INTRODUCTION
Cancer continues to be a leading cause of death worldwide; thus, optimizing survivorship care is a critical priority [1]. As cancer mortality has decreased due to recent screening and treatment advancements [2], survivors are experiencing various sequelae, decreasing the quality of individual life and causing emotional and economic burdens on society [3]. Therefore, interest in managing clinical sequelae to enhance the quality of life of cancer survivors is increasing [2,3]. Patients with cancer experience cognitive impairment, especially after chemotherapy and radiotherapy [2]. Cancer-related cognitive impairment, refers to different kinds of decline in memory and thinking such as slow processing speed and attentional or learning problems, which can be triggered by the cancer itself, its treatments, or related factors [4,5]. One of the mechanisms explaining cancer-related cognitive impairment is inflammatory cytokines induced by chemotherapeutic drugs can cross the blood-brain barrier and hippocampal neurotoxicity [4,6]. However, dysfunction of hippocampally mediated memory tasks can be caused by non-central neurocytoma itself [6]. Therefore, cancer-related cognitive impairment has been proposed as a cancer-related cognitive impairment because cancer alone can cause cognitive dysfunction [5].
The incidence of cancer-related cognitive impairment is reported to be approximately 17% to 70%[7]; however, cancer-related cognitive impairment is relatively considered not serious compared with other side effects. The onset of cancer-related cognitive impairment is not easily detected because 35% of cancer survivors experience late effects of cognitive dysfunction even after completing the treatment [8]. Therefore, appropriate neurorehabilitation programs such as cognitive behavior therapy [9,10], web/app based digital intervention [10-12], and cognitive exercise [11] are essential as cancer-related cognitive impairment limits the activities of daily and becomes a disability factor in returning to a society of cancer survivors.
However, previous studies on these interventions have reported inconsistent results regarding their efficacy [11]. Furthermore, there is a lack of comprehensive evidence regarding the optimal intervention methods, such as specific program types, duration, and session length, for maximizing cognitive recovery [13]. Therefore, this study aimed to systematically review neurorehabilitation program studies conducted on cancer survivors to understand the content and characteristics of the program and to perform a meta-analysis to analyze the effect sizes on cognitive-related outcome variables.
METHODS
1. Study Design
This meta-analysis was conducted to analyze the results of neurorehabilitation programs conducted for cancer survivors according to the Preferred Reporting Items for Systematic Reviews and Meta-analyses 2020 statement [14].
2. Selection and Exclusion Criteria
The selection and exclusion criteria for this study were determined based on core questions, and the literature was selected accordingly. Among all the papers written in English, the literature selection and exclusion criteria were specified focusing on participant, intervention, comparison, outcome, and study design.
Participants: Studies were included that featured adult cancer survivors aged 20 years or older who had experienced cognitive impairment related to their cancer diagnosis or treatment, including both first diagnosis and relapse. These participants were required to be eligible to live daily, participate in neurorehabilitation programs, and have their objective and subjective cognitive functions evaluated using outcome scales.
Intervention: Neurorehabilitation intervention aimed to improve cognitive function of cancer survivors. Regardless of the location, disease duration, or intervention frequency, individual interventions were provided to cancer survivors, interventions with groups or families were included in the review, and patients treated with meditation only were excluded.
Comparison: Comparison was those who did not receive a neurorehabilitation program in this study or who were set as a control group in the previous study as general education, counseling, and wait-list.
Outcomes: Outcomes refer to the quantitative values of the outcome variable measured using the measurement tool after performing the neurorehabilitation program on cancer survivors. When measured several times, the mean and standard deviation values before and after were used to determine the effectiveness of the program. Outcome indicators associated with cognitive function measured in cancer survivors included both positive and negative indicators.
Study design: Considering that finding a randomized controlled group experimental design that identifies the effects of providing programs to specific subjects in historical societies for cancer survivors is difficult [15], this study also included an experimental design study with a control group and a pilot experimental study. However, experimental studies, qualitative studies, descriptive research studies, meta-analyses, and longitudinal studies without a control group were excluded.
