Study Design and Patients

A systematic review of 22 randomized controlled trials (RCTs) compared the communication abilities of 1620 people, some of whom received SLT while others were randomly allocated to receive no SLT (Brady et al., 2016). Five trials randomized people across two SLT interventions and a no treatment group and so the evidence summary is based on 27 paired ‘randomized comparisons’. Participants’ age ranged from 28 to 94 years, they were between 2 days to 29 years post-stroke onset, and their aphasia ranged from mild to severe.

Interventions

Among the SLT interventions contributing to this evidence include groups that received ‘conventional’ SLT, computer-mediated SLT, group SLT, functional SLT, intensive SLT, language enrichment therapy, constraint-inducted aphasia therapy, melodic intonation therapy (MIT), independent training, and volunteer-facilitated SLT.

Functional Communication

Based on pooled data from 10 trials, patients who received SLT performed better on measures of functional language abilities compared with those who did not receive SLT (standardized mean difference [SMD] 0.28, 95% confidence interval [CI]: 0.06–0.49; P = 0.01) (Brady et al., 2016). At 6 months follow-up (based on a small amount of pooled data from 111 participants in two trials) there was no longer any evidence of this effect. In contrast, the RATs-3 trial (n = 152; completed after the Cochrane review was published) found no benefit on measures of functional communication measures for patients within 2 weeks of aphasia onset following 1 hour daily cognitive-linguistic SLT over 4 weeks compared with no therapy access (Nouwens et al., 2017).

Receptive Language

Eleven trials measured participants’ receptive language (Brady et al., 2016). Pooled statistical data from 6 trials (9 randomized comparisons), comparing the auditory comprehension of participants who received SLT to those who did not, showed no evidence of a difference between the groups (SMD 0.06, 95% CI: –0.15–0.26; P = 0.42). Five trials compared the groups’ reading comprehension. Those participants who received SLT performed better on tests of reading comprehension than those who did not receive SLT (SMD 0.31, 95% CI: 0.03–0.59; P = 0.03). In one trial the participants receiving SLT also received an acupuncture co-intervention.

Expressive Language

Seven randomized comparisons evaluated the naming abilities of participants who received SLT compared with those who had not received SLT, but there was no evidence of a difference between the groups (Brady et al., 2016). In contrast, eight randomized comparisons compared participants’ writing abilities and found that the participants who had received SLT performed better on measures of writing abilities than those who had received no SLT (SMD 0.41, 95% CI: 0.14–0.67; P = 0.0003). There was some evidence of statistical heterogeneity in this meta-analysis (heterogeneity: chi² = 11.15, degrees of freedom [df] = 7 [P = 0.13]; I² = 37%). This was no longer evident when the data from two Chinese randomized comparisons were removed from the meta-analysis (I² = 0%), but the overall effect remained (SMD 0.27, 95% CI: 0.03–0.56; P = 0.08). Other measures of expressive language, such as copying, repetition, and fluency, which are less relevant to the functional use of language, were reported by a small number of trials, and did not show any indication of a difference between the groups. At 6 months follow-up there was no evidence of a difference in a group’s expressive language naming skills based on the pooled data from 3 small trials (Brady et al., 2016).

Severity of Language Impairment

Seven trials compared a group that received SLT with one that did not by measuring the severity of the participants’ aphasia impairment using a range of multilingual aphasia test batteries. Based on the pooled data available, there was no evidence of a significant difference between the groups (SMD 0.55, 95% CI: –0.14–1.25; P = 0.12).

Evidence Overview

People with aphasia who received SLT had higher scores on measures of functional communication, receptive language (reading comprehension), and expressive language (general) than people who had no SLT. Significant differences were not evident across all measures nor at follow-up. Sample sizes were small. For ethical reasons, random allocation of patients within the early stages of stroke to groups that have no access to SLT are unlikely in the future, with more recent trials of SLT making comparisons to participants who had access to deferred therapy (e.g. the FCET2EC trial, Breitenstein et al., 2017) or an alternative approach to therapy (the VERSE 3 trial, Godecke et al., 2018).

