The Impact of Reminiscence on Autobiographical Memory, Cognition and Psychological Well-Being in Healthy Older Adults

Design

This study employed a mixed factorial design, with reminiscence condition as the between-participants factor and pre- and post-assessment as the within-participants factor. Participants were randomly assigned on a per-site basis to one of three conditions (reminiscence therapy, active control group, waiting list control group). Those in the control groups were offered participation in a reminiscence group following their completion of the post-control research visit.

Materials

Autobiographical memory was assessed using the Episodic Autobiographical Memory Interview (EAMI; Irish, Lawlor, O’Mara, & Coen, 2008, 2010, 2011). This interview assesses episodic and semantic autobiographical memory, as well as autonoetic aspects of autobiographical memory, for five different life epochs (0-15, 15-30, 31-45 years of age, 46 years of age-last 5 years, last 5 years). Participants were requested to think of different memories than what they had discussed at baseline, or during any reminiscence sessions, when they completed the EAMI for a second time, following the intervention/waiting list. We requested this in order to avoid participants building on any rehearsal of memories they had recently retrieved.

New verbal learning was assessed with the Rey Auditory Verbal Learning Test (RAVLT; Schmidt, 1996). This assesses learning of a 15-word list; there are five trials in which the list is read and participants are required to recall as many words as they can in any order. These five trials are followed by one trial of an interference list. Participants are then requested to recall as many words from the first list, both immediately after the interference trial and at delay (following completion of the EAMI).

We also assessed two aspects of executive function; response inhibition and verbal fluency. We assessed response inhibition using a computerised colour-naming Stroop task (Balota et al., 2010; Doyle, Smeaton, Roche, & Boran, 2018), presented using E-prime^®. Participants were presented with a series of words and asked to name the colour of ink in which they were presented. Although previous research also included a word-naming block, we omitted this to reduce participant burden. We assessed verbal fluency from the Montreal Cognitive Assessment (MoCA; Nasreddine et al., 2005); participants were required to think of as many words beginning with “F” as they could (or “A” for the parallel version) in one minute.

We employed a number of self-report measures of subjective quality of life and satisfaction with life, mental health and affect; these were the Control, Autonomy, Self-realisation and Pleasure quality of life questionnaire (CASP-19; Netuveli, Wiggins, Hildon, Montgomery, & Blane, 2006); Satisfaction With Life Scale (SWLS; Diener, Emmons, Larsen, & Griffin 1985), Beck Depression Inventory (BDI; Beck, Steer, & Brown, 1996), Spielberger Trait Anxiety Inventory (STAI; Spielberger, Gorsuch, Lushene, Vagg, & Jacobs, 1983), and Positive and Negative Affect Schedule (PANAS; Watson, Clark, & Tellegen, 1988).

Participants

We included 36 participants in the analysis for this study; one participant was excluded following baseline assessment due to evidence of cognitive impairment. We assigned 14 to the reminiscence condition (10 females, 4 males; M_age = 69, SEM = 1.3), 11 to the active control (10 female, 1 male, M_age = 68.8, SEM = 1.5) and 11 to the no-intervention control (8 female, 3 male, M_age = 71.9, SEM = 2.0). All participants identified as White Irish. Participants were recruited from Dublin and surrounding counties in the Republic of Ireland, via flyers and newspaper advertisements, as well as via community groups (e.g. active ageing groups, men’s sheds). All participants were over 60 years of age. Exclusion criteria were as follows: severe visual impairments; history of psychological/neurological impairment, severe head trauma resulting in loss of consciousness, history of epilepsy, currently taking psychoactive medication, other relevant medical conditions, history of drug or alcohol problems (See Figure 1 for recruitment flowchart).

Interventions

The reminiscence and active control interventions were delivered in small groups of three to seven participants at a number of different venues. The reminiscence and active control groups comprised six weekly sessions, each lasting one hour. The reminiscence groups used a simple reminiscence approach, adapted from the “remembering yesterday, caring today” program (Schweitzer & Bruce, 2008). This program was used in previous research examining autobiographical memory as an outcome (Woods et al., 2016). The sessions were more conversation-based than the “remembering yesterday, caring today”, which had a stronger focus on hands-on activities for working with clients with memory problems. We did not target a specific function of reminiscence as we were interested in observing which functions of reminiscence would emerge in the group context.