3. Literature Search Method
The literature targeted all papers published until December 17, 2022. When searching for literature, the study design was limited to experimental studies targeting people in addition to search words, the language was limited to English articles, and the publication form was limited to papers published in academic journals (Supplementary file 1). If articles have the same author, the study periods, patients, and variables were evaluated, and non-overlapping articles were included. A bibliography management program (EndNote X7) was used to classify the literature and remove duplicate papers. For literature search, professional search databases CINAHL, Cochrane Library, PubMed and Embase.
4. Data Extraction Techniques
For systematic literature analysis, data on general characteristics (author name, publication year, and country), study patients (number of patients, age, gender, and type of diagnostic cancer), study intervention (program, intervention period, time per intervention, and place of implementation), comparison patients (control type), study results (results and measurement tools), and study design were extracted. Two researchers independently extracted and evaluated the data of each study using a predetermined framework, and the results were compared. If a disagreement ensued, discussions were made until an agreement was found.
As a result, the intervention program was grouped into a similar type and finally classified into four types: cognitive rehabilitation (CR), mindful-based stress reduction (MBSR), computer-based cognitive training (CCT), and strategies training (ST). The study results were grouped by referring to the fact that when measuring results in each study, they were mainly grouped into objective cognitive function (OCF) and subjective cognitive function (SCF) to achieve results.
The following three objective cognitive function subvariables were found: i) function (FC) includes information processing, response speed, etc.; ii) memory (MM) includes working memory, memory composite, etc.; and iii) status (SS) includes neuropsychological status, attention, etc. The subjective cognitive function also consists of four subvariables: i) perceived cognition (PC) includes cognitive difficulties of daily living, perceived cognitive abilities, problems of daily cognitive function etc.; ii) concerns (CC) include comments from others, cognition general concerns, etc.; iii) memory (MM) includes retrospective and prospective memory, etc.; and iv) quality of life (QL) includes illness identity, psychological well-being, etc. (Table 1).
5. Assessment of Bias Risk
Studies presenting incomplete data—specifically lacking pre- and post-intervention means and standard deviations for experimental and control groups—were excluded from the analysis. The methodological quality of the included studies was assessed independently by two researchers using the Cochrane risk-of-bias tool (RoB 2) [16]. If discrepancies occurred, items were discussed until an agreement was reached. The RoB 2 tool assesses risk of bias across five domains, including the randomization process, deviations from the intended interventions, missing outcome data, measurement of the outcome, and selection of the reported result. The signaling questions were answered using four options (Yes, Probably yes, Probably no, No information), and based on these results, each domain was evaluated as low risk, some concerns, or high risk.
6. Meta-analysis Method
In this study, the Comprehensive Meta-Analysis version 4.0 (Biostatinc., Englewood, NJ, USA) program was used to calculate the overall effect size of the neurorehabilitation program for cancer survivors with chemotherapy experience and to verify the publication bias of the entire study. Since the literature selected in this study has heterogeneity in the participants' cancer diagnosis name, age, country, program type, and outcome variable measurement tool, a single effect size cannot be assumed; thus, it was analyzed using a random effects model to analyze the change in the average effect size. To evaluate the effectiveness of neurorehabilitation programs on cancer survivors, the Hedges'g value and 95% confidence interval (CI) were used to verify the statistical significance using the average value and standard deviation of the subjective cognitive function and control groups measured in each literature, and the effect size was visually confirmed using forest plots. Since the study's purpose aims to confirm the summary effect size of each literature on the effect of neurorehabilitation programs on cognitive function, a total of 85 variable values for subjective and objective cognitive function were integrated and analyzed for each paper to avoid duplicate calculations. The scale directions were all the same, and all the effects were the same in this study. Hedges's g method was selected to eliminate the possibility that the effect size could be overestimated when the number of participants was small [17]. The effect size was judged to be small if the total effect size was 0.2, a medium effect size if 0.5, and a large effect size of 0.8 [18]. In addition, to examine potential moderators, pre-planned subgroup analyses were conducted based on the following categories: cognitive function type (objective vs. subjective), program type (cognitive rehabilitation, mindfulness-based stress reduction, computer-based cognitive training, and strategies training), session duration (1 hour or less, over 1 to 2 hours, more than 2 hours), and total intervention duration (4 weeks or less, 5 to 7 weeks, more than 7 weeks). These analyses were conducted only when at least two studies were included per subgroup. A forest plot was performed to determine the degree of heterogeneity in the effect size, and the statistical review (Higgins I2 homogeneity test) value quantifying inconsistency was also confirmed. In general, I2 value of 25% is interpreted as low; 50%, moderate; and 75%, high heterogeneity [19]. To evaluate the degree of publication bias, trim-andfill and visual funnel plot analysis were performed, and sensitivity analysis was additionally performed.