Study Design and Patients

Nine RCTs involving 447 people with aphasia compared SLT interventions (conventional, group, or telerehabilitation interventions) with interventions providing social support (Brady et al., 2016). Aphasia onset ranged from 12 days (an average of those participating in one intervention trial [Bowen et al., 2012]) up to 28 years. Four trials recruited participants within 4 weeks of stroke onset. The remainder of the trials recruited after this acute phase. Description of the participants varied. Their ages ranged from 18 to 97 years. The people who received SLT in one trial (David et al., 1982) were significantly older (mean age [± SD] 70 [± 8.7]) years) compared to those who received social support (65 [± 10.6] years). All trials reported the severity of participants’ aphasia, which ranged from mild to severe.

Interventions

Providing people with social support stimulates conversation and augments functional language use. Interventions in this category included participation in art, dance, or music classes or other non-language-orientated therapeutic interventions (e.g. art, physical or music therapy). Other social interventions typically included providing regular conversational stimulation or alternative informal, unstructured communicative interactions. These interventions did not include targeted ‘therapy’ designed to resolve expressive or receptive language impairments. In some cases, such interventions were included in a trial of SLT as an attention control. However, providing regular (additional) language stimulation is likely to benefit functional language use. Conversational stimulation could be easily implemented and supported by family members or volunteers and is an important adjunct to formal SLT (Brady et al., 2018). Hence, we have presented these data separately.

Social support considered within one Cochrane review was provided by volunteer visitors, nurses, a psychologist, SLT students, research assistants, or in a group manner through external movement classes, arts groups, or church or support groups (Brady et al., 2016). Providers were given training and (in three trials) a manual of permitted activities. Interventions were provided for between 1 and 3 hours weekly for between 1 and 12 months. Most were face-to-face support, although two used an internet-supported videoconferencing tool.

Functional Communication

Using a range of tools, five trials (involving 247 people) found no evidence of a difference in participants’ functional communication based on whether they had access to SLT or social support activities.

Receptive, Expressive, and Severity of Language Impairment

Few data were available to inform our understanding of the benefits of SLT versus social support on participants’ receptive language. Pooled summary data from two small trials (involving 23 people) found no evidence of a difference between the groups’ receptive language skills. Similarly, data based on three small randomized comparisons (n = 33) showed no evidence of a difference in naming skills between the groups. At individual trial level, significant effects were observed favouring both social support and SLT. Other aspects of expressive language such as sentences, picture description, writing, and fluency were informed by even smaller trials (often in isolation), limiting confidence in their findings (Brady et al., 2016).

Adherence to Interventions

A total of 40 people stopped participating in SLT in the trials considered within this comparison compared to 65 participants lost to the social support groups (odds ratio [OR] 0.51, 95% CI: 0.32–0.81; P = 0.005), confounding the findings described above. We identified more participants (n = 45) who did not adhere to their social support intervention than those who failed to adhere to their allocated SLT (n = 11) (OR 0.18, 95% CI: 0.09 to 0.37; P < 0.00001). Thus, social support interventions (or the rationale in their use) were less acceptable for some people with aphasia than SLT interventions, resulting in systematic difference in the groups’ adherence rates (Brady et al., 2018).

Evidence: Overview

Seven trials compared 447 people who received SLT with groups who received social support and stimulation. Differences in language measures observed (favouring the group that received social support over those that received SLT) were derived mainly from one small trial and confounded by significantly higher losses and non-adherence rates in the social support groups (Brady et al., 2018). In contrast, a large rigorously conducted trial found no evidence of a significant difference (Bowen et al., 2012). Additional research should confirm whether social support and stimulation provide benefits to some aspects of language skills and why more people fail to adhere to this intervention. In the meantime, rehabilitation guidelines recommend social stimulation for people with aphasia in the form of conversational support from people with training in supporting impaired language skills and referral to support groups alongside formal SLT interventions (Intercollegiate Stroke Working Party, 2016).

