A controlled clinical effectiveness trial of multimodal cognitive rehabilitation on episodic memory functioning in older adults with traumatic brain injury (TBI)

Experimental design

Results presented in this paper are part of a larger semi-randomized, controlled, before-after study with follow-up at 6 months and blinded outcome measurement clinical trial (study registered at ClinicalTrials.gov, Identifier: NCT04590911). This study is reported according to the TREND statement [15].

Participants

42 French-speaking participants having sustained a TBI (based on the WHO criteria [16]) and meeting the inclusion criteria (see Supplementary material for inclusion/exclusion criteria) were recruited by clinical coordinators from the McGill University Hospital Centre (MUHC), as well as from two outpatient interdisciplinary rehabilitation centres, the Lucie-Bruneau Rehabilitation Centre and Le Bouclier Rehabilitation Centre. Ethical Review Boards of the MUHC as well as from the Centre for Interdisciplinary Research in Rehabilitation to which are affiliated the referring rehabilitation centres provided ethical approval for the study. All participants gave their written informed consent prior to their participation in the study. The study was conducted in compliance with the Helsinki Declaration. A symbolic financial compensation was offered to all participants as a contribution towards their transportation expenses. The CONSORT diagram illustrating participant allocation and flow throughout the trial is presented in Figure 1. Additional details on participant recruitment and allocation are presented as Supplementary material.

• Download figure
• Open in new tab

Figure 1. CONSORT Flow Diagram, modified for partially randomized trial design.

Procedure

The CEP consists of three intervention modules, Introduction and self-awareness, Attention and memory, and Executive functions (strategies detailed in Cisneros et al. [17]; this issue), which include a total of 24 sessions of 90 minutes, delivered twice weekly during in a 12-week time frame. CEP main content and structure, as well as detailed intervention delivery, are presented as Supplementary material. The module targeting memory consists of the MEMO program strategies that have been described elsewhere [12]. MEMO teaches participants to use self-initiated strategies in situations requiring sustained attention and episodic memory. The main goal is to enhance and enrich encoding, and facilitate ulterior recall, by generating new visual images (method of loci, face-name associations), or by analyzing and synthesizing the semantic content of texts (FRST method). In addition to MEMO’s proposed exercises, homework included reading magazines or newspapers articles to practice the FRST method, meeting at least one new person during the week to practice the Face-name method and making one grocery list and one to-do list during the weekend. The CEP was conducted by an experienced clinical neuropsychologist (E.C., experimenter), who had previously received MEMO training, with small groups of 5 participants. Assessments of intervention effects were performed by trained evaluators, who were blinded to group assignment of participants, in two separate sessions lasting about 90 minutes each. There were three time-points: baseline (pre-intervention – T0); 14-weeks (post-intervention – T1); 6-months post-intervention (follow-up – T2). Additional information on assessments can be found in Supplementary material.

Primary outcome measures

Three tasks were chosen as primary outcomes for episodic memory. The same tests as Belleville et al. [12] were used to evaluate the effects of the MEMO program were administered: Face-name association, Word list recall, and Text memory. These tests were selected from the Côte-des-Neiges Computerized Memory Battery [18]. See Supplementary material for more details on the measures used.

Generalization measures

To assess the participant’s perception of changes in memory abilities in daily life after the CEP intervention, we used a self-administered memory questionnaire, the Self-Evaluation Memory Questionnaire (SEMQ) [19]. The SEMQ measures memory performance for 10 dimensions for which are derived mean scores: Conversations, Books & movies, Slips of attention, People, Use of objects, Political & social events, Places, Actions to perform, Personal events, General (this latter dimension assessing the impact of physical, emotional or environmental variables on memory). We also used the French version of the Psychological General Well-Being Index (PGWBI) [20], and a Satisfaction questionnaire asking about the perceived impacts of strategies learned in each CEP module and, overall, on daily life activities.

Control measures

The following measures served as a control of spontaneous recovery. We did not expect significant changes on these measures in any of the groups. Forward and Backward Digit spans and Coding subtests (scaled scores) from the WAIS-III were used. The Montreal Cognitive Assessment (MoCA) was used as a control measure for post-TBI cognitive decline associated with aging [21]. WAIS-III Vocabulary was used as a control measure for dementia suspicion or low intellectual functioning. As a part of general health variables describing the sample, we documented Vascular risk factors of each participant (e.g., arterial hypertension, cardiovascular disease) and family history of these conditions using a health questionnaire.

