Abstract
Background:
Current pharmacological and behavioral treatment options for mild cognitive impairment (MCI) are limited, motivating a search for alternative therapies that might slow the progression of cognitive decline.
Objective:
We investigated the effectiveness of a cognition-focused mindfulness-based intervention.
Methods:
An open-label, three arm randomized controlled trial was conducted at a public tertiary medical center. Older persons (ages 45–75; N = 76) diagnosed with MCI were recruited and randomized into either mindfulness-based training (MBT), cognitive rehabilitation therapy (CRT), or treatment as usual (TAU). Participants in the intervention arms received 8 weekly 2-h sessions delivered in a group setting and engaged in home practice. Primary outcomes measures included changes in index scores for attention, immediate memory, and delayed memory as measured using the Repeatable Battery for the Assessment of Neuropsychological Status (RBANS). Depression was a secondary outcome.
Results:
Using intent-to-treat analysis, we found that participants receiving MBT showed significant improvements in global cognition (d = 0.26; [95%CI 0.03–0.56]) and delayed memory (d = 0.36; [95%CI 0.17–0.57]), with significantly greater improvements in delayed memory than CRT (ηp2 = 0.10). However, there was no benefit of MBT over TAU. No change in depression was observed in the MBT group. Reductions in depression were associated with improvements in cognitive functioning in the MBT group only.
Conclusion:
Our results suggest that a cognition-focused MBT did not improve cognitive functioning in MCI patients substantially more than spontaneous reversion rates, possibly as mood symptoms were not significantly alleviated in this group.
INTRODUCTION
The burden from common age-related brain disorders (e.g., Alzheimer’s disease) has doubled from 1990 to 2016, and is expected to rise with increased longevity [1]. Current treatment options are limited: drugs such as cholinesterase inhibitors have only a slight and short-term effect in mitigating cognitive decline [2, 3], as do behavioral treatments such as cognitive rehabilitation therapy [4]. Against this backdrop, calls have been made to focus on modifiable risk factors for dementia, with an estimate that this might prevent or delay up to 40%of cases [5], and towards this end, meditation and the practice of mindfulness have been proposed as useful tools to mitigate cognitive decline. Aside from evidence that mindfulness training can improve cognition among healthy individuals [6], practicing mindfulness also has known associations with improvements in mood [7], sleep [8], and well-being [9], all of which are protective factors against developing dementia.
Mindfulness, which is cultivated through several formal meditative practices, is a mental state achieved by focusing one’s attention and awareness on the present moment, while calmly acknowledging and accepting one’s feelings, thoughts, and bodily sensations [10]. Mindfulness-based interventions (MBIs) consist of a series of exercises designed to improve these skills, can be implemented in a group setting at low cost, and may have a low risk of negative consequences (however, see [11]). While standard programs emphasize the effects of mindfulness on stress and negative emotions, training protocols can also be modified to highlight the relationship between the mindfulness exercises and the individual’s cognition, specifically attention and memory.
There is currently some evidence that MBIs can slow or reverse cognitive decline in persons with neurodegenerative disorders [12–15], with a recent systematic review suggesting that effect sizes are in the small-to-medium range [16]. In particular, it has been suggested that early intervention, such as at the stage of mild cognitive impairment (MCI), is more effective than when patients progress to more severe disease. MCI is a transitional state between normal aging and very early dementia marked by progressive cognitive decline [17]. While 50%of those with MCI will develop dementia within 5 years, approximately 14–40%of these individuals will return to normal cognitive functioning over time [18]. Although this reversibility indicates that this group is a prime target for early intervention, it also suggests the need for studies that use both active and passive control conditions in order to account for diagnostic instability.
Based on this logic, our aim was to test the potential of a shorter 8-week MBI for clinical use in patients with cognitive impairment using both an active and passive control arm. To achieve this, we used an MBI program modified to focus on how the skills of mindfulness might specifically benefit cognition, and examined the effects of the intervention on individual cognitive domains. Patients with MCI were randomized to receive either an MBI, cognitive rehabilitation therapy or standard care. Our primary outcome variables were scale scores of attention, immediate memory, and delayed memory on a standardized neuropsychological battery, the Repeatable Battery for the Assessment of Neurological Status (RBANS). We hypothesized that patients in the MBI group would experience significantly improved scores on the RBANS following the intervention in comparison with the active and passive control groups, with corresponding increases in mood (i.e., lower depression scores).