Study Design and Patients

A Cochrane systematic review of eight RCTs (involving 263 people with aphasia) compared the delivery of high-intensity SLT interventions to a lower intensity of intervention (Brady et al., 2016).

Interventions

Higher-intensity interventions were quantified by number of hours of direct therapy per week (ranging from 4 to 15 hours). These high-intensity therapies were compared with therapy delivered at a lower intensity (ranging from 1.5 to 5 hours weekly). Home-based practice activities or tasks were not included in this calculation of intensity (Brady et al., 2016).

Functional Communication

Two trials compared high- (7.5 to 10 hours therapy weekly) to low- (1.5 to 2 hours therapy weekly) intensity interventions by measuring participants’ performance on the Functional Communication Profile (Sarno, 1969). Those who received the higher-intensity SLT (n = 45) had better functional communication than those who received SLT at a lower intensity (n = 39) (mean difference [MD] 11.75, 95% CI: 4.09 to 19.40; P = 0.003) (Brady et al., 2016). These benefits were still evident on follow-up at 40 weeks. The FCET2EC trial randomly allocated people (n = 158, aged 70 years or younger) with chronic aphasia (of at least 6 months in duration) to receive 3 weeks of intensive SLT of at least 10 hours weekly versus deferred intensive therapy (but continued usual SLT access). Measuring functional communication, the intensive SLT group significantly improved from baseline compared to the usual care group who did not (Breitenstein et al., 2017).

Receptive Language

Data from two trials compared high- (10 hours weekly) to low-intensity (2 to 5 hours weekly) SLT regimens and, where possible, data were pooled. Performance on measures of auditory comprehension and reading did not differ between the groups.

Expressive Language

There was no evidence of a difference in expressive language skills (measured using naming, repetition, fluency, and writing subtests) between the groups across three trials that had access to high- (10 hours weekly) or low-intensity (2 to 5 hours weekly) SLT. The recent Big CACTUS trial found benefit in measures of word finding among a chronic group of patients (more than 4 months after aphasia onset) but not functional communication when they had access to self-managed, computer-based therapy software (up to 30 minutes’ practice daily over 6 months) compared to usual care (which averaged approximately 1 hour every 2 weeks) (Palmer et al., 2015).

Severity of Language Impairment

Based on the meta-analysis of five RCTs, people who received high-intensity SLT (n = 96; ranging from 5 to 10 hours weekly) experienced a lower severity of aphasia at the end of treatment period compared to people who received low-intensity SLT (n = 91; 1.5 to 5 hours) (SMD 0.38, 95% CI: 0.07–0.69; P = 0.02). One recently completed small trial (n = 30; Stahl et al., 2017) found that intensity of intervention may not be as important in the context of a chronic patient population. Among that trial’s patients who were at least 1 year after aphasia onset, overall language benefits were observed in the context of SLT but there was no additional benefit with the higher-intensity therapy (4 hours daily) compared to the moderate-intensity approach (2 hours daily). In contrast, the VERSE trial compared acute patients’ (n = 246, 12 weeks following index stroke) overall language severity following usual care and two experimental SLT approaches but found no evidence of a difference between the groups’ recovery on overall measures of aphasia severity (Godecke et al., 2018).

Adherence to Interventions

Information on therapy adherence was available for four trials. More people failed to complete the higher-intensity SLT interventions (n = 35/114; 4 to 10 hours SLT weekly) compared to those who received SLT at a lower intensity (n = 17/102; 1.5 to hours SLT weekly). Intensive SLT interventions may not be acceptable or feasible for all people with aphasia after stroke. The data contributing to measures of functional language and severity of impairment only reflect the participants who remained in the trials. Interestingly, all the individuals who dropped out of the intervention were recruited within a few months of stroke onset. Two trials that recruited participants who were 2 or more years after stroke onset did not report any dropouts. Importantly, the three trials that recruited people within 3 months of stroke onset found benefit of high-intensity SLT (measured by aphasia severity) for those participants who remained within the trials (SMD 0.47, CI: 95% 0.05–088). In contrast, the participants recruited several years after stroke did not demonstrate similar evidence of effect of high- compared to lower-intensity SLT (SMD 0.06 95%, CI: –0.67– +0.78).