Statistical analysis

Demographic, clinical characteristics, primary outcomes, generalization and control measures at baseline (T0) were described for each group and compared using Student’s t-test and chi-squared test. Since there was internal variability in both groups in time since injury and the latter could be related to enhanced time-related brain atrophic changes [22,23], we used a mixed model ANCOVA for repeated measures with a repeated factor time (T0 – pre-intervention baseline, T1 – post-intervention, T2 – 6-months follow-up), a factor group (experimental, control) and Time since TBI as a covariate. In this model, the effect of the intervention corresponds to a significant group-by-time interaction. Appropriate post-hoc tests were conducted. Effect sizes were interpreted according to Cohen, as cited by Pallant [24]. Appropriate post hoc comparisons were performed where indicated in this model. Analyses were performed assuming missing data are at random (MAR) with SAS version 9.4 and a significance level of p < .05.

Primary outcome measures

Before intervention (T0), experimental and control groups did not present significant differences on all psychometric variables, except for macrostructure immediate recall on text memory (t(30) = 2.30, p = 0.028) where experimental participants showed higher scores than controls. Table 2 summarizes means and SD at the three assessment times, and ANCOVA group-by-time interaction results after controlling for the effects of the covariate Time since TBI. The analysis showed a strong significant group-by-time interaction for Face-name association, with a large effect size. Supplementary bilateral t-test analysis indicated that scores at T1 were significantly larger for the experimental group than for controls (t(28) = 2.95, p = 0.006; 95% CI [1.05, 5.81]). The pre-post-intervention difference for the experimental group was also significant (t(28) = 2.389, p = 0.024; 95% CI [0.35, 4.62]). These significant differences were, however, not maintained at the 6-months follow-up, even though scores remained higher in the experimental group. Statistics for non-significant results not presented in Table 2 for and completed/attrition participants are presented as Supplementary material.

There was a significant main effect of time on Word list - immediate recall (F(2, 44) = 3.60, p = 0.036), as well as for Word list - delayed recall (F(2, 42) = 9.23, p = 0.001), but there were no significant effects for group or group x time interaction. There was no significant main effect of time, group, nor a group-by-time interaction for Text memory - macrostructure immediate recall. Bilateral t-test showed that the initial baseline (T0) difference between the two groups tended to remain at post-intervention (T1) assessment (t (29) = 1.99, p = 0.056). For Text memory - macrostructure delayed recall, there was a significant main group effect (F(1, 2) = 4.80, p = 0.039), but no effect of time, or significant group-by-time interaction. There were no significant differences for Text memory - microstructure immediate recall and Text memory - microstructure delayed recall.

Generalization measures

No significant between-group differences were found at the baseline assessment on any of the ten dimensions of the SEMQ. There were significant group-by-time interactions on three dimensions: Conversations, Slips of attention, and Political & social Events. The significant group-by-time interaction for Conversations showed a moderate effect size. Confidence intervals adjusted for Time since TBI showed changes in distribution throughout T0, T1 and T2 (95% CIs [-0.1728, 1.2384] [-0.8378, 0.4156] [-0.01142, −1.1234]), indicating some improvement on memorizing conversation content in the experimental group at T1, but showing significance only at the 6-months follow-up (T2). A significant group-by-time interaction was also found for Slips of Attention, with a large effect size. Post-hoc confidence intervals indicated a significant T0-T1 difference (95% CI [−1.07, −0.15]), while the difference between T2 and T0 did not reach significance. Memory for Political & social Events was significantly increased after the CEP intervention in the experimental group, as indicated by a significant group x time interaction with a large effect size. Confidence interval analysis shows that the effect occurred between T0 and T1 (95% CI [-1.56, - 0.10]), while the difference between T0 and T2 did not reach significance.

There were no significant effects for Books & movies, People, Places, and General dimensions of the SEMQ. There was a significant main effect of time for Use of objects (F(2, 40) = 12.44, p < 0.000) and Actions to perform (F(2, 40) = 4.24, p = 0.021), but no significant group effect, nor group × time interaction. No significant effects were found for Personal events, although there was an almost significant group effect (F(1, 21) = 4.24, p = 0.052). In this case, the control group systematically showed less severe problems on this dimension at the three assessment times.

ANCOVA failed to show statistically significant effects on psychological well-being, as reflected in the raw scores of the PGWBI. Additional analyses based on the clinical thresholds (well-being, moderate distress, severe distress) for the PGWBI were performed, and showed clinically meaningful improvements, which are illustrated in Figure 2. Nine (9) participants of the experimental group were clinically classified as severely distressed at baseline (T0) and this number was reduced to 5 after the CEP (T1) and remained relatively stable at 6-months follow-up assessment (T2). Simultaneously, the number of experimental participants classified in the well-being category went from 3 at T0 to 10 at T1 (10/20 participants; 50%) and remained stable six months after (9/17 participants; 53%). In the control group, only one participant progressed toward the well-being category from T0 to T1 and another one at T2. In summary, of the 12 participants having shown improvement, 10 (i.e. 83%) were from the experimental group.

• Download figure
• Open in new tab

Figure 2. Psychological General Well-Being by clinical category at the three assessment times.