For the reminiscence groups, in addition to autobiographical memories, participants were prompted to discuss memories of historical events in the public sphere (e.g. John F. Kennedy’s visit to Ireland in the 1960’s, economic recession in Ireland in the 1980’s). For the active control groups, the topics of discussion were approximately matched to the content of the reminiscence groups, but with a focus on the present and future. For example, the active control group discussed learning new things whilst the reminiscence group discussed schooldays. As certain topics were more likely to provoke discussion of different epochs in autobiographical memory, the order of topics for the groups were reversed for half the groups (i.e. half of the reminiscence groups completed the sessions in one order, and the other half of the reminiscence groups completed them in the opposite order). The order of topics was similarly reversed for half of the active control groups. The weekly topics for the reminiscence group and active control group are outlined in Table 1 below.

Reminiscence sessions were recorded, and, with participants’ permission, highlights from the reminiscence groups were added to an online archive at the Digital Repository of Ireland (see Supplementary Materials). Participants were encouraged to listen to previous entries in the archive, to help trigger memories of their own.

Procedure

Participants completed the cognitive measures in a small testing room one at a time. This testing session lasted approximately 60-90 minutes. Participants completed the questionnaires in their own time prior to the first group session. After completing the reminiscence/active control group, or waiting an equivalent amount of time in the waiting list condition, participants completed the same measures again, 6-8 weeks after the baseline testing session. Parallel versions of tests were used wherever available for the post-intervention study visit, to minimise learning or other carryover effects.

Statistical Analysis

Where assumptions of parametric data were violated, we transformed variables using a square root transformation; where that transformation did not make the data parametric then a natural log transformation was used. Data for participants who competed the baseline assessment but were lost to follow-up were imputed using last observation carried forward.

We used ANCOVA for the primary outcome (autobiographical memory) and MANCOVA (using Roy’s largest root; see Field, 2009) for the secondary outcomes, which were executive function (verbal fluency and Stroop effect reaction time), verbal learning (RAVLT performance indices) and psychological well-being (quality of life, satisfaction with life, depression, anxiety, positive and negative affect). We employed reminiscence condition and pre-post assessment as predictors and age as a covariate. We assessed the interaction between the two predictors to determine if the change in outcome from baseline to post-intervention differed between the reminiscence group and the control conditions. We also examined whether there was a main effect of life epoch on autobiographical memory.

Autobiographical Memory

Autobiographical memory was the primary outcome assessed within this study. There was a significant interaction between pre-post assessment and reminiscence condition for episodic memory, F(2, 32) = 3.98, p = .03, ${\mathrm{\eta }}_{\mathrm{p}}^2$ = .2; this was due to a fall in episodic memory retrieval in the active control group, rather than a change in the reminiscence group. There was no significant main effect of reminiscence condition, F(2, 32) = 0.72, p = .49, ${\mathrm{\eta }}_{\mathrm{p}}^2$ = .04. However, there was a main effect of pre-post assessment, F(1, 32) = 6.46, p = .02, ${\mathrm{\eta }}_{\mathrm{p}}^2$ = .17, with participants performing slightly worse at post-intervention regardless of reminiscence condition (see Figure 2A).

There was no significant effect of reminiscence condition on semantic autobiographical memory; there no significant interaction between pre-post assessment and reminiscence condition, F(2, 32) = 2.67, p = .09, ${\mathrm{\eta }}_{\mathrm{p}}^2$ = .14, nor was there a main effect of reminiscence condition, F(2, 32) = 0.54, p = .59, ${\mathrm{\eta }}_{\mathrm{p}}^2$ = .03 or pre-post assessment, F(1, 32) = 0.08, p = .78, ${\mathrm{\eta }}_{\mathrm{p}}^2$ = .002 (see Figure 2B).

