Introduction

Cognitive impairment is the primary symptom of schizophrenia (Bellani et al., 2019). However, because a diagnosis of schizophrenia is not based on cognitive impairment, patients with cognitive impairment do not receive cognitive enhancement drugs, and the related symptoms are often overlooked. The efficacy of typical antipsychotics (Guercio et al., 2019) and conventional psychological education interventions (Tripathi et al., 2018) has been limited. Nonpharmacological therapies have been a primary focus of recent research (Irazoki et al., 2020). A Japanese study showed aerobic exercise could effectively improve cognitive function in patients with schizophrenia (Shimada et al., 2022). However, environmental, symptomatic, and motivational factors make exercise interventions difficult to implement in these patients, who have physical activity barriers (Chen et al., 2022).

Working memory and executive function in people with schizophrenia are worse than in healthy adults (Alkan et al., 2021) because of reduced functional connectivity among their sensorimotor, basal ganglia, default mode, and visual networks (Penadés et al., 2020). Therefore, their ability to handle complex tasks in daily life is significantly reduced. People with schizophrenia have poor intelligence, memory, and executive function, and patients in long-term care institutions experience more severe cognitive impairment as they age (McCleery & Nuechterlein, 2019). Cognitive impairment negatively impacts important neurocognitive areas such as attention, executive function, and working memory, resulting in weaker interpersonal relationships and independent living ability, leading people with schizophrenia to become socially disabled (Green et al., 2018). Cognitive rehabilitation may help patients with schizophrenia take on stable employment, which is known to facilitate social functioning (Cervello et al., 2021).

Haddad et al. (2021), summarizing previous articles, proposed that people with schizophrenia have cognitive function deficiencies, particularly in terms of attention, working memory, executive functions, social cognition, verbal learning, and memory. Several important areas are especially relevant to cognitive function. The many neurons in gray matter allow it to process information and output new information. The gray matter throughout the central nervous system enables individuals to control their movement, memory, and emotions. Gray matter stores and processes information, affecting the higher-order cognitive function required to perform planning, organizing, and other complex, goal-directed tasks. The prefrontal cortex is activated by most higher-order processing tasks, including sustained attention, working memory (message logging, access, and retrieval), emotion processing, language, and executive function (monitoring, organization, planning, and deciding). Working memory involves the storage, processing, and updating of information (Roalf & Gur, 2017). The medial temporal lobe controls memory, visual, perception, and stimulus recognition. In particular, the hippocampus, parahippocampus, and amygdala are closely involved in the formation of working memory. Cognitive remediation, cognitive rehabilitation, and cognitive training are known to increase plasticity in prefrontal cortex nerves, which has a protective effect on the volume of gray matter in the medial temporal lobe. Thus, a “normalizing” effect on resting-state networks and functional activity in the frontal–temporal network, amygdala, and anterior cingulate cortex can improve cognitive function (Bellani et al., 2019).

Interventions for cognitive impairment include cognitive stimulation, cognitive remediation, cognitive rehabilitation, and cognitive training, with the latter three sharing many treatment elements. Cognitive stimulation draws on the theoretical concept of cognitive stimulation, which exploits the diversity of perception modalities associated with situations in everyday life. It stimulates the encoding, consolidation, and retrieval of messages in visual and semantic modes. Cognitive stimulation therapy, developed by Spector et al. (2003), is a 14-session course for patients with Alzheimer's that uses multisensory stimulation and learning mechanisms to improve cognitive functions such as attention, memory, and orientation. Whether through cognitive remediation, rehabilitation, or training, the goal of these sessions is to stimulate new learning of cognitive tasks, develop strategies to reduce errors in the learning process, and improve the cognition, memory, and executive function of patients. The intervention methods used in schizophrenia treatment have evolved from paper, pencil, and task-based interventions to a method that integrates therapist-led cognitive improvement tasks with computer-based training and monolithic computer software training. Regardless of whether it is conducted individually or in a group, therapists are expected to maintain a positive attitude when discussing cognitive improvement strategies and giving encouragement and feedback (Zhu et al., 2022). In recent years, electronic games have also emerged as a tool in schizophrenia treatment. Although playing electronic games can increase the volume of gray matter in the brain and represents a novel treatment for cognition (Meiland et al., 2017), the research results are inconsistent with regard to improving cognitive function (Jahshan et al., 2019; Powers et al., 2013). Thus, further studies are needed.

