Interventions to improve the occupational performance of people with post stroke upper limb apraxia – A systematic review and meta-analysis

Abstract

Introduction: Upper limb apraxia is a post stroke disorder affecting the persons’ ability to perform everyday activities. This review aimed to determine the effectiveness of interventions on occupational performance outcomes. Method: A systematic review of literature (2000-2022) across five electronic databases was conducted. PRISMA guidelines were applied. Data were pooled using RevMan. Findings: Four studies reporting findings from three randomised controlled trials were included. The methodological quality of studies was low. Three treatment approaches were reported: (1) strategy training (2) gesture training (3) combined gesture and strategy training. Strategy training alone or in combination with gesture training was significantly more effective than control interventions in improving occupational performance scores (FEM, mean difference: 1.08, 95% confidence interval: −6.01–8.16, I2 = 0%). Conclusion: This review provides low quality evidence to support the use of strategy training alone, or in combination with gesture training, by Occupational Therapists to improve occupational performance and apraxia scores post intervention among people with post stroke upper limb apraxia.

Introduction

Limb apraxia, a subtype of apraxia (Leiguarda and Marsden, 2000), is a common, complex, higher-order cognitive-motor deficit post-stroke with reported prevalence rates ranging from 30% to 50% (Dovern et al., 2012). Limb apraxia is operationally defined as a ‘disorder of learned skilled movements that cannot be explained by sensorimotor or comprehension deficits’ (Foundas and Duncan, 2019). Limb apraxia is commonly associated with lesions of the left (dominant) hemisphere, is often coexistent with aphasia and is often present in both limbs (Bounis and Binkifski, 2023). People after stroke with limb apraxia may be unable to gesture or may experience difficulty participating in activities of daily living (ADLs) (Bounis and Binkifski, 2023) due to sequencing errors, spatial temporal movement errors and tool use/content errors (Van Heugten et al., 2000a; Bieńkiewicz et al., 2014).

Terminology is used inconsistently in the field of apraxia research, and it is noted that confusion about the term ‘apraxia’ persists (Foundas and Duncan, 2019). Understanding of limb apraxia is developing based on neuroanatomy research that has demonstrated the hierarchical organisation of the disorder and the role of several visuomotor pathways involved in integrating perceptual and semantic information necessary for the performance of purposeful movements (Binkofski, 2020). Clinicians and researchers predominantly reference two types of apraxia; ideational and ideomotor (Buxbaum et al., 2008; Goldenberg, 2013; Lindsten-McQueen et al., 2014). Both ideational and ideomotor subtypes of limb apraxia can be described (Leiguarda and Marsden, 2000). Ideational apraxia occurs when patients experience a breakdown with the knowledge required for a task and present with difficulties performing a sequence of actions requiring the use of objects in the correct manner and order to complete their intended goal, for example, making a cup of coffee with tea and sugar. Ideomotor apraxia describes compromised ability to pantomime actions, mimic tool use without holding objects and/or difficulty with gesture production. Gesture production is usually divided into transitive (object/tool use) and intransitive (non-tool related) acts. The idea or plan of action is not impaired (i.e. the patient knows what to do), but the translation into the required motor programme is disrupted (i.e. the patient does not know how to do it) (Pérez-Mármol et al., 2015).

A comprehensive review of interventions for limb apraxia has described 10 treatment approaches; multiple cues, error reduction, six-stage task hierarchy, conductive education, strategy training, transitive/intransitive gesture training, rehabilitative treatment and errorless completion + exploration training (Buxbaum et al., 2008). In the same year, a Cochrane review of interventions for motor apraxia after stroke was also published including three trials with a total of 132 participants. All three trials used different treatment interventions: strategy training, gesture training and a transfer of training approach. The review concluded that there was insufficient evidence to support or refute the effectiveness of specific therapeutic interventions for motor apraxia after stroke (West et al., 2008). A further systematic review of interventions for apraxia was published in 2014 and concluded that the best practice for the treatment of apraxia appears to be task-specific strategy training (Lindsten-McQueen et al., 2014). A limitation of this review was the use of FAME scoring (Pearson, 2004) to rate the quality of research evidence included, this approach is not widely used and is qualitative in nature. Furthermore, the database searches for this review were concluded in 2012. Alashram et al. (2021) published a systematic review concluding that the effects of strategy training and gesture training on apraxia post-stroke are promising. This review also has some limitations, it was not prospectively registered, it includes studies without control groups, PEDro was utilised to assess the methodological quality of randomised controlled trials (RCTs) and non-RCTs included in the review, a meta-analysis was not conducted, and the search concluded in September 2020. Confidence that the intervention of interest is the reason for changes or improvements in participant outcomes is affected by the inclusion of studies without control groups (Malay and Chung 2012).