MATERIALS AND METHODS
This study was an open-label, assessor-blinded, parallel, randomized clinical trial. Ethical approval was obtained from both the SingHealth Centralised Institutional Review Board (cIRB 2015/3049) as well as the National University of Singapore Institutional Review Board (NUS-IRB: H-19-048) and was conducted in accordance with the 1964 Helsinki declaration and its later amendments. The study was pre-registered at ClinicalTrials.gov, identifier: NCT04000984. All participants provided written informed consent and were reimbursed up to SG$180 for their time, based on their attendance.
Participants
Study participants were recruited through five main channels: 1) through the outpatient memory clinic at a tertiary hospital; 2) from members of the community who attended public events on healthy aging and cognition, 3) using posters within the tertiary hospital, 4) through online forum pages targeted at older people, and 5) through email groups related to dementia. Members of the community who expressed interest in participating in the study were assessed with a diagnostic neuropsychological test battery (Supplementary Material) administered by a clinical psychologist.
Inclusion criteria
To be eligible for this study, participants needed to be in their early stages of mild memory loss or an impairment to other cognitive abilities such as language or visuospatial perception, while maintaining their ability to perform most of their daily living activities independently. They were required to be 1) between the ages of 45 to 75, 2) fulfill the Diagnostic and Statistical Manual of Mental Disorders version five (DSM-V) diagnostic criteria for Minor Neurocognitive Disorder or Mild Cognitive Impairment: that is to have at least one impairment across the domains tested for neurocognitive functioning, and 3) have a Clinical Dementia Rating Score (CDR) = 0.5 and 4) have a Mini-Mental State Examination (MMSE) score of >20.
Exclusion criteria
Participants were excluded from the study if they had 1) any presence of a major neurological conditions such as epilepsy, stroke, Parkinson’s disease and/or traumatic brain injury, 2) presence of any major psychiatric conditions such as major depression or schizophrenia, 3) contraindications for fMRI scanning (e.g., pacemakers, metallic implants, claustrophobia), 4) moderate to severe levels of depression measured by the Geriatric Depression Scale (GDS) [19], or 5) were unable to give independent consent.
Power analysis
The current study was powered to detect differences at post-intervention testing between MBT and each control condition, with a large effect size (d = 0.7) at β= 0.80. Based on these parameters, we decided that a minimum sample size of 25 participants per group was required.
Data acquisition
Eligible participants were scheduled for two visits to collect pre- and post-intervention data, which included neuropsychological assessments and questionnaires, and a 1-h MRI scan (not reported in this paper). All pre-intervention data collection occurred no earlier than 3 months from the first intervention session and all post-intervention data collection session occurred no later than 3 months after the last intervention session.
Randomization and blinding
Participants were randomly assigned to each treatment arm 1 week prior to the start of the intervention, which was conducted when a sufficient number of participants had been recruited to form 3 treatment groups. The initial 24 participants were randomized to groups in a 1:1:1 ratio. Due to difficulties in subject recruitment, a pragmatic decision was made to change this ratio to 2:2:1 (with fewer individuals randomized to treatment as usual [TAU]) for the remainder of the trial. Pre-intervention testing was double blinded while the post-intervention testing session was conducted by a blinded team member.
Interventions
Mindfulness-based training program
The mindfulness-based training (MBT) program consisted of eight weekly sessions (Supplementary Table 1); each session was approximately an hour and a half in duration. Weekly group sessions included an introduction and practice of mindfulness exercises, a discussion of participants’ experiences and how mindfulness relates to their cognitive abilities, and an assignment of the activities to do on their own between sessions. To help incorporate mindfulness into their daily lives, participants were asked to spend a minimum of 15 min a day practicing these exercises and to record them in a journal that was given to them at the start of the program. The diaries were implemented as a guide and resource, while participants were encouraged to complete it weekly it was not mandatory. Diaries also recorded weekly practice. Participants were also given audio tracks of the various exercises taught during the session to aid their practice at home; these guided practices were no longer than 30 min. Unlike programs such as Mindfulness-Based Stress Reduction [10], the intervention in this study did not include a retreat.
Cognitive rehabilitation training program
The cognitive rehabilitation training (CRT) program also consisted of eight weekly sessions, each lasting up to an hour and a half (Supplementary Table 1). Each session had the following components: 1) identifying and working on at least one personal rehabilitation goal related to everyday life that is associated with cognitive difficulties; 2) reviewing and building on the use of practical memory strategies, and/or introducing and teaching the use of a new strategy or memory aid; 3) introducing techniques for learning new information and associations, identifying the preferred strategy, and encouraging the application of this strategy in daily life; 4) providing practice in maintaining attention and concentration; and 5) exploring current ways of coping with stress and anxiety as well as providing relaxation techniques to aid with coping [20].
