This meta-analysis was conducted according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. The systematic review methodology employed in this study has been registered on the Prospero platform, although it is currently awaiting approval.
Search strategy
A comprehensive scoping search was conducted to gather research published until May 11, 2023. Two independent authors (WBJ and JLM) searched the PubMed, Embase, Cochrane Library, Web of Science, China National Knowledge Infrastructure (CNKI), China VIP Database (VIP), China Biomedical Literature Database (CBM), and Wanfang Database. To ensure the effectiveness of the search, a combination of MeSH terms and relevant keywords was employed. The search terms for this review were ("dementia" or "Alzheimer's disease" or "dementia *" or "Alzheimer's disease *" or "senile dementia") and ("creativity" or "creative expression *" or "creative expression therapy" or "timeslips"). The retrieval strategy was jointly determined by two evaluators and involved a manual search of references in selected studies to identify articles that fulfilled the predetermined criteria.
Qualifications
The inclusion criteria for this review are determined by the "PICOS" principle. The target population consists of individuals aged 60 and above, who were diagnosed with dementia according to the established diagnostic criteria. The intervention focuses on the experimental group receiving creative story therapy, specifically storytelling intervention, while the control group receives routine nursing intervention. The primary outcome of interest centers around overall cognitive function, which can be assessed through standardized scales such as the Mini Mental State Examination (MMSE), the Montreal Cognitive Assessment (MoCA), and the Alzheimer's Disease Assessment Scale Cognitive Subscale (ADAS-cog). Secondary outcomes encompass depression, which can be evaluated using the Kangnai Depression Scale, and quality of life, which can be assessed using validated and standardized quality of life-specific scales for dementia patients, such as the Alzheimer's Disease Quality of Life Scale (QOL-AD). Communication skills can be evaluated through the Functional Assessment of Communication Skills (SFACS). Moreover, the included studies must be randomized controlled trials published in either Chinese or English. On the other hand, the exclusion criteria encompass reviews, qualitative research, conference abstracts, and experimental protocols. Additionally, original research articles without full-text availability or missing key outcome indicator data were excluded. Repeatedly published articles by the same author or team were also excluded. Furthermore, studies with low-quality literature evaluations are not considered in this review.
Research Selection and Data Extraction
The collected research was imported into the EndNoteX9.1 software. Duplicate studies were promptly identified and removed. The title and abstract of each article are then carefully assessed, allowing the reviewers to select studies that meet the predetermined inclusion criteria. To maintain accuracy and consistency, two independent reviewers, WBJ and JLM, screened and extracted the literature. They conduct thorough cross-checking to ensure the reliability of the selected studies. In case of any differences or disagreements, JLM and LYM engaged in thoughtful discussions to reach a consensus. Alternatively, if needed, they seek advice from a third-party expert to resolve any remaining discrepancies. The following characteristics were extracted : author information, publication year, patient age, sample size, details of the intervention measures, duration of the intervention, and assessment scales.
Quality assessment
Two evaluators, WBJ and JLM, independently employed the esteemed Cochrane Risk Bias Assessment Tool[23]. This tool, specifically designed for randomized controlled studies, comprises seven essential questions that comprehensively evaluate various aspects of study quality. These questions encompass critical domains, including random sequence generation, allocation concealment, blinding of implementation, blinding of outcome assessment, completeness of outcome data, selective outcome reporting, and potential biases. Each item is carefully classified as either low risk, unclear, or high risk, indicating the level of potential bias associated with the study. To ensure accuracy and consistency, any discrepancies or differences in evaluation are meticulously addressed. In such cases, the Third Reviewer, LYM, serves as an impartial mediator, providing invaluable expertise to resolve any remaining disagreements and guide the final assessment.
Statistical analysis
We used two software tools, RevMan 5.4 and Stata software (version 14), to perform comprehensive statistical analysis. All the outcome indicators examined in this meta-analysis consist of continuous variables. Depending on the consistency of the evaluation tools employed across studies, we will utilize different effect indicators. Specifically, when the evaluation tools are consistent, we will employ the mean difference (MD) as the effect indicator. Conversely, when variations in evaluation tools are inconsistent, we will employ the standardized mean difference (SMD) as the effect indicator. To determine the magnitude of the effect size, we will utilize Cohen's criterion. According to Cohen's criterion, effect sizes between SMD (or MD) ≥ 0.20 and < 0.50 are considered small, SMD (or MD) ≥ 0.50 and < 0.8 are considered moderate, and SMD (or MD) ≥ 0.8 are considered large, thereby offering a comprehensive understanding of the impact.
Heterogeneity among the included studies was assessed through the I2 statistic and chi-square test statistics, and a random effects model was used when heterogeneity was more than 50%. Otherwise, a fixed effects model will be applied to combine the data. subgroup analysis was conducted to further investigate the potential sources of heterogeneity and sensitivity analysis was conducted to assess the impact of individual studies on the overall findings and the stability of the results. Additionally, to evaluate the potential publication bias, funnel plots will be utilized, providing a visual representation of the distribution of studies and their corresponding effect sizes. Furthermore, to detect and quantify any publication bias present, we will employ both funnel plots and Egg's test, a widely recognized method for assessing publication bias.[24]