Given the prevalence of post-stroke limb apraxia and the impact of apraxia on occupational performance, a comprehensive, contemporary and robust systematic review to evaluate the totality of evidence in relation to the effectiveness of interventions for upper limb apraxia post-stroke to influence occupational performance is warranted.

Methods

Design and search strategy

The Preferred Reporting Items for Systematic Reviews Meta Analysis (PRISMA) guidelines were followed in the conduct and reporting of this systematic review (Page et al., 2021). The study protocol was registered on Prospero in July 2020. Six electronic databases were searched by the lead author in March 2021 (The Cochrane Library, PubMed/Medline, CINAHL (Ebsco), Embase, PsycINFO, Google Scholar), and an updated search was completed in July 2022. Search terms utilised are shown in Appendix 1. The search terms included three core concepts: stroke, apraxia and interventions and were developed based on the Cochrane review on this topic (West et al., 2008). The search strategy was limited to include adults with a diagnosis of stroke and peer-reviewed articles in the English language published from January 2000 to July 2022. The year 2000 was the year of the earliest RCT included in the Cochrane review. Reference lists of the selected studies were reviewed to check for any additional eligible studies.

Screening and study selection

Database searches were uploaded to Mendeley reference manager (2019) and duplicates were removed. Articles were screened by title and abstract based on the inclusion criteria by two independent authors (ST and KL). Studies included were then full text screened for eligibility. Results from the search strategy and screening process are displayed in Figure 1.

Figure 1.

PRISMA flow chart showing database search (Moher et al, 2009).

Selection criteria

Studies were included if they met the following inclusion criteria:

• Population: Participants with a diagnosis of stroke and upper limb apraxia (confirmed using standardised assessment, both ideational and ideomotor subtypes of limb apraxia were included).

• Intervention: Any rehabilitation intervention targeting the upper limb. Studies in which interventions targeted speech apraxia only were excluded. Studies evaluating deep brain stimulation, cognitive assistive technologies (e.g. Cogwatch, Coach) and surgery or pharmaceutical interventions were also excluded.

• Comparison: Active or usual care control groups were included. The definition of usual care operated by each individual study is described.

• Outcomes: The primary outcome measure was a change in occupational performance measured using a standardised or non-standardised assessment. Measures of apraxia were also included.

• Study design: All RCTs, quasi-RCTs, controlled before and after studies and interrupted time series designs were included. Case studies were excluded.

Data extraction

Data on the following were extracted: study design, level of evidence, participant characteristics, details of the intervention/comparator, outcome measures and results relating to apraxia and occupational performance.

Risk of bias and quality assessment

The Cochrane collaboration risk of bias tool was used (Higgins et al., 2016). Two independent reviewers (ST and RG) independently assessed each study’s risk of bias using defined domains: sequence generation; allocation concealment; blinding of participants, personnel and outcome assessors; completeness of outcome data; selective reporting; other bias and the overall risk of bias. Disagreements were resolved by consensus.

Statistical analysis

Data were extracted separately for the meta-analysis (see Table 1). For each study, outcome data were extracted at baseline, post-intervention and follow-up time points. For our primary outcome of occupational performance, data were extracted on the mean and standard deviation (SD) values, across the intervention and control groups as well as the number in each group at baseline. Pooled mean differences (MD) and 95% confidence intervals (CI) were calculated to determine the treatment effect. In instances where studies reported change values, the changes from baseline were pulled across the intervention and control groups. As per the Deeks et al. (2022) on pooling data, we pooled absolute scores post-intervention and changed scores from baseline to post-intervention (Higgins et al., 2022). Standard methods were used to convert CIs to SDs (Higgins et al., 2022). In instances where the 20-item Barthel Index was reported, this value was multiplied by 5 to ascertain a value out of 100 (Wade and Collin, 1988).

Table 1.

Intervention outcomes.