To facilitate their understanding and practice of the strategies taught and discussed during the session, participants were provided with a training book. The training book also contained logs to help them record, monitor, and evaluate their progress.
Treatment as usual (TAU)
The third group received treatment as usual, which was the standard care offered by the memory clinic. Participants received no intervention and were prospectively monitored on a yearly basis by their attending neurologist.
Outcome measures
Neuropsychological tests
The Singapore validated versions of the MMSE [21] and the Montreal Cognitive Assessment (MoCA) [22] were both used as global cognitive screening tests.
The RBANS [23] was used to assess multiple areas of cognitive functions and includes different test forms for use before and after the interventions. Test forms A and B were alternately used as either the pre- or post-intervention measure to reduce test order effect on the results. The pre-registered primary outcome measures for this study were the subscale scores for attention, immediate memory, and delayed memory. For completeness, we also analyzed the language and visuospatial subscales of the RBANS in secondary analysis. Details of the subtests comprising each subscale score are reported in the Supplementary Material.
Mood assessment
We measured mood using the 20-item Modified Geriatric Depression Scale (GDS-M) [24]. GDS-M has good internal consistency (α= 0.85) [25].
Trait mindfulness
Trait mindfulness was assessed using the Mindful Attention and Awareness Scale (MAAS) [26] which is a 15-item questionnaire that measures an individual’s level of awareness and attention to present events and experiences. The MAAS has shown good internal consistency (α∼= 0.90) across several studies [27, 28].
Statistical analysis
Participants were included in the final analysis if they contributed data at pre-intervention and were randomized to an intervention arm. Missing data for participants who dropped out after randomization were filled using intent-to-treat analysis, using the pooled average of five imputed values derived from an iterative Markov Chain Monte Carlo method with predictive mean matching. Demographic and pre-intervention clinical measures were used as predictor variables.
As the MBT was our treatment of interest, data were analyzed using 2×2 repeated-measures ANOVA with time (pre/post-intervention) as a within-subjects factor and group as a between-subjects factor. Two sets of analyses were conducted: MBT against active control (CRT), and the second compared MBT against TAU. We considered comparisons to be statistically significant at Cronbach α= 0.05.
Finally, to explore the correlations between cognition and mood, we conducted Pearson’s correlations on the change in GDS scores and the change in MMSE scores within each intervention arm.
RESULTS
Participant flow
Figure 1 depicts the CONSORT flowchart for this trial, and Table 1 reports baseline demographic and clinical characteristics. 605 participants were assessed for initial eligibility, 191 were consented to be screened, 81 were enrolled into the study and 5 participants withdrew from the study prior to randomization. Data from 76 participants were included in the analysis. Participants ranged between 53 and 75 years old (M = 67; SD = 5.3), spoke either English (N = 45) or Mandarin (N = 31) and were predominantly female (56.3%). All but 3 participants had at least primary education (>6 years).
Consort flow diagram of the study. MBT, mindfulness-based treatment; CRT, cognitive rehabilitation therapy; TAU, treatment as usual; ITT, intent-to-treat.
Demographic variables of intent-to-treat sample
Thirty-two participants were randomized into the MBT group (male = 16; meanage(SD) = 67.6 (5.3); mean attendance = 6.3 sessions), 27 participants were randomized into the CRT group (male = 10; meanage(SD) = 67.1 (3.4); mean attendance = 6.1 sessions), and 17 participants were randomized to receive TAU (male = 8; meanage(SD) = 66.3 (6.6)). One participant who completed the CRT intervention was lost to follow-up and 2 participants in the TAU group were lost to follow-up.
Data were collected between April 2016 and September 2019. Recruitment stopped when we reached our predetermined sample size.
Missing data
Three participants did not contribute GDS data at baseline. Data were also not available for 2 participants for the RBANS-visuospatial scale score as they lacked the manual dexterity to complete the tasks.
Baseline characteristics
Using one-way ANOVA on the intent-to-treat sample (ITT), we found no significant differences among the three groups in any demographic or clinical variables (p > 0.05).
We compared dropouts and completers on demographic characteristics (age, gender) and scores on primary clinical outcomes (MMSE, MoCA, RBANS subscale scores) using independent-samples t-tests to assess for risk of bias. Dropouts from the study tended to be younger than completers (64.3 versus 67.7; t = 2.15; p = 0.04). No other significant differences were found in these tests (p > 0.05).
Means and standard deviations of all measures pre- and post-intervention are presented in Table 2.
