Measuring Therapy Effectiveness in the Rehabilitation Stage of Aphasia

(1)

Measuring Therapy Effectiveness in the Rehabilitation Stage of

Aphasia

Name: Valerie van Dongen

Student Number: S3469530

Master’s Thesis Neurolinguistics

Supervisor 1: D.A. De Kok

External Supervisor: J.F. Feiken

University of Groningen

Word Count: 12.346

(2)

Voorwoord

Voor u ligt mijn scriptie van de master Neurolinguïstiek, die ik heb gevolgd aan de

Rijksuniversiteit van Groningen. Het schrijven van deze scriptie is voor mij het einde van een lange studieweg. Al tijdens mijn bachelor Engelse Taal en Cultuur aan de UvA wekte de neurologische kant van taal mijn interesse. Ik koos een minor in Logopediewetenschap in Utrecht, waarna deze interesse alleen maar groter werd. Voor de liefde van het vakgebied verhuisde ik naar Groningen om de master Neurolinguïstiek te volgen in combinatie met logopedie. Ik heb geen moment spijt gehad van deze keus, ik heb veel geleerd en ben alleen maar enthousiaster geworden over de neurologische kant van taal, met name afasiologie. Ik wist dan ook vanaf het begin van mijn master dat ik mijn scriptie wilde richten op dit onderwerp. Dit is gelukt en ik wil daarvoor graag meerdere mensen bedanken die mij geholpen hebben om mijn scriptie tot stand te brengen.

Allereerst wil ik Judith Feiken bedanken voor de mogelijkheid om mijn scriptie bij UMCG Centrum voor Revalidatie, locatie Beatrixoord te schrijven. Wij hebben regelmatig samen gebrainstormd wat ervoor zorgde dat ik steeds weer verder kon. Ondanks het feit dat ik door omstandigheden het onderzoek twee keer flink om moest gooien, bleef zij optimistisch en vol goede moed. Hierdoor lukte het om om te gaan met tegenslagen en door te gaan met schrijven, al dan op een andere manier dan ik voorafgaand voor ogen had. Verder wil ik Dörte de Kok graag bedanken voor de fijne begeleiding, de goede inzichten en de waardevolle feedback. Daarnaast wil ik mijn ouders bedanken, die mij de mogelijkheid geven om te studeren en mij daarnaast altijd steunen rondom studie en daarbuiten. Tot slot wil ik mijn vriend bedanken voor zijn steun en toeverlaat en de vrienden en familieleden die mijn scriptie lazen en voorzagen van feedback.

Ik hoop dat ik met mijn scriptie kan bijdragen aan het opzetten van het grote

effectiviteitsonderzoek naar Module Specifieke Therapie, en dat deze ook van waarde zal zijn voor andere studies binnen de revalidatiefase van afasie.

Valerie van Dongen Juli, 2020

(3)

Declaration of Academic Integrity

Verklaring Masterscriptie Faculteit der Letteren, Rijksuniversiteit Groningen Naam student: Valerie van Dongen

Studentnummer: S3469530

Masteropleiding – programma: Neurolinguïstiek

Titel scriptie: Measuring Therapy Effectiveness in the Rehabilitation Stage of Aphasia Naam scriptiebegeleider: D.A. de Kok

Hierbij verklaar ik ondubbelzinnig dat de door mij ingeleverde scriptie gebaseerd is op eigen werk en de vrucht is van een onafhankelijk wetenschappelijk onderzoek. Ik verklaar dat ik geen gebruik heb gemaakt van ideeën en formuleringen van anderen zonder bronvermelding; dat ik geen vertalingen of parafraseringen van tekstgedeelten van anderen als onderdeel van mijn eigen betoog heb ingezet; dat ik de tekst van deze scriptie of een vergelijkbare tekst niet heb ingediend voor opdrachten van andere opleidingsonderdelen.

Datum: 02-07-2020

Plaats: Groningen, Nederland Handtekening student:

(4)

Abstract

Little is known about applying single-case research designs in the rehabilitation stage of aphasia in order to determine therapy effectiveness. Therefore, a literature research was conducted, with the goal to discuss the different single-case research designs in light of validity, feasibility and ethical responsibility concerning the rehabilitation stage of aphasia. In this way, it was investigated how therapy effectiveness can be validly measured in the

rehabilitation stage of aphasia. That is, it was determined that employing a cross-over design in combination with a control task design is best suited for measuring therapy effectiveness, given that no baseline measurements are required, but still spontaneous recovery can be ruled out by using the assessments of independent language modules as control tasks for each other, and, if possible, a third, independent control task. In the end, these designs were also

recommended to use for an effectiveness research to the Module Specific Therapy as developed by Feiken, Links and Hüttman (2020).

Keywords: aphasia, rehabilitation stage, chronic stage, language modules, Module Specific Therapy, therapy effectiveness, single-case research designs, withdrawal design, control task design, item-specific design, multiple baseline design, cross-over design.