Study	Measures	Baseline	Post-treatment	Follow-up	Univariate analysis of variance across three time points
Aguilar-Ferrándiz et al. 2021	ADL		Week 8	Week 20
	Barthel index	68.42 ± 27.74	74.21 ± 26.10	73.42 ± 27	F = 0.143, p = 0.867
		67.37 ± 25.95	69.72 ± 25.29	71.39 ± 23.75
	Lawton and Brody IADL scale	3.53 ± 3.26	4.16 ± 3.48	4.21 ± 3.54	F = 0.092, p = 0.912
		2.58 ± 2.04	2.67 ± 1.88	2.78 ± 1.99
	ADL observation and scoring	18.26 ± 14.69	14.84 ± 14.51	15.05 ± 14.86	F = 0.039, p = 0.962
		18.37 ± 13.32	16.56 ± 12.12	16.67 ± 12.09
	Apraxia
	De Renzi tests for ideational and ideomotor	27.00 ± 3.15	31.00 ± 2.75	33.00 ± 2.75	F = 5.212, p = 0.007
		28.00 ± 3.25	27.50 ± 3.71	27.50 ± 3.62
	De Renzi imitating gestures test	50.37 ± 8.71	63.05 ± 8.08	64.16 ± 8.28	F = 11.256, p = 0.001
		55.95 ± 7.40	53.39 ± 7.55	53.72 ± 8.99
	The recognition of gestures test	7.74 ± 1.28	9.11 ± 1.24	8.84 ± 1.50	F = 3.852, p = 0.024
		8.00 ± 1.60	7.78 ± 1.52	7.56 ± 1.54
	TULIA	188.53 ± 30.04	219.53 ± 24.06	224.26 ± 18.89	F = 3.583, p = 0.031
		206.84 ± 23.67	208.67 ± 19.12	205.61 ± 30.76
Donkervoort et al. (2001)	ADL		week 8	week 20
	Barthel Index	n = 53, 10.7 (4.9) vs n = 53, 11.2 (5.0)	n = 45, 2.44 vs n = 48, 1.15 p < 0.001	n = 40, 3.00 vs n = 48, 2.83 p = 0.11
	ADL observations	n = 51, 2.2 (0.5) vs n = 50, 2.3 (0.4)	n = 43, 0.24 vs n = 39, 0.12 p = 0.03	n = 39, 0.21 vs n = 36, 0.22 p = 0.37
	ADL judgements OT	n = 53, 3.2 (1.3) vs n = 53, 3.1 (1.4)	n = 47, 0.90 vs n = 48, 1.04 p = 0.48	n = 37, 1.20 vs n = 35, 1.48 p = 0.42
	Participant	n = 43, 4.1 (1.5) vs n = 51, 4.4 (1.6)	n = 38, 0.65 vs n = 44, 0.54 p = 0.25	n = 37, 0.72 vs n = 39, 0.56 p = 0.27
	Apraxia
	De Renzi Apraxia	n = 54, 57.3 (21.2) vs n = 54, 62.0 (17.9)	n = 44, 4.20 vs n = 48, 2.21 p = 0.25	n = 42, 6.26 vs n = 41, 4.02 p = 0.16
Geusgens et al. (2006)	ADL ADL observations The Barthel Index	10.7 (4.9) vs 11.2 (5.0)	week 8 Non-trained Strategy training: n = 41, 0.307 (0.45) Usual Care: n = 40, 0.143 (0.37) p-value: 0.04	week 20 Non-trained Strategy training: n = 38, 0.271 (0.54) Usual Care: n = 36, 0.241 (0.48) p-value: 0.40
	Apraxia The Apraxia Test, based on a test by De Renzi.	57.3 (21.2) vs 62.0 (17.9)	Trained Strategy training: n = 32, 0.208 (0.53) Usual care: n = 29, 0.108 (0.46). p-value: 0.22	Trained1 Strategy training: n = 30, 0.242 (0.54). Usual care: n = 27, 0.218 (0.53). p-value: 0.43
Smania et al. (2006)	ADLs Non-standardised questionnaire completed by caregiver (0–80).		ADL Intervention p < 0.001 Control NS	ADL Intervention NS
	Apraxia Non-standardised praxis function test.	Ideational apraxia 9.17 (4.16) vs 11.07 (3.37) Ideomotor apraxia 27.67 (14.68) vs 34.93 (12.19) Gesture comprehension 6.22 (2.29) vs 7.07 (2.09)	Ideational apraxia Intervention p < 0.01 Ideomotor apraxia Intervention p < 0.01 Control p = 0.08 Gesture comprehension Intervention p < 0.01

Meta-analysis was performed using Review Manager 5 (RevMan5) (computer programme). Version 5.4 Copenhagen: The cochrane Collaboration, 2020. In terms of dealing with heterogeneity across studies, forest plots were visually examined alongside reporting of the I² statistic. In pooled analyses where the I² is less than 50%, a fixed-effects model was reported. No sensitivity analyses were conducted. Where statistical pooling is not possible, the findings are presented in table and narrative form.