Means and standard deviations of outcome variables pre- and post-intervention
MBT, mindfulness-based treatment; CRT, cognitive rehabilitation therapy; TAU, treatment as usual; MMSE, Mini-Mental State Examination; MoCA, Montreal Cognitive Assessment; RBANS, Repeatable Battery for the Assessment of Neuropsychological Status; GDS, Geriatric Depression Scale; MAAS, Mindful Attention and Awareness Scale.
Neuropsychological outcomes
MBT versus CRT
Among the primary outcome variables (Table 3), repeated-measures ANOVA of the two active treatment groups showed that participants improved from pre-to post-intervention on the MMSE (p = 0.03), MoCA (p = 0.009), and the RBANS delayed memory subscale (p = 0.009). We further found a significant time by group interaction on the RBANS delayed memory subscale (p = 0.01), driven by a larger improvement in the MBT group. Post-hoc analyses demonstrated that the MBT group alone showed significant improvements on all of these outcomes except the MOCA (MMSE: t31 = 2.09, d = 0.26, [95%CI, 0.03–0.56], p = 0.044; MOCA: t31 = 1.94, d = 0.25, [95%CI, –0.06 to 0.49], p = 0.06; RBANS-delayed memory: t31 = 3.96, d = 0.36, [95%CI, 0.17–0.57], p < 0.001).
Repeated measures ANOVA F values for ITT analysis (MBT versus CRT)
ITT, intent-to-treat; MBT, mindfulness-based treatment; CRT, cognitive rehabilitation therapy; MMSE, Mini-Mental State Examination; MoCA, Montreal Cognitive Assessment; RBANS, Repeatable Battery for the Assessment of Neuropsychological Status; GDS, Geriatric Depression Scale, MAAS, Mindful Attention and Awareness Scale.
In the two secondary outcome variables, we found a significant improvement over time in the RBANS-visuospatial subscale (p = 0.03), with no time by group interaction. Post-hoc analysis showed that MBT participants improved significantly on this subscale (RBANS-visuospatial: t30 = 2.20; d = 0.33; [95%CI, 0.01–0.65]; p = 0.04).
MBT versus TAU
Among the primary outcome variables (Table 4), repeated-measures ANOVA of the MBT and passive control group showed that participants improved from pre- to post-intervention on the MoCA (p = 0.01), the RBANS attention subscale (p = 0.001), and the RBANS delayed memory subscale (p = 0.001). We further found a significant time by group interaction on the RBANS attention subscale (p = 0.04), driven by a larger improvement in the TAU group.
Repeated measures ANOVA F values for ITT analysis (MBT versus TAU)
ITT, intent-to-treat; MBT, mindfulness-based treatment; TAU, treatment as usual; MMSE, Mini-Mental State Examination; MoCA, Montreal Cognitive Assessment; RBANS, Repeatable Battery for the Assessment of Neuropsychological Status; GDS, Geriatric Depression Scale, MAAS, Mindful Attention and Awareness Scale.
In the two secondary outcome variables, we found a significant improvement over time in the RBANS-visuospatial subscale (p = 0.007), with no time by group interaction.
Self-report outcomes
No changes in trait mindfulness were observed in comparisons of MBT with either of the two control groups (Tables 3-4- and Tables 3-4).
With regard to depression, we observed significant decreases in GDS scores over time in the comparison of MBT and CRT (p = 0.03), and a significant time by group interaction driven by a greater decrease in the CRT group (p = 0.03) (Table 3).
Associations between mood and cognition
To explore the relationship between mood and cognition, we correlated change in GDS scores from pre- to post-intervention with change in MMSE scores (Supplementary Figure 1). In this analysis, we found a significant relationship in the MBT group (r = –0.38; p = 0.04) but not CRT (r = –0.14; p = 0.49) or TAU (r = –0.12, p = 0.65). However, correlation coefficients did not differ significantly between MBT and CRT (z = –0.91; p = 0.18) and between MBT and TAU (z = –0.85; p = 0.20).
DISCUSSION
In this randomized controlled trial, we studied the effects of a mindfulness-based intervention against both active and passive control conditions on a comprehensive, repeatable neuropsychological battery. Compared with CRT, MBT participants showed greater improvements in delayed memory, significant but non-superior improvements in global cognitive functioning (measured by the MMSE) and visuospatial processing. Surprisingly, these outcomes were not superior to TAU, as participants in this group showed substantial amounts of spontaneous improvement in cognitive functioning. Overall, these results are equivocal, and suggest that the lack of diagnostic stability in MCI may have played a large role in the cognitive improvements observed in this trial. Notwithstanding this possibility, our data also suggest that MBT is a marginally more effective intervention than CRT for patients comparing these two options.
