Rehabilitation of Cognitive Deficits Poststroke: Systematic Review and Meta-analysis of Randomized Controlled Trials

Systematic Review and Meta-Analysis of Randomized Controlled Trials

Mairead O'Donoghue, BSc; Siobhan Leahy, PhD; Pauline Boland, PhD; Rose Galvin, PhD; John McManus, MD; Sara Hayes, PhD


Stroke. 2022;53(5):1700-1710. 

In This Article


This comprehensive systematic review examined the totality of evidence regarding the effectiveness of nonpharmacological interventions on cognitive deficits poststroke. Findings exhibited the range of rehabilitation approaches that may be effective in improving cognitive deficits poststroke. Multiple component interventions demonstrated favorable effects on outcomes of general cognitive functioning and memory when compared with standard care. Physical activity interventions and NIBS protocols were also effective on a range of cognitive and secondary outcomes. Despite research involving over 4000 individuals across 64 randomized controlled trials, the evidence base presented in this review demonstrated several methodological shortcomings and unanswered questions regarding the optimal design and delivery of interventions to rehabilitate cognitive deficits poststroke. Furthermore, the overall moderate certainty of evidence and high risk of bias for various methodological domains must be acknowledged.

The most consistent evidence in this review supported multiple component interventions, with significant improvement demonstrated for general cognitive function and memory. Evidence from three pooled studies using multiple component interventions indicated a positive effect in general cognitive functioning. Within this analysis of multiple component interventions, cognitive rehabilitation training was used in conjunction with a form of physical activity, such as systemic coordination training,[31] conventional occupational therapy/physiotherapy,[32] or an acupuncture treatment.[33] However, given the very low clinically meaningful effect of this outcome,[34] the practical and clinical importance of multiple component interventions on outcomes of general cognitive function should be interpreted with caution.

Evidence from 2 pooled studies indicated a large effect for spatial and visual memory outcomes post computerized cognitive rehabilitation,[35] in conjunction with traditional rehabilitation therapy consisting of physical therapy and occupational therapy.[36] Previous evidence suggests that a combination of physical activity and cognitive rehabilitation may facilitate greater improvements in relation to cognitive impairment in both older adults and individuals poststroke, respectively.[37,38] Although there is relatively limited literature regarding the use of multiple component interventions in individuals poststroke, combined cognitive interventions and physical activity have shown positive effects on cognitive functioning in persons with dementia[39] and traumatic brain injury.[40] In their review of evidence-based cognitive rehabilitation, Cicerone et al[40] recommended that programs of comprehensive, holistic neuropsychological rehabilitation can improve outcomes relating to community integration and functional independence and recommends that such programs be provided to reduce both cognitive and functional impairments. However, it should be noted that these recommendations were based on studies examining a mixed cohort of individuals poststroke and post-traumatic brain injury and should be extrapolated with caution. Furthermore, given the heterogeneity of interventions across studies in the current review, it remains difficult to make recommendations regarding the optimal design and delivery of a multiple component intervention for individual's poststroke. Nevertheless, the provision of two or more interventions may be more effective compared with usual care and as such, a combination is, therefore, more likely than many single-component interventions to reduce cognitive dysfunction poststroke.

In addition to the benefits of physical activity when combined with a form of cognitive training, this review found favorable results of single-component physical activity interventions. Trials by Pandian et al[41] and Thieme et al[42] found a significant improvement on outcomes of neglect post mirror therapy interventions of 4 and 5 weeks duration respectively. Both these studies were conducted in the acute to subacute phases poststroke and were in comparison with an active control group of sham mirror therapy. Mirror therapy is a feasible and effective intervention that can facilitate both motor and sensory improvements in the acute, subacute, and chronic phases poststroke.[43] It remains unclear whether the intensity of exercise training affects cognitive outcomes poststroke.[26]