Environmental conditions and the standardized curriculum used in rehabilitation units allow patients few opportunities to improve their cognitive impairment. Therefore, this study was designed to develop and investigate the efficacy of a group cognitive stimulation training (GCST) intervention on cognitive function, working memory, executive function, and social function.

Methods Study Design

This was a randomized controlled trial with convenience sampling. Following the requirements of the Consolidated Standards of Reporting Trials statement, the principal investigator was the interventionist, and the names of the participants who had completed the pretest and information on their ward of treatment were sent to the noninterventionist researcher by email (see Figure 1). Random Allocation Software Version 2 was used to randomly assign 38 people each to the experimental group (EG) and control group (CG). The data were collected by trained researchers who had completed consistency training. Both the EG and CG received conventional treatments, whereas only the EG received the GCST.

Figure 1:

Recruitment and Participant FlowchartNote. MoCA = Montreal Cognitive Assessment Scale, Taiwan version; GCST = group cognitive stimulation training; EG = experience group; CG = control group; AAD = against advice discharge.

Participants and Setting

The participants were recruited from seven chronic wards of a psychiatric hospital in northern Taiwan. The data collection period was from January 1 to May 3, 2021. The inclusion criteria were as follows: (a) diagnosis of schizophrenia or schizoaffective disorder > 2 years earlier based on diagnostic and statistical manual of mental disorders, fifth edition, (b) 20–65 years old, (c) regular use of antipsychotic drugs, and (d) Montreal Cognitive Assessment (MoCA) score of 10–25. The exclusion criteria were (a) brain injury, (b) intellectual disturbance, (c) substance abuse, (d) severe psychotic symptoms and emotional instability, (e) receiving cognitive behavioral therapy, and (f) visual or hearing impairment.

G*Power Version 3.1.2 was used for sample size estimation (Faul et al., 2007). On the basis of the results of the repeated-measures analysis of variance between-factors model, to achieve 80% power at a 5% level of significance (two-sided) per arm with an effect size of 0.29 (Vita et al., 2021), two measurements, and a correlation value of .05, a minimum sample size of 72 was needed. A measure of effect size, Cohen's d, was estimated for the Group × Time interaction by comparing change scores through t test and then converting the t statistic using the formula d = 2(t)/sqrt(df).

Intervention Interventionist training and preparation

The interventionist had a master's degree and 18 years of clinical nursing experience in psychiatry. The GCST activity manual was designed by referring to the structured activity pattern of CST (Spector et al., 2003). Between August and October 2020, the 14 GCST courses were conducted in a non-study-targeted psychiatric ward on a small group. A second facilitator participated as an observer. Discussions and corrections followed each group session in preparation for the formal study. The appropriateness of GCST course content and implementation was evaluated by experts who had studied and practiced in the cognitive training group (clinical psychologist and occupational therapist) using a 1–5 point scale. In this study, courses with a content validity index above .80 were retained, with minor revisions made based on expert recommendations.

Group cognitive stimulation training program

The GCST was a 7-week activity with 14 sessions. Each session lasted 60 minutes and was led by the first author. The EG was further divided into six groups of four to eight people who attended sessions in the activity rooms of different wards. The therapeutic elements of GCST include (a) cognitive stimulation, focusing on multisensory stimulation, learning mechanisms, and problem solving, and (b) cognitive remediation/rehabilitation/training, that is, individualized cognitive improvement strategies that use cognitive aids and computerized training to engage in communication and supportive discussions in small groups.

Each GCST course (Table 1) began with a 10-minute opening session, self-introduction, and warm-up, followed by a 10-minute “passing and catching” activity, for which the therapist designated a theme (e.g., animals). The first participant had to name a kind of animal and then toss a ball to another participant, who said the animal mentioned by the previous person as well as a new one, with each subsequent catcher required to name all of the animals named so far. This game strengthened working memory through repeated practice.