Results

The results are presented in the PRISMA flowchart (Figure 1). Of the 11,321 studies screened, four studies comprising data from three original RCTs are included in the synthesis. Geusgens et al. (2006) completed a secondary analysis of data from an original Dutch trial completed by Donkervoort et al. (2001). In their secondary analysis, Geusgens et al. (2006) explored whether cognitive strategy training for stroke patients with upper limb apraxia transferred from trained to non-trained tasks. The two further original trials were conducted in Spain (Aguilar-Ferrándiz et al., 2021) and Italy (Smania et al., 2006). Table 2 presents the descriptive characteristics of the included studies.

Table 2.

Descriptive characteristics of included studies.

Author/year	Level of evidence Study design Risk of bias	Participants Inclusion Criteria study setting	Intervention and control groups	Outcome measures	Results
Aguilar-Ferrándiz et al. (2021) https://doi.org/10.1016/j.apmr.2020.12.015	Level 1B RCT Risk of bias High	Participants N = 38, (M age, 74.8; 50% female) M mo post-stroke, 12.23 Inclusion criteria Community-dwelling adults (aged 25–95) with mild–moderate unilateral stroke effects and ULA lasting 2 months. Intervention setting Participants own homes in Spain	Intervention 1: (N = 19) OT delivered a combined functional rehabilitation programme for 3 sessions/week @ 30 min over 8 weeks. 2 sessions/week focused on a restorative approach (gesture training). 1 session focused on strategy training (teaching the patient internal/external compensatory approaches to assist ADL performance) Intervention 2: (N = 19) An OT delivered monthly education workshop for patients and caregivers focused on ULA.	Completed at baseline, post-treatment (8/week) and follow-up (20/week) N = 37 ADLs: • Barthel Index • Lawton and Brody IADL Scale • Van Heugten ADL observations Apraxia: • De Renzi tests for ideational and ideomotor Apraxia • De Renzi imitating gestures test, • The recognition of gestures test • TULIA	Significant findings Significant differences in favour of the experimental group were noted in the apraxia tests relating to ideomotor apraxia, imitating gestures, recognition of gestures and comprehension of gesture production. Non-significant findings The intervention did not significantly impact occupational performance or quality of life.
Donkervoort et al. (2001) https://doi.org/10.1080/09602010143000093	Level 1B RCT Risk of bias High	Participants* N = 113, (M age, 65; female 46%) M mo post-stroke, 3.2 Inclusion criteria Adults (25–95 years) with left hemisphere CVA and ULA. Setting Rehabilitation centres and nursing homes, Netherlands	Intervention 1: (N = 56) Strategy training (teaching the patient-internal/external compensatory approaches to assist ADL performance) was integrated into usual occupational therapy for 8 weeks. Average of 15 h of therapy over 25 sessions Intervention 2: (N = 63) usual Occupational therapy. Average of 19 h of therapy over 27 sessions.	Completed at baseline, post-treatment (8/week) and follow-up (5 mo) N = 86) ADLs: • Barthel Index • Van Heugten ADL Observations • ADL judgement by 1. Patient 2. Therapist Apraxia: • De Renzi tests for ideational and ideomotor Apraxia	Significant findings Therapeutic intervention produces a small improvement on the Barthel immediately after intervention. The study did not find evidence of a lasting effect on functional performance after 5 months (control group still receiving OT). Non-significant findings consistent with the aim to reduce disability not impairments – no improvement in apraxia
Geusgens et al. (2006) https://doi.org/10.1080/09602010500172350	Level 1B RCT Risk of bias High	Participants* N = 113, (M age, 65; female 46%) M mo post-stroke, 3.2 Inclusion criteria Adults (25–95 years) with left hemisphere CVA and ULA. Study setting Rehabilitation centres and nursing homes, Netherlands	Intervention 1: (N = 56) Strategy training (teaching the patient-internal/external compensatory approaches to assist ADL performance) was integrated into usual occupational therapy for 8 weeks. Average of 15 h of therapy over 25 sessions Intervention 2: (N = 63) usual Occupational therapy. Average of 19 h of therapy over 27 sessions	ADLs • Barthel Index • Van Heugten ADL Observations Apraxia • De Renzi tests for ideational and ideomotor Apraxia	Significant findings In both groups, the scores on the ADL observations for the non-trained tasks improved significantly (p < .00). Change scores of non-trained tasks were larger in the intervention group than in the control. Non-significant findings Patients scored lower on the non-trained tasks compared with the score on trained tasks.
Smania et al. (2006) https://doi.org/10.1212/01.wnl.0000247279.63483.1f	Level 1B RCT Risk of bias High	Participants N = 33 (M age, 65; 30% Female) M mo post-stroke, 13.5 Inclusion criteria Left hemisphere CVA and ULA Study setting Neuropsychological rehabilitation unit, Italy	Intervention 1: (N = 18) Gesture training 3 × 50 min session × 3 times/week. Intervention 2: (N = 15) Conventional aphasia therapy	Baseline, post-treatment and follow-up (2 months; N = 17) ADLs • Caregiver ADL questionnaire Apraxia • De Renzi tests for ideational and ideomotor Apraxia • The Recognition of Gesture Test	Significant findings Improvements in apraxic tasks and reported on ADL questionnaire; positive treatment effect after 2 months Non-significant findings none