RESULTS
1. Results of Literature Selection
According to the data selection criteria, 10 studies were included in the systematic literature review, and the data selection process was as follows. The number of papers searched through the search strategy was 10,288. Of the searched papers, duplicate papers of 3,605 were excluded, and the remaining papers of 6,683 were first selected by reviewing the titles and abstracts. Of the first selected papers, 62 were secondarily selected, except for 6,621 for non-total papers and abstracts. The original text was reviewed for 62 papers selected for the second time, and of these, 7 studies that did not meet the selection criteria, including non-adult and cancer survivors, 26 conducted interventions other than neurorehabilitation programs at the same time, 12 that presented only some of the pre- and post-research results of the experimental group and the control group, 7 conducted non-experimental studies. Finally, 10 that could be quantitatively synthesized were meta-analyzed (Figure 1) (Appendix 1).
2. Results of the Quality Assessment of the Literature
Results of the bias risk assessment for 10 documents finally selected for meta-analysis shows high risk with 5 studies and some concerns with 5 studies. There were low bias in deviations from the intended intervention and in the measurement of the outcome. However, a high risk of bias was observed in some studies regarding the randomization process and selection of the reported results (Appendix 2).
3. Characteristics of the Included Literature
This study selected 10 studies on the effectiveness of neurorehabilitation programs for cancer survivors through a systematic literature review. As for the publication year of the studies, six (60.0%) papers were published in 2017~ 2022 [A2,A5-A8,A10] and four in 2012~2016 [A1,A3,A4,A9]. Five studies were conducted in the United States of America (50.0%) [A1,A3,A4,A7,A10], three in Australia (30.0%) [A5,A6,A9], and two in Canada (20.0%) [A2,A8]. Seven were randomized control experimental studies (70.0%) [A1-A5,A8,A10] and three non-random control experimental studies (30.0%) [A6,A7,A9]. The total number of participants was 448, with 249 in the experimental group (55.6%) and 199 in the control group (44.4%). Six (60.0%) studies set the control group as wait-list control [A2,A3, A5-A8], and 4 (40.0%) did not [A1,A4,A9,A10]. The average age of the study participants was 53.6 years in the experimental group and 53.9 years in the control group, and 410 (91.5%) were females and 38 (8.5%) were males. Excluding one paper that did not indicate the exact cancer diagnostic name, five (55.6%) of the nine papers conducted a study on one cancer [A2,A3,A7,A8,A10], two (22.2%) on two cancers [A4,A5], and two (22.0%) on six to seven cancers [A6,A9]. The most common cancer diagnosis of participants was breast cancer in eight and colon cancer in three papers. In addition, various cancers such as leukemia, primary brain tumor, endometrial cancer, lymphoma, appendix cancer, ovarian cancer, and testicular cancer were included.