The benefits of physical activity on cognitive function are widely known in older adult populations.[44,45] Physical activity and exercise prescription are recommended in the management of not only multiple physical but also psychosocial domains poststroke.[46] However, uncertainty remains regarding the effects of such interventions specifically on cognitive function poststroke. A review by Cumming et al[26] evaluated the effect of increased physical activity on cognitive functioning in individuals poststroke. Although this review demonstrated a significant effect favoring the use of physical activity interventions compared with active control groups on outcomes of neglect and balance, there was large variability between trials regarding the optimal frequency, intensity, timing, and type of physical activity intervention provided. Moreover, the assessment of cognitive function was rarely the primary outcome measure of these trials. Given the methodological shortcomings and small pool of studies within this review by Cumming et al,[26] there was insufficient evidence to support clear recommendations for clinical practice. In their updated review examining the effects of fitness training poststroke, Saunders et al[47] noted that outcomes of cognitive function in particular lack investigation, despite being ranked as the most important research priority relating to life after stroke.[7]

Although meta-analytic evidence from the current review failed to demonstrate favorable effects of cognitive rehabilitation interventions on cognitive or secondary outcomes, Cicerone et al[40] recommend external compensation strategies as a practice standard for the rehabilitation of memory deficits poststroke. However, in their updated review of cognitive rehabilitation for memory deficits poststroke, das Nair et al[17] report insufficient evidence for memory rehabilitation due to poor methodological quality of current studies. Similar to the current review, das Nair et al[17] noted lack of long-term effects of cognitive intervention on memory outcomes. In their updated meta-analysis, Loetscher et al[48] examined 6 randomized controlled trials comparing cognitive rehabilitation with a standard care control group on outcomes of attention. Although there were no significant improvements in global attentional measures, the domain of divided attention improved significantly, but these effects did not persist long-term.[48] Their review highlighted insufficient data to determine whether outcomes varied according to the frequency, intensity, or type of intervention.[48] Given the key features of intervention design. that is, frequency, intensity, time, and type of cognitive rehabilitation interventions have yet to be established,[8] it is unclear whether the intensity and duration of these interventions were justified.

Meta-analytic evidence from this review suggests that NIBS protocols have a favorable effect on outcomes of neglect and performance in activities of daily living poststroke. NIBS, specifically rTMS, was found to have a large effect on neglect outcomes in the acute and subacute phase poststroke when compared with a sham stimulation group.[49–51] Two of these studies combined NIBS with another treatment such as conventional therapy or sensory cueing respectively.[49,51] Our review adds to the existing body of evidence and supports the findings of Salazar et al[52] where NIBS, specifically rTMS and transcranial direct current stimulation, in conjunction with other therapies (such as visuospatial scanning training, conventional rehabilitation treatment, or feedback training) improved neglect in the acute and subacute phase poststroke. Thus, this combined approach, much like that of multiple component interventions, may result in more favorable clinical improvements on symptoms of neglect in individuals poststroke. However, this effect is limited as a result of the large heterogeneity observed in these analyses, suggesting that further studies are needed to determine the effects of NIBS protocols on outcomes of neglect.

Subgroup analysis demonstrated favorable results of occupational-based interventions conducted less than 3 months poststroke on general cognitive functioning. Individual trials included in this review by Ntsiea et al[53] and Skidmore et al[54] found favorable effects in functional ability after a specific workplace intervention and a strategy training intervention respectively. These interventions incorporated goal setting and self-management strategies. These findings are in contrast to a previous Cochrane review of occupational therapy poststroke which found no differences between groups for outcomes of general cognitive functioning or basic activities of daily living as assessed using the Barthel Index.[55] However, it is of note that only one trial met the criteria for inclusion in this review.[55]

Unlike previous research in this area, meta-analytic evidence from the current systematic review evaluated an extensive range of rehabilitation interventions, including multiple component interventions, physical activity interventions, cognitive rehabilitation interventions, NIBS protocols, occupational interventions, and other interventions such as art therapy, prism adaptation, and music therapy. Given the breadth of interventions included in this review, there was extensive heterogeneity in relation to the intervention content, the duration, and dose of the intervention, the delivery of the intervention, and the domain of cognition targeted. To address this, we categorized intervention type accordingly and limited the heterogeneity in the meta-analyses. Furthermore, many of the included studies employed a standard care control group, with few studies describing the actual standard treatment delivered to the control group. This variation between studies raises challenges when aiming to determine the effectiveness of these types of rehabilitation interventions on cognitive deficits poststroke. Studies should provide a comprehensive description of intervention design to facilitate reliable intervention implementation and also to allow for replication in subsequent research.[56] Future trialists should consider the Template for Intervention Description and Replication checklist to guide comprehensive and systematic reporting of interventions.[57]