Evidence: Overview

Functional language and aphasia severity improved following access to high-intensity SLT (compared to SLT at a lower intensity), but the findings were confounded by the greater number of participants dropping out from the therapy delivered at a higher intensity. There is some indication that timing of intensive SLT for aphasia after stroke may be important for tolerance of high-intensity interventions (and perhaps benefit), though numbers of trials and participants informing these meta-analyses were small.

Study Design: SLT Delivery Models

Few RCTs compare group SLT, 1-to-1 SLT, computer-facilitated, or volunteer-facilitated SLT to SLT delivered directly by a professional therapist.

Evidence Overview: SLT Delivery Models

There is no evidence of a difference in the functional communication skills (3 trials involving 43 people with aphasia) or severity of aphasia (4 trials involving 122 people with aphasia) of people who received SLT on a 1-to-1 versus group basis. A small number of trials based on small numbers of participants found no evidence of a difference in the provision of SLT interventions facilitated by volunteers or computers (under the direction of professional therapists, where volunteers received appropriate training and had access to relevant therapy materials and therapeutic intervention plans) compared to SLT delivered by a professional therapist. Clinical guidelines support the provision or augmentation of therapy provision by trained volunteers, family members, and computer programs (Intercollegiate Stroke Working Party, 2016).

Constraint-induced Aphasia Therapy

Constraint-induced aphasia therapy (CIAT) involves creating ‘constraints’ on the use of communication components that would naturally support verbal expression (e.g. visual feedback, gesture, or facial expressions). Constraints are created by placing a physical barrier or screen between speaker and listener and ‘forcing’ the use of specific verbal output to communicate meaning. CIAT is usually provided in a group format and at a high level of intensity. Five RCTs (174 people) compared CIAT with conventional 1-to-1 SLT, other group therapy, or a semantic approach to SLT. In most cases, the durations of therapies were matched. Only two trials controlled for the intensity, duration, and dose (total number of therapy hours provided) across the two arms of the trials (Sickert et al., 2014) (Wilssens et al., 2015). A third controlled for the duration and dose of therapy (Ciccone et al., 2016.). Meta-analyses of four trials (where data permitted) showed no evidence of a difference between CIAT and other theoretical approaches to SLT provision on measures of functional communication, receptive language, expressive language, severity of aphasia, or quality of life.

Melodic Intonation Therapy

Melodic intonation therapy (Albert et al., 1973) involves the repetitive singing of short phrases while tapping the rhythm of the phrase. MIT is generally thought optimal for the language rehabilitation of severe non-fluent aphasia. The rationale for this approach is based on melody activating regions of the right hemisphere of the brain that could support language use and the use of rhythm in language. The current evidence base for this approach is poor and based primarily on case studies and case series (Hurkmans et al., 2012; van der Meulen et al., 2012). One small RCT (n = 27) compared the use of MIT with conventional SLT approaches but found no evidence of a difference between the groups (van der Meulen et al., 2014).

Semantic-based SLT

Semantic approaches to the rehabilitation of people with aphasia are based on the cognitive neuropsychological model of language processing in the brain (Whitworth et al., 2005). Semantic therapy aims to improve processing at the level of word meaning, which will in turn lead to improved use and comprehension of language. Four small RCTs (n = 177) compared participants who received semantic SLT interventions to participants who received an alternative SLT (phonologically based, communicative SLT, a repetition in the presence of a picture approach, or CIAT approach). There was no evidence of a difference between the participants on any language measures.

Evidence Overview: Theoretical Approaches to SLT

Some trials addressing such comparisons have recently been reported or are ongoing. To date, comparisons are based on a small number of trials involving few participants (typically fewer than 20). No one theoretical approach has been demonstrated to be more effective than another. Similarly, national clinical guidelines do not as yet recommend one theoretical approach over another (unlike many other aspects of stroke rehabilitation). Additional large, rigorous, well-funded trials are required to further inform our understanding.