Experimental group participants also completed the Satisfaction questionnaire. The mean score for the question “To which extent did the memory module of the CEP have a positive impact on your daily life?” was 4.00 (SD = 0.741) out of a maximum of 5, indicating a perceived positive impact on daily life habits requiring memory abilities. The general satisfaction level for the impact of the entire CEP intervention was 4.75 (SD = 0.476).

Control measures

There were no significant main effects of time, group, nor group-by-time interaction for Forward Digit Span. For Backward Digit Span there was a significant main effect of time (F(2, 40) = 3.33, p = 0.04), but no significant main effect of group, nor group-by-time interaction. There was no significant main effect of time, group, nor group-by-time interaction for Coding.

Effectiveness of the CEP on episodic memory

The findings of this study show that the multimodal CEP intervention improves various modalities of episodic memory in older TBI adults that received the training, compared to active TBI controls who received usual rehabilitation focused on resuming daily activities and social roles, but no cognitive rehabilitation. Comparison of baseline measures with post-CEP ones showed a strong group-by-time interaction on one of three primary outcome measures, Face-name association, as well as for three dimensions of a real-world generalization measure of memory functioning (SEMQ), Conversations, Slips of attention, Political & social events. Experimental participants significantly improved their face-name association abilities and the size of the effect was large. During the CEP intervention, participants learning to associate a person’s name to an interactive image related to a particular feature of the person’s face. In the loci and face-name association methods used in the CEP, the vivid visual imaging representation of a stimulus interacting with other more familiar objects seeks to create novelty at the time of encoding. Such self-initiated novelty has been associated with activation of frontal cortex and its connections, enhancing attention and facilitating encoding, while the presence of multi-modal stimuli boosts simultaneous activation of brain regions associated with verbal (names, words) and visual (facial traits, imagined objects) processing [25]. Our finding of a significant intervention effect on memory of Face-name associations and not word lists are compatible with the fact raised by Bottiroli et al. [26] that, contrary to memorizing word lists or reading material, meeting new people is a natural and usual situation where this method, which is directly associated to the context, can be easily practiced.

Three measures of generalization to daily memory functioning were also improved by the CEP: memorizing Conversations contents and Political & Social events, and Slips of attention, showing moderate or large effect sizes. This indicates that these memory dimensions in daily life activities benefited from the learned CEP strategies that aimed at improving the allocation of sustained and selective attention to significant sources of information, such as conversations and news information, with the intention to memorize it; that is, self-enhancing attention during encoding. As we had hypothesized, in addition to improving neuropsychological test performance related to CEP training, the intervention also increased real-world memory function of experimental participants, whereas TBI controls not receiving the CEP did not demonstrate these effects.

The strategies put forward in the CEP target ‘secondary’ memory which is the type of episodic memory often used in daily life situations, where important amounts of information must be retained [27]. Experimental participants thus learned to enhance self-activation when using the memory strategies not only during the training sessions, but also in daily life contexts, where regular practice of the learned strategies constituted a central part of the intervention. They then shared with the group their accomplishments made during the week using the learned methods. Our results thus indicate that older individuals with TBI can benefit from a cognitive rehabilitation program with a strong episodic memory component, such as that put forward by the CEP. Positive impacts of memory training on neuropsychological and self-report memory measures have also been shown in older healthy individuals as well as persons with MCI [8]. It is interesting to note that validation studies using the SEMQ have shown that in healthy aging the Political and social events and Conversation dimensions items explained most of the variance for the entire measure [28], and that Conversation and Books & movies were the only items related to actual cognitive deficits in MCI [29]. Increased Slips of attention and decreased memory of Political and social events, which are the two other SEMQ dimensions that were significantly modified by intervention in our study, could possibly represent an added effect of TBI during aging, as compared to healthy aging or MCI. Those two SEMQ dimensions were previously found to be related to mild TBI in older adults [30], while Political and social events and Books and movies dimensions were reported as being affected in young adults with mild TBI [31].

With respect to the absence of intervention effects for the other two primary outcome variables, Word list learning and Text memory, some experimental participants showed difficulties with learning the memory strategies for words and texts. Furthermore, we observed that a few participants had difficulties in learning the reading memory method (FRST). Semantic interpretation, verbal abstract reasoning and verbal synthesis generation problems can be present following TBI. In that respect, Cicerone et al. [32] recommended cognitive interventions for specific language impairments such as reading comprehension after TBI.