ADLs: activities of daily living; CVA: cerebral vascular accident; IADL: instrumental activities of daily living; M: mean; Mo: months; OT: occupational therapist; RCT: randomised controlled trial; TULIA: Test of Upper Limb apraxia; ULA: upper limb apraxia.

^*Same sample participants for Donkervoort et al. (2001) and Geusgens et al. (2006).

A total of 184 participants with upper limb apraxia after stroke were included in the studies. All studies had mixed-gender samples. Mean time post-stroke varied from 3.2 months (Donkervoort et al., 2001) to 13.5 months (Smania et al., 2006). The participants were recruited from rehabilitation units (Donkervoort et al., 2001; Smania et al., 2006), homes (Aguilar-Ferrándiz et al., 2021) and nursing homes (Donkervoort et al., 2001).

Methodological quality of included studies

The methodological quality of the included studies was poor overall. Issues around selection bias (randomisation and allocation concealment) were addressed in two of the three trials (Aguilar-Ferrándiz et al., 2021; Donkervoort et al., 2001). Detection bias was also minimised across all three studies. Performance bias was unclear/evident across all studies. All studies included in this review experienced a high risk of bias (see Table 3).

Table 3.

Risk of bias table.

Author (year)	Selection bias			Performance bias		Detection bias		Attrition bias	Reporting bias	Overall risk of-bias assessment
	Random sequence generation	Allocation concealment	Baseline differences between intervention groups	Blinding of participants during the trial	Blinding of study personnel during the trial	Blinding of outcome assessment: self-reported outcomes	Blinding of outcome assessment: objective outcomes	Incomplete outcome data	Selective reporting
Aguilar-Ferrándiz et al (2021)	+	+	−	−	−	−	+	+	+	H
Donkervoort et al (2001)	+	+	−	?	−	?	+	+	+	H
Geusgens et al (2006)	+	+	−	?	−	?	+	+	+	H
Smania et al. (2006)	?	?	?	?	?	?	+	−	?	H

Categories for risk of bias are as follows: low risk of bias (+), unclear risk of bias (?), high risk of bias (–). Scoring for overall risk of bias assessment is as follows: 0–3 min, low risk of bias (L); 4–6 min, moderate risk of bias (M); 7–9 min, high risk of bias (H).

Source: Higgins et al. (2016).

Outcome measures

A range of standardised and non-standardised assessment tools was used throughout the studies to diagnose apraxia and measure changes in occupational performance, detailed in Table 1. Donkervoort et al.’s (2001) study is the only study from the Cochrane review (West et al., 2008) included in this review secondary to outcome measures utilised.

Apraxia measures

Two studies (Aguilar-Ferrándiz et al., 2021; Donkervoort et al., 2001) included in the meta-analysis used standardised apraxia tests, based on De Renzi et al. (1968), De Renzi (1980), which assesses ideational and ideomotor apraxia. In addition, Aguilar-Ferrandiz et al. (2021) completed a Recognition of Gestures Test (Smania et al., 2000) and the Test of Upper Limb Apraxia (Vanbellingen et al., 2010). Smania et al. (2006) used a non-standardised limb praxis evaluation focusing on ideational, ideomotor, gesture and constructional abilities.