Background to Interventions

Memory impairments appear to be the most common form of cognitive deficit after stroke. Jokinen and colleagues (2015) found that 60% of patients had impairment of memory 3 months after stroke. The presence of memory impairment has been found to be specifically associated with dependence and functioning in everyday life (Tatemichi et al., 1994; Middleton et al., 2014). Memory is not a unitary system and specific forms of memory difficulty may occur (e.g. visual vs verbal impairment), but the major distinction is between working memory (holding and manipulating information in mind) and long-term memory (remembering information and events anywhere from a few minutes ago to many years ago). In addition, a distinction is drawn between forgetting what has happened (‘retrospective memory’ – e.g. what I did yesterday; what I was told this morning; how to find a place I visited last week) and forgetting to do things (‘prospective memory’ – e.g. forgetting to pay a bill, go to an appointment, pass on a message or take medication; das Nair and Lincoln, 2007). In terms of everyday functioning, while impairments in retrospective and prospective memory may both cause difficulties, deficits in prospective memory are likely to have the most serious consequences for everyday life (Evans, 2013).

Interventions

Reflecting the different types of memory impairment, interventions for memory deficits have taken several forms. Most work is in relation to long-term memory, but recently there has also been a focus on working memory, though the latter tends to be included with attention training because of the overlap between the constructs of attention and working memory. In relation to long-term memory, some interventions are aimed at enhancing initial learning of information so that it will be better retained. Other interventions are focused on improving the likelihood that a person will carry out an intended task. For enhancing learning, one intervention that has been investigated relatively extensively in people with acquired brain injury, albeit not extensively in people with stroke, is errorless learning. This refers to a general principle of attempting to avoid making errors while learning something new. The rationale is that if a person with a memory impairment makes an error while learning something new, they will find it difficult to remember that the response was an error, but also be more likely to make the same error again as a result of implicit memory processes (Baddeley and Wilson, 1994). Hence, errorless learning techniques aim to minimize the likelihood that a mistake is made during learning. There is a range of other mental strategies that are designed to improve the encoding of information into memory, with the intention that the information will be stored, retained, and retrieved more effectively (Evans, 2009). The other major approach to memory rehabilitation involves the use of external memory aids of some form. These include diaries, notebooks, wall charts, and electronic reminding devices such as pagers, personal digital assistants, and smart phones (Evans et al., 2003).

Study Designs

Although the number of studies of memory rehabilitation has increased somewhat over the last decade, there are still very few high-quality RCTs of memory rehabilitation interventions for stroke. In their original Cochrane review of CR for memory impairment in stroke, das Nair and Lincoln (2007) identified just two trials that met inclusion criteria. In the updated Cochrane review (das Nair et al., 2016), 13 studies were included, but most were graded as moderate in quality and reporting quality was judged to be poor in many studies. Sample sizes were typically small, and outcome measures varied considerably between studies. The current NICE guideline (National Institute for Health and Care Excellence [NICE], 2013) included two RCTs. Cicerone and colleagues’ (2011) series of three systematic reviews of CR (for TBI and stroke) considered 70 papers relating to memory interventions, but of these only 7 were considered to be well-designed prospective RCTs and most of the studies included mainly patients with TBI. Of the rest of the studies, 3 were quasi-randomized trials, 7 were cohort studies without randomization, and 53 were case series or studies using single subject methods. One might, in fact, argue that the class allocation for well-conducted single-case experimental design studies ought to be higher (Tate et al., 2014), but even then very few studies would be considered high-quality single-case experimental designs examining memory interventions in stroke.