Intra-group variability could also have been partly responsible for the absence of significant effects for memory of word lists and texts. Variability of cognitive performance being a characteristic effect of TBI, it could increase the pre-existing inter-individual heterogeneity in cognitive aging [33]. Additionally, contrary to methods of loci and face-name association, the testing condition of memory for texts was dissimilar to the learning one. In fact, participants learned to write a synthesis of main and secondary ideas, create a new title for the story, read their notes and then recalled the story. In the testing situation, participants read a text in silence and then recalled the content orally. It is possible that those differences influenced the results in a TBI sample who may have presented flexibility problems. In this regard, Bottiroli et al. [34] suggested that strategy-adaptation training in cognitive rehabilitation programs for older TBI individuals could help enhance the effects of cognitive rehabilitation. Furthermore, older TBI individuals could probably benefit from training that puts more explicit emphasis on the initiation aspect of using the strategies, since these were not prompted during assessments [32,35].

Since individuals with a more severe TBI have increased difficulties in encoding information, older persons with severe TBI may need a more individualized approach. For instance, a higher number of training sessions could be necessary to develop proficiency in all methods. Some studies reporting effective interventions have invested 15 to 30 hours of training on specific memory strategies [36]. On the other hand, it has been found that encoding words in divided attention conditions is more difficult for older healthy adults [30]. Older persons with severe TBI could then have difficulties in learning and using these strategies because they require dividing attention between cognitive tasks.

Stability of change over time

The effects of face-name memory were not as stable as hypothesized. Bottiroli et al. [26,34] suggested additional training sessions to facilitate long-term stability of effects of memory training, as well as specific instructions to elders concerning the more suitable circumstances to apply the learned methods. While the CEP intervention explicitly addressed the latter aspect, additional or ‘booster’ sessions and more diversified real-world practice circumstances could have been profitable.

The significant effects shown on the primary outcome and on daily life generalization can be considered as clinically important. However, differences between experimental and control groups changed at the 6-months follow-up assessment. Attrition in the control group at 6-months follow-up could have, in our view, influenced the results. Although there were no significant differences in primary outcomes and memory generalization measures at baseline assessment between the attrition participants and their counterparts who completed the 6-months follow-up assessment in both groups (see Supplementary material), the control group seemed to improve because of the departure of weaker counterparts. Frenette et al. [37] found that mean MoCA scores in the early phases of mild TBI was highly correlated to cognitive functioning, age, schooling, and severity, where individuals with uncomplicated mild TBI obtained higher scores that those with complicated mild TBI. On the other hand, some studies have shown that TBI individuals who succeed in rehabilitation tend to have higher MoCA scores at the beginning than the participants who don’t achieve rehabilitation goals [38]. This seemed to be the case for the control group having completed the study; these participants obtained a mean MoCA score of 26, while attrition control participants had a mean score of 23.67. Otherwise, it is possible that improvement of control group participants was linked to their receiving usual care in the form of rehabilitation interventions focusing directly on daily activities and life habits.

Generalization to psychological well-being

Psychological factors were not addressed by the CEP interventions. However, we identified potentially clinically important improvements in psychological well-being in experimental participants. Virtually all experimental participants showed an overall improvement in psychological well-being from T0 to T1. While only 3 experimental participants showed a score reflecting well-being status at baseline, this number rose to 10 after the intervention, and this number was stable (i.e., 9) at the 6-months follow-up assessment. Some of the improvements reflected a reduction of psychological distress from severe to moderate. Winocur et al. [39] proposed that self-efficacy could be enhanced by developing strategies that improve cognitive function. Hence, cognitive and memory mastery developed in the experimental group could have had a positive influence on psychological well-being. This was accompanied with a high satisfaction level with regards to positive impacts of memory training, as well as of the overall CEP intervention, on participants’ daily lives. Group experiences also could be an important factor influencing these changes: mutual support and sharing experiences are some of the benefits of group participation, in addition to the already demonstrated cognitive benefits. Considering that depression is frequently seen following TBI even if age still a controversial risk factor [40], group interventions may represent a promising approach to influence mood through a cognitive training.

Study limitations

The present study has some limitations that should be taken into consideration when interpreting our findings. It is impossible to determine if only the memory module of the CEP intervention is related to the observed changes in the experimental group. Such a multimodal approach to cognitive rehabilitation is warranted clinically and is evidence-based considering the important variability in the aging process and clinical manifestations of TBI. However, the fact that intervention effects were found on tests measuring memory abilities that were targeted by the memory module of the CEP suggests that there was a direct effect of memory intervention. Also, the small size or our sample, with the largest proportion of participants having a mild TBI, does limit the scope of conclusions, and we have also pointed out earlier the possible influence of attrition on the results. However, effects sizes of significant changes were moderate to large, indicating that these were strong effects that were related to the memory interventions. Our experimental group received the CEP intervention as well as usual care (holistic rehabilitation interventions without cognitive rehabilitation) while active controls also received usual care, albeit without any cognitive rehabilitation. There thus remains the possibility that some effects of CEP interventions were masked by the positive influence of this holistic rehabilitation approach. At the same time, we considered that any significant change favoring the experimental group would mean that interventions effects were more robust.