Occupational performance outcome measures

Two studies (Aguilar-Ferrándiz et al., 2021; Donkervoort et al., 2001) used a set of standardised ADL observations (Van Heugten et al., 1999; Van Heugten et al., 2000b). Performance in ADLs is scored on four aspects: independence, initiation, execution and control. Four levels of ability are recorded for each aspect (e.g. help required to complete the task as independently as possible). Donkervoort et al. (2001) study reports a total mean score for this assessment ranging from 0 = totally dependent to 3 = fully independent. Aguilar-Ferrándiz et al. (2021) report a total score ranging from 0 = total independence to 48 = total dependence. The Barthel ADL index (Wade and Collins 1988) is also utilised in both studies. In addition, Aguilar-Ferrandiz et al. (2021) used the Lawton and Brody Instrumental ADL (Lawton and Brody, 1969) which is a self-administered questionnaire used to assess the instrumental activities necessary for living independently. Smania et al. (2006) used a novel ADL questionnaire completed by the caregiver, while Donkervoort et al. (2001) also availed of ADL questionnaires based on the Rivermead ADL scale (Lincoln and Edmans, 1990) which was completed by both the occupational therapist and the patient. Donkervoort et al.’s (2001) study is the only study from the Cochrane review (West et al., 2008) included in this review secondary to outcome measures utilised.

Interventions

Interventions included strategy training (Donkervoort et al., 2001), gesture training (Smania et al., 2006), and a combined functional rehabilitation approach of gesture and strategy training (Aguilar-Ferrándiz et al., 2021).

The strategy training method is tailored to each individual and focuses on the training of compensatory techniques during the performance of ADLs as per published guidelines (Stehmann-Saris et al., 2005; Van Heugten et al., 2000a). It entails teaching the use of internal compensatory strategies (i.e. self-verbalisation) and external compensatory strategies (i.e. use of pictures, instructions) to maximise independence in selected everyday activities (Aguilar-Ferrándiz et al., 2021; Donkervoort et al., 2001).

Gesture training focuses on the training of transitive, intransitive symbolic and intransitive non-symbolic gestures (Smania et al., 2000). The treatment for transitive and intransitive gesture training is designed in three phases, with the difficulty level increasing. For example, in phase A of transitive gesture training, the patient is required to demonstrate the use of common tools. In phase B, the patient is shown a picture illustrating a transitive gesture and is then required to imitate it. In phase C, the patient is presented with a picture showing a common tool and is then asked to demonstrate how the tool is used. Each phase contains 20 items. When the patient is able to correctly perform at least 17 of the 20 items in a phase, the next phase is initiated. The intervention ends when the patients adequately can perform the three intervention sections or when 35 sessions are reached. The training for intransitive non-symbolic gestures involves meaningless gestures presented by the therapist. In all, 12 gestures involving six proximal and six distal joints are delivered, both static and dynamic in nature. The patient is required to imitate. The therapist provides facilitation to allow for correct achievement (Pérez-Mármol et al., 2015; Smania et al., 2000).

Intervention effects

Changes in occupational performance

Table 1 presents the quantitative findings across the studies. Two trials (Aguilar-Ferrandiz et al., 2021; Donkervoort et al., 2001) examined occupational performance using the Barthel Index post-intervention and provided data suitable for meta-analysis. Interventions included strategy training (Donkervoort et al., 2001) and a combined functional rehabilitation approach of gesture and strategy training (Aguilar-Ferrandiz et al., 2021). There was a significant improvement in favour of the intervention group at 8 weeks (Finite Element Method (FEM), MD: 6.24, 95% CI: 0.92–11.57, I² = 0%) but this was not maintained in the follow-up assessment at 20 weeks (FEM, MD: 1.08, 95% CI: −6.01–8.16, I² = 0%). See Figure 2. These findings correspond with those of Smania et al. (2006) where significant improvements were noted in ADL post-intervention for patients who received gesture training (ADL measure p < 0.001) but this difference did not persist at follow-up.

Figure 2.

Occupational performance of the intervention and control groups at (a) 8 weeks and (b) 20 weeks. (a) Functional status as measured by the Barthel Index at 8 weeks. (b) Functional status as measured by the Barthel Index at 20 weeks.