Evidence and Overview

Conclusions with regard to which, if any, memory interventions are recommended vary between reviews and guideline documents. The Cochrane review (das Nair et al., 2016) and other reviews (e.g. Gillespie et al., 2015) that draw upon well-conducted RCTs conclude that there is limited evidence to support or refute the effectiveness of memory rehabilitation for stroke. Gillespie and colleagues (2015) note one report on data relating to stroke patients from an RCT of a pager-based reminding system, but argue this needs replicating before making recommendations. Das Nair and colleagues (2016) conclude that subjective measures of memory show some benefit of memory rehabilitation interventions but only at the first assessment point following intervention and no benefit is shown at longer-term follow-up (at least 3 months). No specific recommendation is made therefore with regard to specific interventions. On the other hand, Cicerone and colleagues (2011) recommend use of external compensations with direct application to functional activities for people with severe memory deficits after TBI or stroke, albeit the level of this recommendation is not at the strongest level. The NICE guideline (2013) notes the limited evidence to draw upon and suggests that one should use interventions for memory and cognitive functions that focus on relevant functional tasks, taking into account the underlying impairment. The NICE guideline also notes a range of possible interventions including interventions to increase awareness of the memory deficit; enhancing learning using errorless learning and elaborative techniques; using external aids (e.g. diaries, lists, calendars, and alarms); and making changes to the environment that will support memory.

Background to Interventions

Deficits in attention are common after stroke (Loetscher and Lincoln 2013). Prevalence estimates vary and depend upon time since stroke, but in terms of longer-term outcomes up to 50% of people may experience attention deficits (Barker-Collo et al., 2010; Loetscher and Lincoln, 2013). There are several forms of attention dependent upon different anatomical systems, and therefore a person may have a deficit in one system but not another. Although conceptualizations of attention differ, the systems are commonly described as arousal, alertness, selective attention, sustained attention, divided attention, and spatial orienting (dealt with later in terms of unilateral neglect).

Interventions

Interventions for attention deficits have mainly involved attempts to improve, or restore, normal attention functions through practising attention-demanding tasks, sometimes using computerized training programmes (Gillespie et al., 2015). Other interventions have involved learning mental strategies to manage attention in functional situations, but these interventions have not been the focus of RCTs in stroke patients.

Study Designs

As with memory, there are very few RCTs of attention training interventions in stroke. Loetscher and Lincoln’s (2013) Cochrane review includes six studies, four of which were parallel group studies and two were cross-over designs. The NICE guideline (2013) includes two RCTs (Westerberg et al., 2007; Barker-Collo et al., 2009). Most studies had small sample sizes and methodological limitations. Cicerone and colleagues (2011) included eight studies, with six of them being Class III level of evidence.

Evidence and Overview

Six studies including 223 participants compared CR with usual care (no treatment) (Loetscher and Lincoln, 2013). Meta-analyses demonstrated no significant long-term effect of CR on subjective measures of attention (two studies, 99 participants; SMD 0.16, 95% CI: –0.23–0.56; P = 0.41). In terms of effects at end of treatment, there was a borderline significant effect for subjective measures of attention (two studies, 53 participants, SMD 0.53, 95% CI: –0.03–1.08; P = 0.06). A statistically significant effect was found in favour of CR for immediate effects on standardized measures of divided attention (four studies, 165 participants; SMD 0.67, 95% CI: 0.35–0.98; P < 0.0001). No significant effects were found for other domains of attention including alertness (four studies, 136 participants, SMD 0.14, 95% CI: –0.20–0.48; P = 0.41), selective attention (six studies, 223 participants, SMD –0.08, 95% CI: –0.35–0.18; P = 0.53), or sustained attention (four studies, 169 participants, SMD 0.39, 95% CI: –0.16–0.94; P = 0.16). There were no significant effects on functional abilities in daily living (two studies, 75 participants, SMD 0.29, 95% CI: –0.16–0.75; P = 0.21). Loetscher and Lincoln concluded that the effectiveness of CR remains unconfirmed.

Cicerone and colleagues (2011) concluded that computer-based interventions could be considered as an adjunct to clinician-guided treatment of attention deficits after stroke. But relying on repeated practice on computer-based tasks without some involvement and intervention by a therapist was not recommended. SIGN 118 (2010) notes that there is not yet sufficient evidence to support or refute the benefits of CR for patients with problems of attention. NICE 162 (2013) suggests that ‘attention training’ may be considered, and alternatively interventions should focus on training people on functional tasks, and use strategies to manage the environment including using prompts that are relevant to the task.