision to constrain the number of initial sessions in the memory group was determined by feedback about the feasibility of group length from participants in our earlier pilot study [27]. Furthermore, in their systematic review of intervention studies, Reijnders et al. [29] found no dose-relationship between hours of intervention and effectiveness. Alternatively, some test specific effects were found in the ACTIVE study when booster training was provided [23] suggesting that this approach may provide the greatest advantage for consolidating skill development in use of strategies.
Of course, increasing knowledge alone will not guarantee changing behavior; therefore, providing skills in the actual techniques and application of memory strategies is necessary to persuade older adults that they can manage their everyday memory more effectively [39]. Furthermore, Dixon and de Frias [59] reported that in a six-year observational study of older adults, participants who were found to have a mild memory deficit were increasingly less likely to use memory strategies than older adults with normal memory function. In this context, it was notable that our participants with aMCI, as well as the HOA group, reported that their strategy use increased following intervention, and long-term gains were found in both internal and external strategies. The intervention, following the principles of promoting behavior change [37], focused on modeling of strategies to real-world scenarios and provided opportunity for mastery of skill through practice and weekly assignments. Its multi-strategy approach endorsed the position that different people can use different strategies to assist performance of everyday tasks [60]. Therefore, participants were encouraged to develop skills in multiple strategies rather than focus on one approach; thereby, providing more secure scaffolding for real world demands [24].
Indeed, the positive shift in reported increased strategy use was reflected in evidence of improvement in objective memory performance. There were long-term gains at six-month follow-up on prospective memory test performance for both groups, and the HOA group showed a small effect in prospective memory immediately after intervention. The relevance of prospective memory, or remembering to perform an intended action at some point in the future, is that memory complaints of older people are frequently about prospective memory errors [12] and prospective memory has been associated with capacity to maintain independent living [61]. The prospective memory test used in the study (CAMPROMPT) allowed the use of memory strategies to provide a more naturalistic approach to memory testing, which may account for why early intervention effects were noted on this test but not on the word-list memory test (CVLT-II). Furthermore, when questioned about how they approached memory testing, many participants reported that they would not use strategies on these tests as that would be “cheating”. This illustrates the dilemma of a disconnection between performance in real life activities and performance on clinical tests. The objective of the memory group was to manage memory difficulties in the real world rather than to effect a shift in performance on traditional clinical tests where there is often little opportunity to put into practice strategies that are relevant for functional problems. A counter-issue is that caution is also needed in interpreting the apparent long-term gains on neuropsychological tests as although alternate versions of tests were used and counterbalanced across assessment phases, the contribution of practice effects from familiarity with test protocols must be considered. While neither group reported improved memory ability in daily activities by six-month follow-up, the HOA group did report benefit immediately after intervention. Troyer et al. [31] proposed a staged model of intervention effects, whereby an increase in knowledge of what to do, is followed by a behavioral change in increased use of strategies, and finally by translation of behavioral change into task performance change. In our study, there is support for the translation of behavioral change (knowledge and use of strategies) into task performance change for the HOA group immediately following intervention and qualified support for long-term task gains. However, although some task gains were observed by six-month follow-up, the aMCI group may need more extensive intervention to be clipped onto a standard memory group program to reinforce translation of strategy use into functional behavior. This may take the form of combining individualized goal-oriented sessions [62] with the group sessions in order to reinforce the practical application of strategy use for personalized goal attainment in everyday activities.
Finally, a component of the positive effect of the memory groups is captured by the gains in wellbeing, including increased self-efficacy or self-confidence in managing memory performance (also see [39]) and indexed through increased contentment with memory performance (MMQ Contentment). Although the aMCI group could not maintain their early gains by six-month follow-up, the HOA group was able to show sustainable wellbeing effects from participation in the memory group. Following intervention, most participants reported being reassured that they had been able to meet with others who experienced similar memory challenges in daily activities, and many participants commented that as a result of attending the group, they regained self-confidence and were able to re-engage with former activities. For example, one participant described returning to volunteer driving for meals on wheels after he had mastered a technique for remembering the routes to be taken for dropping off the meals. It is the re-engagement with previous life activities as much as engagement with new activities that represents an important contribution in the pathway to better mental health, as social re-engagement is potentially more sustainable.
In relation to methodology, the major strengths of our study are that the sample includes a large, well-defined group of older people with aMCI, and that the multi-component intervention was developed on evidence-based techniques for improving memory performance [17] and combined with current models of effecting behavior change [37]. In addition, the study methodology carefully adhered to CONSORT recommendations and included a six-month follow-up to evaluate gains in outcome measures after the intervention had ceased. However, it should be noted that in order to address our research questions, we elected to use a randomized waitlist control group rather than an active control. We used this approach as recruitment of the aMCI group was through memory clinics, where there was a reluctance to commit to an extended delay (including the follow-up period) in offering the memory groups for patients struggling with memory difficulties. As a result the beneficial effects of the intervention may have been moderated by the positive expectances of the waitlist group, who were anticipating involvement in the memory group. Nevertheless, we were still able to detect small to medium effects following the randomized intervention in most of the outcome measures. A further constraint of the waitlist control group is that we are unable to specify which component of the intervention was more potent in effecting behavior change, including the impact of therapeutic attention alone. However, by selecting discrete assessments of memory strategy knowledge, strategy use, and memory ability as well as the more general outcome measure of wellbeing, we minimized the risk of over-interpreting the potential for simply generalized effects from therapeutic attention.
In conclusion, our study findings provide support for the effectiveness of memory groups as a feasible intervention in an older age population to reduce and limit cognitive difficulties in daily life. The value of memory groups is that they provide a low cost method for focusing on the engagement of older adults into techniques for maintaining cognitive health. However, our study findings suggest that a standard six-week model of intervention needs adaptation to maximize the benefits for people with aMCI. Although there was no clear evidence of translation of intervention effects into improved memory ability for the aMCI group, there were intervention benefits (increased knowledge and use of strategies, improved wellbeing and self-confidence in memory) that provide important contributions to maintaining independence and social engagement in situations of memory difficulty. Future research is needed to evaluate how these early training effects can be sustained over extended periods of time by extending the sessions of the primary memory group; or, more specifically, to evaluate whether the beneficial effects of the intervention for people with aMCI can be enhanced by incorporating periodic goal-oriented booster sessions to consolidate new skills acquisition and also to improve self-confidence in applying skills in daily activities. From a cost perspective, the memory group intervention has relatively low resource requirements and can be readily incorporated into existing health and community services. Such sustainable practice is particularly important within the context of meeting the concerns of our aging communities.
Footnotes
ACKNOWLEDGMENTS
The authors wish to acknowledge the study participants who participated in the memory groups. We also acknowledge the support of the Cognitive, Dementia and Memory Services at Austin Health, Barwon Health, Bendigo Health, Bundoora Extended Care Centre, Caulfield Hospital, Melbourne Health, St. George’s Hospital and Wantirna Hospital, as well as Associate Professor Michael Woodward and Dr. Alastair Mander for their referral of aMCI participants to the study. We would also like to thank Dr. Sarah Price, Dr. Nadia Petruccelli, Sam Parsons and Fenny Muliadi for their contribution to the organization of the study, and the research assistants who aided in data collection. This work was supported by a NHMRC, Australia grant (487318) to GJK, DA, ES, BO, MMS, LC, EM, and ER. KEP is supported by a NHMRC Clinical Research Training Fellowship (602543).