Table 1 - Group Cognitive Stimulation Training Program No. Practice Skills/Section 1. Opening and warm-up (every time) 2. Pass and catch the ball (every time) 3. Cognitive remediation/cognitive rehabilitation/cognitive training module: three sessions  Session 1: Listening and writing practice: Sequential memory of numbers and reverse memory of numbers  Session 2: Visual and auditory memory training  Session 3: (1) Clap your hands after listening to the command. (2) Graphical discrimination. (3) Word discrimination. 4. Cognitive training module: seven sessions  Session 4: Electronic games—Magic Piano Tiles  Session 5: Electronic games—everyday brain exercise (reaction and judgment training)  Session 6: Electronic games—Air Hockey  Session 7: Electronic games—Right Brain Tempo (image learning game)  Session 8: Electronic games—Moles Attack  Session 9: Electronic games—animals memory game  Session 10: Electronic games—Star Memory (graphics pairing, sequence and location memory training) 5. Cognitive stimulation module: four sessions  Session 11: Object classification  Session 12: Associate memory  Session 13: Quick pairing; logic training  Session 14: Happy clock drawing and creation 6. Experience share and feedback (every time)

The first three sections included three modules: cognitive remediation, cognitive rehabilitation, and cognitive training. Session 1 covered listening and writing practice as well as the recall and reverse recall of number sequences. After listening to the therapist read out a sequence of three to eight numbers in sequential and inverse orders, the participants immediately used a pen to write the numbers in those orders on a small white board. Session 2 covered visual and auditory memory training. The visual training component asked participants to arrange unordered cards into the right position of the cards they had seen 30 seconds earlier. The auditory training component asked participants to listen to the therapist's instruction and then quickly move the card to the designated correct position. Session 3 included asking the participants to clap their hands after hearing a command (i.e., after hearing a number spoken by the therapist, participants quickly clapped their hands for an amount of times equal to that number). This session also included a graphical discrimination task, in which the participants used a pen to circle the differences between two images on a card, and a word discrimination task, in which the participants were tasked to find specific words among a large number of characters sharing similar glyphs.

The next seven sessions (Sessions 4–10) were cognitive training modules. The therapist explained the rules of a video game using a PowerPoint presentation and then showed how to use the game software and its operating procedures. The participants were then invited to play the electronic game according to the instructions. During play, the therapist observed their reactions, provided timely guidance, and gave encouragement. The seven electronic-game-based sessions included magic piano tiles, an everyday brain exercise (reaction and judgment training), air hockey, right brain tempo (image-learning game), mole attack, animal memory game, and star memory (graphical pairing, sequence and location memory training).

Session 11 covered object classification, in which the therapist successively showed five to six pictures in each of five groups, and the participants were asked to read and remember them. After putting the cards away, the participants recalled and named each picture seen. In addition, the participants were asked to classify the shapes and functions of the picture images. Session 12 covered associated memory, in which participants were asked to group Taiwanese scenes with their associated local cuisines, with more words resulting in higher scores. Session 13 covered quick pairing (arranging together cards with the same colors and shapes) and logic training (zodiac signs and sorting each county and city in order or reverse order). Session 14 covered happy clock drawing and creation. Finally, each session included 10 minutes at the end for the participants to share experiences and feedback.

Control group

Participants in the CG continued to perform routine therapeutic activities in their wards, including walking, singing, and watching television, during the study period.

Measures Participant demographic data

The nine demographic characteristics studied as control variables included gender, age, marital status, employed status, duration of disease, years of education, benzodiazepine usage, antidepressant usage, and antipsychotic dose equivalents (Inada & Inagaki, 2015).

Primary outcomes

Primary outcomes included cognitive function (MoCA), working memory (Wechsler Memory Scale–Third Edition [WMS-III]), and executive function (Taiwan Frontal Assessment Battery [TFAB]).

1. Interrater reliability

MoCA, WMS-III, and TFAB were used as the research instruments in this study. The research assistant was trained by a clinical psychologist in the hospital and followed the operation manual before taking the test. Interrater reliability was .994–1.000.

2. Cognitive function

MoCA, developed by Nasreddine et al. (2005) to screen for mild cognitive impairment, has eight cognitive domains, including attention, executive function, memory, language, visual–structural skills, abstract thinking, calculation, and orientation. The total possible score is 30, with ≥ 26 interpreted as normal, 18–26 as mild, 10–17 as moderate, and < 10 as severe cognitive impairment. MoCA has a good Cronbach's alpha (.83) and test–retest reliability (r = .92, p < .001) and has a good correlation coefficient with the Mini-Mental State Examination (r = .87, p < .001). In this study, the Taiwanese Chinese version of MoCA was used, which earned a total scale reliability (Cronbach’s alpha) value of .86 and was highly correlated with Mini-Mental State Examination (r = .92, p < .001; Tsai & Fuh, 2010).