The study results were largely divided into the subjective cognitive function (k=10) and objective cognitive function (k=9), and a meta-analysis was conducted. The neurorehabilitation programs were categorized into four types: cognitive rehabilitation (k=3), mindfulness-based stress reduction (k=2), strategies training (k=2), and computerbased cognitive training (k=3). Regarding the intervention setting, only four studies provided specific descriptions; of these, three (75.0%) were conducted in independent spaces, and one (25.0%) was conducted in a hospital. Regarding the session duration, nine studies were analyzed excluding one study that reported only total duration. Session times ranged from 30 minutes to 2.5 hours. The most common duration was over 1 to 2 hours (k=4, 44.5%), followed by 1 hour or less (k=3, 33.3%), and more than 2 hours (k=2, 22.2%). Regarding total intervention duration, the length ranged from 4 to 10 weeks. The most common duration was more than 7 weeks (k=5, 50%), followed by 5 to 7 weeks (k=3, 30%), and 4 weeks or less (k=2, 20%).
4. Results of Meta-analysis
1) Effectiveness of neurorehabilitation program for cancer survivors
To analyze the magnitude of the effects of neurorehabilitation programs on the cognitive function of cancer survivors, a random effect model was applied for metaanalysis of 10 finally selected documents. As a result, the overall effect size of the neurorehabilitation program on the cognitive function of cancer survivors was found to be small (Hedges' g=0.45, 95% CI=0.29~0.61), which was statistically significant (Z=5.57, p<.001). Therefore, cancer survivors with neurorehabilitation programs have higher cognitive function than cancer survivors who do not (Figure 2).
2) Subgroup effectiveness of neurorehabilitation programs for cancer survivors
In the meta-analysis, the degree of heterogeneity of all 10 studies was high (I2=80.1%, Q=45.20, dF=9, p<.001), and subgroup analysis was conducted. The subgroups were categorized based on cognitive function types, program types, intervention hours per session, and intervention period (Table 2).
Subgroup analysis was first performed on objective cognitive function and subjective cognitive function. Both domains showed statistically significant small-to-moderate effect sizes. Specifically, subjective cognitive function demonstrated a slightly larger effect size (Hedges' g=0.45, 95% CI=0.28~0.63, Z=5.05, p<.001) compared to objective cognitive function (Hedges' g=0.43, 95% CI=0.24~0.61, Z=4.57, p<.001).
Regarding intervention modalities, analyses were conducted on four intervention. Cognitive rehabilitation (Hedges' g=0.56, 95% CI=0.22~0.89, Z=3.27, p=.001), mindfulness-based stress reduction (Hedges' g=0.54, 95% CI=0.10 ~0.97, Z=2.41, p=.016) and strategies training (Hedges' g=0.51, 95% CI=0.06~0.96, Z=2.22, and p=.027) all demonstrated statistically significant moderate effect sizes. Specifically, cognitive rehabilitation showed the largest effect size, followed by mindfulness-based stress reduction and strategies training. However, computer-based cognitive training did not yield a statistically significant effect size.
The analysis of session duration revealed that sessions lasting over 1 to 2 hours yielded the most optimal and statistically significant moderate effect size.(Hedges' g=0.51, 95% CI=0.19~0.84, Z=3.11, p=.002). Sessions of 1 hour or less also showed a significant, albeit slightly lower, effect size (Hedges' g=0.49, 95% CI=0.14~0.84, Z=2.72, p=.006). In contrast, sessions more than 2 hours were not statistically significant, suggesting that excessively long sessions may not be effective.
In the analysis of total intervention duration, interventions lasting 5 to 7 weeks demonstrated the largest statistically significant moderate effect size (Hedges' g=0.56, 95% CI=0.23~0.89, Z=3.32, p=.001). The subgroup of more than 7 weeks also showed a significant but comparatively smaller effect size (Hedges' g=0.48, 95% CI=0.21~0.75, Z=3.44, p=.001). However, intervention periods of 4 weeks or less did not produce a statistically significant effect, suggesting that a sufficient duration is required to achieve cognitive improvement.
3) Publishing bias
To evaluate the publication bias of the study on the effectiveness of neurorehabilitation programs in cancer survivors, a funnel plot was generated for visual inspection. The plot showed that the individual effect sizes were distributed relatively symmetrically around the central line, suggesting no substantial publication bias. However, two studies appeared slightly displaced from the central axis, prompting further assessment using a trim-and-fill procedure. The trim-and-fill analysis indicated that no additional studies were imputed to adjust for potential publication bias, and the adjusted effect size remained identical to the original estimate, confirming the absence of publication bias (Figure 3). As a result of the sensitivity analysis, no changes in the overall effect size were observed, except for the study with the largest weight (Hedges' g=0.45, 95% CI=0.29~0.61, p<.001).