Another factor adding to the diversity across studies was the lack of consensus between trials regarding assessment of cognitive impairment poststroke. The recent consensus-based recommendations from the Stroke Recovery and Rehabilitation Roundtable highlighted that no consensus was reached regarding a single approach to assessment of cognitive function poststroke.[8] The need for assessments that are easily implementable and target the assessment of cognition across multiple cognitive domains was advocated.[8] The National Institute for Neurological Disorders and Stroke and the Canadian Stroke Network recommend the use of common protocols for neuropsychological assessment in stroke-related research,[58] which would address a current methodological limitation within stroke rehabilitation trials when such assessments cannot be compared.[22] In the current review, there was substantial variability in both the quality and the types of cognitive outcome measures used. The MoCA assesses 6 cognitive domains including episodic memory, working memory, attention/concentration, executive function, language, and visuospatial ability.[59] Pooled data from a recent review demonstrated that the MoCA displays excellent sensitivity but poor specificity.[60] To this end, the MoCA may be particularly useful in people poststroke given the diffuse nature of cognitive impairment poststroke while also showing to have acceptable responsiveness and criterion validity in people poststroke receiving rehabilitation.[34]

Furthermore, it is not clear whether improvements in cognitive functioning lead to improvements in daily functioning or quality of life for the person poststroke. Nine out of the 42 included studies in our meta-analysis demonstrated improved measures of quality of life and/or daily functioning outcomes, along with improved cognitive outcomes. Given the relatively small number of trials that reported improvements in these measures, there are insufficient trials to support or refute the use of the above interventions to improve quality of life and functional outcomes poststroke while simultaneously improving cognitive function. Although it is important to show the functional significance of these rehabilitation interventions, a causative effect of improved cognitive outcomes leading to improved quality of life or daily functioning outcomes cannot be assumed from these results.

Strengths and Limitations

To our knowledge, this is the first systematic review that examined all types of rehabilitation interventions across multiple domains of cognitive function for people poststroke. This review complied with the Preferred Items for Systematic Reviews and Meta-Analyses guidelines[28] and employed a comprehensive search and screening strategy of peer-reviewed research. Unlike previous research which has focused on outcomes within a specific cognitive domain, we have included six domains of cognitive function, as outlined in the Australian Clinical Guidelines for Stroke (2020), as well as general cognitive functioning. Moreover, the comprehensive intervention categorization of included studies provides more in-depth analysis of the effectiveness of a breadth of rehabilitation approaches.

This review has several important limitations. Concerning the domains of cognition examined, this review did not capture the multi-dimensional constructs of each cognitive domain. For example, the domain of memory can be subclassified into working memory, episodic memory, procedural memory, semantic memory, and prospective memory.[61] Although we recognize the interconnected neural network and hierarchical structure of cognitive performance,[23] the analysis of inter-domain interactions occurring during recovery poststroke was beyond the scope of this review. Furthermore, findings need to be interpreted with caution and considered in the context of high risk of bias across the following domains of bias: selection bias, performance bias, detection bias, attrition bias, reporting bias, and other biases.

The comprehensive and pragmatic intervention categorization of included studies provides an in-depth analysis of the effectiveness of rehabilitation approaches. However, the possibility of misclassification of intervention categorization within the analyses is acknowledged. Furthermore, the variation across trials also limited further possible analyses and although we are adding to the novelty of this evidence base, the existence of few homogeneous trials is a limitation.


The evidence regarding the effects of rehabilitation interventions to improve cognitive impairment poststroke remains uncertain. The evidence base presented demonstrates moderate and high risk of bias and suggests there is insufficient evidence to support or refute the effectiveness of rehabilitation interventions on cognitive impairment poststroke. Multiple component interventions, particularly those with a physical activity component, seem to have a favorable effect on memory and overall cognitive functioning. There is some evidence in r of physical activity interventions and NIBS protocols on the cognitive domains of memory and neglect. However, these findings must be interpreted with caution given the moderate certainty of evidence and high risk of bias for various methodological domains, extensive heterogeneity of interventions as well as outcome measures included in these studies.