3. Working memory

WMS-III was exclusively revised and published in Taiwan by NCS Pearson Inc. under authorization from the Chinese Behavioral Science Corporation from the original work of Wechsler (1997). WMS-III was designed to measure the therapeutic effect on cognitive function. The Number–Word Sequence Test subscale of WMS-III is designed to measure working memory in people with chronic schizophrenia, with a total score of 0–21, a split-half reliability of .74–.96, and a test–retest reliability of .47–.83. Regarding criterion-related validity, Wechsler Memory Scale–Revised had r = .36–.72, Children's Memory Scale had r = .26–.74, and WMS-III had r = .26–.82. Spatial Memory Span, a visual-based memory span test, addresses both sequential and reverse-order spatial memory and has a total score range of 0–32. The score of the Number–Word Sequence Test may be used as an indicator of working memory, with higher scores representing better working memory. In this study, the Cronbach's alpha of the WMS-III was .68.

4. Executive function

Frontal Assessment Battery was developed by Dubois et al. (2000). It has shown good internal consistency (Cronbach's alpha = .78), good concurrent validity with the Mattis Dementia Rating Scale (r = .82, p < .001), and good discriminant validity with progressive supranuclear palsy. The TFAB was used in this study, which includes the six dimensions of conceptualization, mental flexibility, motor programming, sensitivity to interference, inhibitory control, and environmental autonomy (Wang et al., 2015). The total possible score ranges from 0 to 18, with higher scores representing better executive function. The TFAB previously earned a Cronbach's alpha of .68 and a split-half reliability of .53 and showed good criterion-related validity and construct validity (Wang et al., 2015). In this study, the Cronbach's alpha of the TFAB was .51.

Secondary outcome

Social function: The Social Function Scale was developed by Birchwood et al. (1990). The Social Function Scale–Taiwan short version (SFST) was translated into Traditional Chinese and psychometrically evaluated by Song (2001). This scale covers 36 items in the seven dimensions of social/withdrawal, interpersonal communication, independence-ability, independence-performance, entertainment, sociality, and occupation/employment, with a score ≥ 71 indicating high functioning, 52–70 indicating moderate to excellent, 34–51 indicating fair to moderate, and ≤ 33 indicating low functioning. The SFST has a Cronbach's alpha of .52–.86, a test–retest reliability of .75–.94, and good discriminant and construct validity (Song, 2001). In this study, the Cronbach's alpha of the SFST was .86.

Data Collection

We first selected patients who met the inclusion criteria and then explained the purpose and procedures of the study. The participants took the MoCA test after signing the informed consent form, and those who scored 10–25 then completed the pretest demographic characteristics datasheet and SFST. Medical record data collection and working memory and executive function tests were completed within 1 week. Data for this study were collected both before and after the intervention (Week 8). The outcome indicators were collected by a trained research assistant who did not know the grouping of the research participants.

Ethical Considerations

This study was approved by the institutional review board (IRB-1081209-04) of the enrolled hospital and was registered in the ClinicalTrials.gov Protocol Registration and Results System (NCT-04916483). In line with the principles of the Helsinki Declaration, the participants kept a copy of the written informed consent form and had the right to terminate their participation at any time without providing any reason and with no effect on their future medical care.

Data Analysis

Data analyses were performed using IBM SPSS Statistics 24.0 (IBM Inc., Armonk, NY, USA). Descriptive statistics included mean (SD), frequency, and percentage. The data used intention-to-treat analysis to deal with the loss or change in samples after random assignment. Continuous variables were analyzed using the independent-sample t test, and categorical data were analyzed using the chi-square test or Fisher's exact test. Outcome indicators included cognitive function, working memory, executive function, and social function. For changes in outcome indicators, we used generalized estimating equation (GEE; Liang & Zeger, 1986), after controlling for nine demographic characteristics, to explore the effects of the Group × Time interaction on cognitive function, working memory, executive function, and social function after the GCST intervention. Data from 68 participants in the EG and CG were available for and included in the analysis. The effect size of GCST on cognitive function was 0.45, the correlation between pretest and posttest was .68, and the post hoc power was .98.

Results Participant Demographic and Clinical Characteristics

In this study, 131 participants were recruited and screened (Figure 1). The average age of the 76 included participants was 51.84 years, and the average number of years of disease was 28.5. The average MoCA score was 19.75 (SD = 3.74), and the average working memory index score was 86.14 (SD = 16.40). The average TFAB corrected score was 8.53 (SD = 3.88). The average SFST total score was 46.92 (SD = 12.42).