Changes in apraxia

The De Renzi tests for ideational and ideomotor apraxia were reported across all three studies and two studies (Aguilar-Ferrandiz et al., 2021; Donkervoort et al., 2001) provided data suitable for meta-analysis. There was a significant improvement in apraxia in favour of the intervention group at 8 weeks (FEM, MD: 3.21, 95% CI: −1.34–5.07, I² = 0%) and this significant difference was still evident at 20 weeks (FEM, MD: 5.12, 95% CI: 3.20–7.24, I² = 12%). See Figure 3. Smania et al. (2006) also noted significant improvements in apraxia post-intervention for patients who received gesture training (ideational apraxia p < 0.01; ideomotor apraxia p < 0.01) but not at the 2-month follow-up.

Figure 3.

Levels of apraxia among the intervention and control groups at (a) 8 weeks and (b) 20 weeks. (a) Apraxia as measured by the De Renzi tests for ideational and ideomotor apraxia at 8 weeks. (b) Apraxia as measured by the De Renzi tests for ideational and ideomotor apraxia at 20 weeks.

Discussion

This systematic review examined the impact of rehabilitation interventions for upper limb apraxia post-stroke. Interventions across the four included studies are as follows: (1) strategy training, (2) gesture training and (3) a combination functional rehabilitation programme (strategy and gesture training).

Strategy training alone, or in combination with gesture training as part of a combined functional rehabilitation approach, was found to be significantly more effective than control interventions at improving function and apraxia scores after intervention. Improvements in apraxia scores were maintained at follow-up, whereas improvements in occupational performance were not. A single study of gesture training, not included in the meta-analysis, reported significant improvements in apraxia and function after intervention but not at follow-up. However, the methodological quality of the included studies was poor, limiting the external validity of the findings.

In the included studies, interventions were delivered by occupational therapists in two (Aguilar-Ferrándiz et al., 2021; Donkervoort et al., 2001) of the three trials. It is not clear who delivered the intervention in the trial of gesture training by Smania et al. (2006). All studies included in this review revealed a high risk of bias, which is often the nature of rehabilitation studies.

Strategy training, outlined in detail in the Dutch Occupational Therapy Guidelines for the assessment and treatment of apraxia following left hemisphere stroke (2005), aligns well with the philosophy of occupational therapy and is feasible to incorporate into everyday practice in hospital or community settings. The aim is to ‘restore impaired occupational performance by making use of the training of strategies to become more independent in the performance of. . .meaningful tasks’ (Stehmann-Saris et al., 2005: 24). The intervention is tailored to the particular problems estimated during the standardised ADL observations. Through this observation spatial errors, object/content errors and/or temporal errors at the orientation, execution and control phases of activities are identified. Intervention may include instruction, support and feedback to enable people to achieve the highest level of independence. Successful outcomes are dependent on intact metacognitive processes (Stehmann-Saris et al., 2005).

By contrast, the restorative approach of gesture training is somewhat decontextualised and potentially less meaningful for clients. Therefore, it could be considered less well aligned with occupational therapy as it involves repetitive use of picture cards and subsequent practice of imitation and initiation of both meaningful and meaningless gestures with and without tools (Smania et al., 2006).

Across the included studies, there was little attention to the satisfaction or experience of people post-stroke in relation to the interventions. Qualitative research with people after stroke, healthcare providers and carers has emphasised the importance of ensuring that activities used during cognitive rehabilitation are meaningful to patients and tailored to their individual abilities and goals (Merriman et al., 2019). Strategy training as described by Van Heugten and Geusgens (2009) involves the individual choosing meaningful activities to perform for the training sessions. Future research should ensure patient’s perspectives of interventions are evaluated.

The majority of participants in the included studies appear to have intact global cognitive abilities outside the apraxia domain. Aguilar-Ferrándiz et al. (2021) excluded those with cognitive impairment as per the mini-mental state examination. Similarly, all participants in Donkervoort et al. (2001) scored above the published cut-off score on the Cognitive Screening Test. No global cognitive screen was reported by Smania et al. (2006). This limits the generalisation of these findings to the broader stroke population who are very likely to experience other significant cognitive and language impairments coinciding with limb apraxia which may limit engagement with the outlined treatments.

Across the included studies, the time since stroke ranged from 2 to 36 months. Clinical guidelines support the initiation of therapy as soon as possible post-stroke (Stroke Foundation, 2021). Future research on apraxia rehabilitation interventions should consider the impact of the timing of intervention after stroke given the known heightened sensitivity to rehabilitative experience that the post-stroke brain displays around 5 days after ischaemic stroke (Biernaskie, 2004).