DISCUSSION
This study conducted a meta-analysis to verify the effectiveness of the intervention program on the cognitive impairment experienced by patients with cancer. Studies included in the research analysis showed relatively geographical bias with three US studies, two in Canada, and three in Australia. In addition, most studies on the cognitive function of cancer survivors were patients with breast cancer. Female patients with breast cancer may accelerate their transition to menopause due to cancer treatment and often fall into the transition period to menopause [20]. Since menopause is known to be associated with cognitive impairment [20], distinguishing between cognitive impairment and clarity caused by cancer treatment is difficult. Therefore, interventions for cognitive problems associated with various types of cancer should be performed in the future. However, since the number of research papers analyzed as the research selection criteria in this study was not large, further research including other criteria should be performed in future studies. In addition, various cognitive rehabilitation programs with interest should be developed and applied in the cognitive problems of cancer survivors. However, a systematic literature review reported that the correlation between self-reported cognitive symptoms and objective cognitive problems was insufficient [5]. Therefore, the types of cognitive impairment should be distinguished, and the intervention determined.
Remarkably, neurorehabilitation programs were effective in objective cognition function, although it was small in effect size. Recent brain imaging studies have reported that it affects brain structure changes and network connectivity after chemotherapy [21]. Functional connectivity of the anterior cingulate cortex by calculating resting-state functional magnetic resonance imaging (fMRI) of breast cancer survivors was lower than the healthy group. Rehabilitation of a damaged brain can help activate the reconnection of neural circuits [22]. Restorative, comprehensive, and contextualized cognitive rehabilitation programs can improve recovery from brain damage by utilizing the principles of neuroplasticity [23], thereby directly targeting and enhancing objective cognitive function sub-domains such as function and memory. Therefore, neurorehabilitation programs using principles of neuroplasticity can improve the cognitive ability of cancer survivors.
Neurorehabilitation programs also showed a significant small effect on subjective cognitive function. A previous study revealed that breast cancer survivors reported subjective execution dysfunction and worse memory ability compared with healthy participants [21]; thus, effective intervention is necessary. In particular, psychological interventions such as mindfulness-based stress program are inferred to have primarily contributed to the improvement of subjective cognitive function sub-domains like perceived cognition and quality of life by alleviating anxiety, fatigue, and depression, which are strong factors influencing cognitive impairment [5,24]. Given that subjective cognitive impairment is strongly influenced by psychological distress, the small yet significant improvement observed in subjective cognitive function suggests that neurorehabilitation programs may play an important role in enhancing survivors' perceived cognitive abilities. In addition, cognitive impairment significantly degrades the quality of life of patients and their families [5,25]. Therefore, these findings suggest that participation in neurorehabilitation programs may help survivors manage cognitive challenges while also promoting meaningful improvements in their overall well-being.
Among the types of neurorehabilitation programs, cognitive rehabilitation, mindfulness-based stress reduction, and strategies training programs had a moderate effect, whereas interestingly, computer-based cognitive training had no effect. The non-efficacy of computer-based cognitive training is likely attributable mainly to patient characteristics, such as the low digital literacy of older patients, and limitations in the intervention mechanism, such as a lack of real-time feedback and social interaction [12]. However, since the COVID-19 pandemic, virtual reality or metaverse-based non-face-to-face educational intervention programs have been actively developed due to advancements in computer technology, warranting further verification of their effectiveness [26]. Therefore, future computer-based cognitive training development should focus on complementing personalized difficulty adjustment and social connectivity. In addition to intervention type, our sub-group analysis highlighted the importance of intervention dosage. We found that interventions lasting 5 to 7 weeks with sessions of over 1 to 2 hours yielded the largest effect sizes. This suggests that such duration strikes an optimal balance, providing sufficient time for cognitive adaptation while preventing fatigue related to limited physical endurance.