Despite an increase in the no intervention condition, there was no significant interaction between pre-post assessment and reminiscence condition for extent of re-experiencing emotion associated with autobiographical memory, F(2, 32) = 2.49, p = .1, ${\mathrm{\eta }}_{\mathrm{p}}^2$ = .14. There was a marginally significant main effect of pre-post assessment, F(1, 32) = 4.16, p = .05, ${\mathrm{\eta }}_{\mathrm{p}}^2$ = .12, with participants generally reporting higher levels of re-experiencing emotion at post-intervention. There was no significant main effect of reminiscence condition on re-experiencing emotion, F(2, 32) = 0.94, p = .4, ${\mathrm{\eta }}_{\mathrm{p}}^2$ = .06 (see Figure 2C).

There was no significant interaction between pre-post assessment and reminiscence condition for extent of reliving the memory, F(2, 32) = 0.2, p = .82, ${\mathrm{\eta }}_{\mathrm{p}}^2$ = .01. There was a marginally significant main effect of pre-post assessment, F(1, 32) = 3.89, p = .06, ${\mathrm{\eta }}_{\mathrm{p}}^2$ = .11, with participants generally reporting lower levels of reliving the memory at post-intervention. There was no significant main effect of reminiscence condition on reliving the memory, F(2, 32) = 1.03, p = .37, ${\mathrm{\eta }}_{\mathrm{p}}^2$ = .06 (see Figure 2D).

There was some evidence of lower autobiographical memory performance for the childhood period (0-15 years), although there was no significant main effect of life epoch on episodic memory, F(4, 128) = 0.69, p = .60, ${\mathrm{\eta }}_{\mathrm{p}}^2$ = .02 (see Figure 3A), or semantic memory, F(3.05, 97.52) = 0.67, p = .57, ${\mathrm{\eta }}_{\mathrm{p}}^2$ = .02 (Greenhouse-Geisser adjusted; see Figure 3B).

Life epoch did have a main effect on re-experiencing of emotion, F(4, 104) = 3.14, p = .02, ${\mathrm{\eta }}_{\mathrm{p}}^2$ = .11, with participants reporting less re-experiencing of emotion for childhood memories (see Figure 3C). Despite a simliar trend for reliving of the event, there was not a significant main effect of life epoch, F(4, 104) = 2.2, p = .07, ${\mathrm{\eta }}_{\mathrm{p}}^2$ = .08 (see Figure 3D). These main effects were not significantly moderated by condition or pre-post assessment.

Cognition

Using Roy’s Largest Root, there was no significant interaction between pre-post assessment and reminiscence condition for executive function (verbal fluency and Stroop effect RT), Θ = .12, F(2, 28) = 1.72, p = .2, ${\mathrm{\eta }}_{\mathrm{p}}^2$ = .11. There was no significant main effect of pre-post assessment Θ = .03, F(2, 27) = 0.35, p = .71, ${\mathrm{\eta }}_{\mathrm{p}}^2$ = .03, nor was there a significant main effect of reminiscence condition, Θ = .17, F(2, 28) = 2.32, p = .12, ${\mathrm{\eta }}_{\mathrm{p}}^2$ = .14 (see Table 2).

There was no significant interaction between pre-post assessment and reminiscence condition for new learning, Θ = .23, F(7, 27) = 0.88, p = .53, ${\mathrm{\eta }}_{\mathrm{p}}^2$ = .19. There was no significant main effect of pre-post assessment, Θ = .31, F(7, 26) = 1.15, p = .37, ${\mathrm{\eta }}_{\mathrm{p}}^2$ = .24, nor was there a main effect of reminiscence condition, although this approached significance, Θ = .57, F(7, 27) = 2.21, p = .07, ${\mathrm{\eta }}_{\mathrm{p}}^2$ = .36. (see Table 3).

Psychological Well-Being

There was no significant interaction between pre-post assessment and reminiscence condition for psychological well-being, Θ = .14, F(6, 14) = 0.33, p = .91, ${\mathrm{\eta }}_{\mathrm{p}}^2$ = .12. There was no significant main effect of pre-post assessment, Θ = .92, F(6, 13) = 1.99, p = .14, ${\mathrm{\eta }}_{\mathrm{p}}^2$ = .48, nor was there a main effect of reminiscence condition, although this approached significance, Θ = 1.2, F(6, 14) = 2.8, p = .053, ${\mathrm{\eta }}_{\mathrm{p}}^2$ = .55 (see Table 4).