Several previous RCTs have been conducted comparing MBIs with an active control condition, and with neuropsychological testing among their outcomes. Smaller pilot studies have generally not found positive effects [29, 30]. In a two-year trial, Quintana-Hernández et al. [13] randomized N = 120 Alzheimer’s disease patients to receive mindfulness training (Mindfulness-Based Alzheimer’s Stimulation), cognitive stimulation therapy, progressive muscle relaxation, or treatment with Donepezil. Mindfulness-Based Alzheimer’s Stimulation and cognitive stimulation therapy were superior to progressive muscle relaxation in maintaining cognitive functioning as early as six months and were superior to the other two control conditions at the study endpoint. Yu et al. [15] randomized MCI patients to receive either an MBI or a health education program. No significant changes in cognition were observed at three-month follow-up. At nine-month follow-up, a time by group interaction was observed on a forward digit span task, and on interference effects on the Color Trails Test.
Overall, the literature of using MBIs to combat cognitive decline in the early stages of dementia is highly mixed, comprising small effects that are heterogeneous across cognitive domains [16]. Consistent with this pattern, we did not observe a clear benefit of MBI over TAU, even though the treatment was effective in improving global cognition (MMSE scores) and superior to CRT in improving delayed memory.
We posit two reasons for the modest effect of MBI in our study compared with previous work. First, our protocol was focused more sharply on the cognitive aspects of mindfulness, which paradoxically may have detracted from the effectiveness of the program by de-emphasizing its other important components. For example, mindfulness training is an effective means in reducing depression [31], which is a risk factor for the development of MCI [32]. In fact, a meta-analysis of MBIs for patients with dementia showed that depression (but not anxiety or stress) is effectively targeted by this treatment [33]. Contrary to this finding, we did not observe changes in depression in the MBI group in our study, while significant decreases were in fact seen in both CRT and TAU. However, we did observe that decreases in depression were correlated with improvements in global cognition in the mindfulness group only (albeit this relationship did not differ significantly from the slopes in CRT and TAU). This pattern of results suggests that the MBI administered in this protocol was not as effective as previous interventions in improving mood, possibly because of its strong emphasis on the relationship between mindfulness and cognition, and that paradoxically, programs that are more effective in targeting negative mood may also have greater effects on cognition through the mediating effect of lowered depression.
Second, participants in the TAU group showed a strong effect of spontaneous reversion in our sample. Prior studies have shown that a substantial percentage of patients revert to normal cognitive functioning with time without any pharmacological or non-pharmacological intervention. This is true even in the relatively short-term; for example, Diniz et al. [34] reported that 23%of patients diagnosed with MCI reverted to normal cognitive functioning over a 12-month period, and Han et al. [35] reported a reversion rate of 28.5%. This effect may be enhanced through participation in research studies (such as ours) that promoted engagement in activities that are cognitively stimulating, even among those assigned to the “treatment-as-usual” arm who are required to contribute data and interact with research staff. In particular, the increased levels of social interaction experienced by the MBT and CRT groups may have accounted for symptom improvement.
Generally speaking, engaging in mindfulness-based interventions, particularly in the context of a research study, requires a relatively high level of functioning, and may be particularly challenging for those with low educational attainment. For instance, the daily practice and journaling requirements of our MBI may have been relatively burdensome for cognitively impaired individuals, which could have contributed to the high dropout rate in the current study. Simplifying course content and minimizing participant burden should be important considerations for future intervention trials of this nature.
The strengths of the trial are its inclusion of both an active and passive control group, and its relatively low risk of bias. The trial also has a number of limitations. First, we were unable to conduct longer-term follow-up, and thus cannot ascertain whether the benefits to cognition were retained over time. It is conceivable that the benefits of MBI over CRT and TAU may only be seen on the order of months following the interventions. Second, the trial had relatively high dropout rates, particularly before participants were randomized to condition, increasing the likelihood of attrition bias. Third, our intervention was relatively short compared with other studies [12, 14], and the negative results may be due to this low treatment dose. Finally, as the trial was open-label, participants were not blind to condition, or to the interventions being administered in the other study arms, which may have led to expectancy effects.
To conclude, in the context of prior positive results, the current study suggests that a more careful assessment is needed to identify the circumstances under which mindfulness training benefits persons with MCI. These may include population or intervention characteristics, or disease severity/subtype. Such trials may permit more confident recommendations for if and when MBIs should be recommended as an aid to improve cognition in this patient group.