However, even after accounting for these program delivery methods, the overall heterogeneity remained high. This suggests that other clinical and methodological diversities across the included studies also contributed to the variability in effect sizes. First, diagnostic heterogeneity is a likely source. While a significant portion of the included studies focused on breast cancer, our analysis also encompassed various other cancer types, such as brain tumors and hematologic malignancies. Variations in cancer type, stage, and associated treatments including different chemotherapy regimens and radiation influence the underlying mechanism and severity of cognitive impairment, thereby leading to differences in responsiveness to interventions [27]. Second, diversity in intervention modality contributed to heterogeneity. Distinct modalities target different cognitive domains and utilize unique mechanisms, creating divergence in effect sizes [28]. Third, the mode of delivery differed, with a mix of in-person and web-based interventions. In-person sessions generally offer stronger social interaction and immediate feedback that can enhance motivation, whereas web-based formats might result in varying levels of participant adherence and engagement, contributing to the variability in outcomes [29]. Finally, the diversity of assessment measures played a crucial role. Specifically, the included studies utilized both self-reported scales and objective neuropsychological batteries. Previous literature suggests a discrepancy between subjective symptoms and objective performance, with subjective measures often being more strongly correlated with psychological distress than with actual cognitive deficits [30]. Consequently, pooling these disparate metrics contributes to statistical heterogeneity
The findings of this study demonstrated that neurorehabilitation programs are effective in enhancing both objective and subjective cognitive functions among cancer survivors. However, since all studies included in this meta-analysis were conducted outside of Korea, the generalizability of the findings may be limited. Although both domestic and international studies were initially considered during the data collection process, domestic studies were ultimately excluded due to the absence of control groups and the limited availability of neurorehabilitation programs specifically designed to improve cognitive function in cancer survivors. Therefore, future meta-analyses that incorporate domestic research are warranted. Furthermore, given the period during which this study was conducted, the studies included in this review were carried out before the widespread adoption of advanced artificial intelligence (AI)-based interventions. As a result, this review was unable to reflect the presence or effectiveness of neurorehabilitation programs incorporating the latest AI technologies, representing an additional limitation. To address this gap, future research should include recent domestic and international studies employing AI-based neurorehabilitation programs to provide a more comprehensive and evidence-based analysis. In addition, the development and implementation of advanced psychiatric nursing interventions, such as those incorporating artificial intelligence and digital-based neurorehabilitation programs, should be actively pursued to improve cognitive function among Korean cancer survivors.
CONCLUSION
This meta-analysis provides clear evidence that neurorehabilitation programs are effective in enhancing both general and specific cognitive functions among cancer survivors, despite the modest overall effect size. Optimal outcomes were achieved through structured interventions consisting of sessions lasting over 1 to 2 hours, delivered over 5 to 7 weeks. Cognitive challenges in cancer survivors, although often underestimated due to their limited association with survival, significantly influence daily functioning, psychological health, and quality of life. As the number of cancer survivors continues to grow, addressing cognitive dysfunction has emerged as a critical component of comprehensive survivorship care. Therefore, it is necessary to identify the types of cognitive impairments and provide individualized, evidence-based interventions that promote cognitive recovery and adaptation. From a psychiatric nursing perspective, these findings emphasize the importance of early assessment and tailored neurorehabilitation nursing interventions to improve cognitive health among cancer survivors. Cancer-related cognitive impairment is closely associated with psychological distress, including anxiety, depression, and fatigue, which may exacerbate cognitive decline. The demonstrated effectiveness of neurorehabilitation programs, particularly those utilizing neuroplasticity and mindfulness approaches, underscores the necessity of holistic psychiatric nursing interventions that address both cognitive deficits and emotional well-being. Furthermore, integrating such evidence-based approaches into nursing practice can enhance holistic and patient-centered care, contributing to an improved quality of life in cancer survivorship.
PDF Links
PubReader
ePub Link
XML Download
Full text via DOI
Download Citation
Supplement