This study applied intention-to-treat analysis. There were 38 participants each in the EG and CG. Demographic characteristics at baseline and outcome indicators for the two groups are compared in Table 2, showing no significant difference between the groups. The GCST attendance rate was 79%–100%. Sixty-three participants completed the entire study (dropout rate: 10.53%), and no injury or unexpected results occurred during the experiment. The post hoc power was .98.

Table 2 - Comparison of Demographics and Outcome Indicators Between the Control Group (CG) and Experimental Group (EG) at Baseline (N = 76) Variable CG (n = 38) EG (n = 38) χ2 p n % n % Gender 0.48 .488  Female 23 60.5 20 52.6  Male 15 39.5 18 47.4 Marital status 2.11 .147  No 22 57.9 28 73.7  Yes 16 42.1 10 26.3 Employment status 0.23 .634  No 25 65.8 23 60.5  Yes 13 34.2 15 39.5 First-generation antipsychotics 0.08 .783  No 30 78.9 29 76.3  Yes 8 21.1 9 23.7 Second-generation antipsychotics 0.16 1.000 a  No 4 10.5 3 7.9  Yes 34 89.5 35 92.1 Variable Mean SD Mean SD t p Age 52.58 6.62 51.11 7.87 −0.88 .380 Duration of disease (years) 29.47 8.00 27.53 10.22 −0.93 .358 Years of education 10.21 2.77 9.26 2.93 −1.45 .152 CPZ (100 mg) equivalent 469.84 273.49 477.63 283.88 −0.12 .903 Diazepam (10 mg) equivalent 5.07 5.65 5.33 9.60 −0.15 .885 Imipramine (50 mg) equivalent 0.00 0.00 0.99 6.08 1.00 .324 MoCA 19.39 3.60 20.11 3.88 0.83 .411 Working memory 84.92 15.21 87.37 17.63 −0.65 .519 TFAB 8.48 3.81 8.57 3.99 0.10 .919 SFST 48.95 11.50 44.89 13.12 −1.43 .156

Note. CPZ = chlorpromazine; MoCA = Montreal Cognitive Assessment Scale, Taiwan version; TFAB = Taiwanese version of the Frontal Assessment Battery; SFST = Social Function Scale–Taiwan short version.

a Fisher's exact test.

Primary Outcome Effects of group cognitive stimulation training

There were three primary outcome indicators. The pattern of results for the effect of GCST is displayed in Figure 2. Table 3 shows the differences in the changes within and between the two groups during the intervention as well as the results of GEE, adjusted for the effects of the nine covariates. The size of these effects (as indexed by Cohen's d) is also shown in the table.

Figure 2:

Changes in Outcomes, Between-group Comparisons for Cognitive and Social Function (N = 68)Note. GCST = group cognitive stimulation training; MoCA = Montreal Cognitive Assessment Scale, Taiwan version; TFAB = Taiwan Frontal Assessment Battery; SFST = Social Function Scale–Taiwan Short Version.

Table 3 - Effects of Group Cognitive Stimulation Training on Cognitive Function, Working Memory, and Executive Function at Baseline and Posttest Variable Control Group 95% CI Experimental Group M (SD) B SE M (SD) p ES Primary Outcome: MoCA  T0 19.39 (3.60) 20.11 (3.88)  T1 20.86 (3.69) 22.94 (3.79)  Intercept 20.64 3.15 [14.47, 26.82] < .001  Group 0.62 0.81 [-0.97, 2.22] .444  Time 1.32 0.45 [0.44, 2.21] .003  Group x Time 1.33 0.65 [0.06, 2.59] .040 0.45  Scale 11.68 Primary Outcome: Working memory  T0 84.92 (15.21) 87.37 (17.63)  T1 87.83 (13.98) 95.15 (16.50)  Intercept 73.61 12.06 [49.97, 97.25] < .001  Group 3.42 3.15 [-2.75, 9.59] .277  Time 2.03 1.86 [-1.60, 5.67] .273  Group x Time 4.79 2.66 [-0.42, 10.00] .071 0.40  Scale 175.21 Primary Outcome: TFAB  T0 8.48 (3.81) 8.57 (3.99)  T1 10.17 (3.24) 10.71 (4.28)  Intercept