Despite the prevalence of limb apraxia and the known impact of apraxia on function only one trial conducted in the past decade was identified. There is clearly a need for more research in this area and previous systematic reviews (Alashram et al., 2021; Lindsten-McQueen et al., 2014; Worthington, 2016) echo the need for further research.

Recent studies excluded from this review due to study design included a study of Naturalistic Action Therapy that trains object selection and application with an errorless learning approach (Buchmann et al., 2019), an evaluation of occupational therapy delivered combined training with physical practice followed by mental practice (e.g. reaching for a cup with a person-specific graded approach followed by mental practice of the same task using an audiotape) (Wu et al., 2011) and an action semantics intervention which targets the action knowledge deficit observed in many individuals with apraxia by strengthening the associations between tools, their related actions and other information in a putative semantic network (Stoll et al., 2020). These studies may indicate future directions in the field of post-stroke limb apraxia rehabilitation. Of concern, it is not clear who delivered the intervention in the studies by Buchmann et al. (2019) and Stoll et al. (2020). Reporting guidelines for intervention studies specify that the expertise, background and training of those who provided/delivered the intervention should be reported (Hoffman et al., 2014). One of the studies included in this review (Smania et al., 2006) also did not report who delivered the intervention. Future intervention research in this area should adhere to intervention reporting guidelines to aid the replication of studies and implementation of findings in practice.

All interventions in this review were delivered over an 8-week period with some variation in the intensity of therapy. Given the robust evidence base for stroke rehabilitation in chronic stroke (Teasell et al., 2012) and the lack of sustained improvement in occupational performance at follow-up found in this review future research could evaluate the impact of intervention delivered over a longer time period with tapering rather than an intensive block of intervention.

Occupational therapy literature often describes how to distinguish ideomotor and ideational apraxia (Jackson and Wolff, 2010). Across the included studies, treatment approaches were not tailored according to these subtypes of limb apraxia (i.e. ideomotor or ideational limb apraxia) and findings provide some support for therapists adopting a generic approach to the rehabilitation of limb apraxia regardless of sub-type.

Implications for occupational therapy practice

Findings from this review provide low-quality evidence to support the use of strategy training alone or in combination with gesture training by occupational therapists for the rehabilitation of upper limb apraxia post-stroke. It must be noted that our review did not find evidence that changes in occupational performance were sustained at longer-term follow-up (20 weeks). In line with their professional philosophy, occupational therapists should consider the meaningfulness to clients of activities used in cognitive rehabilitation.

Conclusion

This systematic review and meta-analysis examined the impact of rehabilitation interventions for upper limb apraxia post-stroke. Interventions identified included strategy training, gesture training and a combination functional rehabilitation programme (both strategy and gesture training). Strategy training alone, or in combination with gesture training, significantly improved apraxia scores and occupational performance post-intervention, when compared to control groups. Improvements in apraxia scores were maintained at follow-up, whereas improvements in occupational performance were not. This review provides low-quality evidence to support the use of strategy training alone, or in combination with gesture training, by Occupational Therapists to improve occupational performance and apraxia scores post-intervention among patients with post-stroke limb apraxia. Results must be interpreted cautiously, as there was a high risk of bias identified across all studies reviewed. The findings from this review add to an under-researched field of study, and further research is needed in this area. Future research should evaluate the impact of limb apraxia interventions delivered over a longer time period with tapering of the intervention to explore whether changes in occupational performance can be sustained longer term.

Key findings

This review provides low-quality evidence that strategy training alone, or in combination with gesture training is significantly more effective in improving occupational performance and apraxia scores in the short term (post-intervention) in people with upper limb apraxia post-stroke than control interventions.

What this study has added

This study critically reviews available literature regarding the rehabilitation of upper limb apraxia post-stroke, supporting the use of strategy training alone or in combination with gesture training for short-term improvement in occupational performance by occupational therapists.

Appendix 1.

Search terms.

Main concepts	Terms used*
Stroke	Stroke or cerebrovascular accident or CVA or post-stroke
Apraxia	Apraxia or apraxias or apraxic or psychomotor disorders or psychomotor performance or Motor skills or task performance or task analysis or cognitive impairment or cognitive disorders or praxis or dyspraxia or dyspraxis or Disorders of performance or action disorganisation syndrome
Intervention	Intervention or interventions or strategies or best practice or best practices or treatment or therapy or programme or management or care or rehabilitation

^*Search terms utilised were developed based on Cochrane review (West et al., 2008).