Abbreviations:

UMCG University Medical Center Groningen

PWA Person with aphasia

SLP Speech and language pathologist

PALPA The Psycholinguistic Assessment of Language Processing in Aphasia

MST Module Specific Therapy

STAPP Speech Therapy Application

(5)

1. Introduction

1.1. Aphasia

Aphasia is an acquired language disorder as a result of brain damage that occurs after the language has been developed (Bastiaanse, 2010). This brain damage is located in the dominant hemisphere, which is the left hemisphere for almost all the right-handers and for 70% of the left-handers. The most common cause of aphasia is a stroke, but aphasia can also gradually evolve consequently of a brain tumor, infection, or dementia, according to El Hachioui (2012). Aphasia depicts a central language disorder that, on the one hand can affect both the language production and language comprehension, and, on the other hand can affect both spoken and written language (Bastiaanse, 2010). Consequently, the severity of aphasia ranges from having difficulty with “infrequent words, complex sentences and texts, to being completely unable to speak, comprehend, read, or write” (El Hachioui, 2012, p. 13). As a result, aphasia has a great impact on a person’s communicative skills and with that on his or her social life (El Hachioui, 2012). According to Berns et al. (2015), in the Netherlands, the prevalence of aphasia is estimated at 30.000 and the incidence at 10.000. In addition to impaired language functions, aphasia often occurs together with impairments in other cognitive domains, such as executive functioning and memory (El Hachioui, 2012).

Moreover, aphasia should be distinguished from speech disorders, which depict problems in articulation. However, brain damage can also affect articulation, either due to damage to the speech musculature in dysarthria, or due to problems with planning or managing speech movements in apraxia of speech. It is very common that a person with aphasia (PWA) also has difficulty with articulation. That is, aphasia could be combined with either dysarthria or apraxia of speech or both (Bastiaanse, 2010).

1.2. The Model-Based Approach

Recently, a considerable amount of literature has been published concerning assessment and treatment of aphasia following a model-based approach. That is, according to Cardell and Chenery (1999), due to the growth of cognitive neuropsychology during the late 1980s and 1990s, the assessment of a PWA’s language deficits and remaining abilities became “a theoretically motivated and dynamic process”, which has a foundation of a detailed model of language processing (p. 354). By way of illustration, an example of a theoretical approach

(7)

based on models and theories from cognitive neuropsychology that can be used in the assessment of language disorders, is that of Ellis and Young (1988) (See figure 1).

Figure 1. Composite model for the recognition and production of spoken and written words (Ellis & Young, 1988, p. 195).

Such a model reflects the human mind and brain as an organized set of modules responsible for, for example, recognizing words, speaking words or understanding them. In this way, “every module engages in its own form of processing independently of the activity in modules other than those it is in direct communication with” (Ellis & Young, 1988, p. 9). Moreover, modules are also distinct within the brain, so that brain injury can damage the functioning of some modules, while the functioning of other modules is left intact. For example, a person can have difficulty with recognizing faces due to a stroke, without necessarily experiencing difficulties with reading (Ellis & Young, 1988). Or, in terms of language processing

specifically, a PWA can have difficulty with speaking but is still able to write or comprehend. For an in-depth description of the functioning of each module in language processing, see Ellis and Young (1988) or Feiken et al. (2020).

In line with the view of the brain as an organized set of modules as illustrated by Ellis and Young (1988), Cardell and Chenery (1999) demonstrated the “usefulness of employing model-based assessment to expose the underlying mechanisms responsible for an individual’s

(8)

disordered language” by identifying the underlying affected modules in a person with dysgraphia (writing disorder) with the Psycholinguistic Assessment of Language Processing in Aphasia (PALPA) (Kay, Lesser & Coltheart, 1996) (p. 380). Specifically, the PALPA is a model-based assessment resource used to identify the different processing mechanisms that may underlie the same surface symptoms in different PWAs (Cardell & Chenery, 1999). In other words, “the misnaming of pictured objects may be due to a disruption to the semantic specification of that concept, or a difficulty in retrieving the lexical label for that concept, or an inability to produce the necessary phonological form of the word” (Cardell & Chenery, 1999, p. 354). Therefore, it is important to know which underlying process is damaged and causes the language disorder, since only then treatment can be optimally effective, as argued by Cardell and Chenery (1999). To be specific, the PALPA is a diagnostic instrument that consists of 60 subtests, which are based on a “psycholinguistic approach to the interpretation of processes in the recognition, comprehension and production of spoken and written words and sentences” (Cardell & Chenery, 1999, p. 360) and is designed to assess language processing abilities in people with aphasia (Kay, et al. 1996). That is, various tests can be selected from the PALPA test battery to assess the modules that might be underlying to the language disorder as it includes specific subtests devoted to the assessment of individual language modules.

1.3. Module Specific Therapy

In line with the importance of a model-based assessment of aphasia, as demonstrated by Cardell and Chenery (1999), Feiken et al. (2020) developed a model-based therapy

programme for rehabilitating persons with aphasia: Module Specific Therapy (MST: Module

Specifieke Therapie). There are two main approaches in the treatment of aphasia, which are

cognitive-linguistic treatment (CLT) and communicative treatment. With CLT, the speech and language pathologist (SLP) sets up a therapy plan based on a linguistic analysis of test data, i.e. data from the PALPA (De Jong-Hagelstein et al., 2012, Feiken et al., 2020). In this way, CLT focuses on the linguistic level that is held responsible for the symptomatology, which are semantics, phonology and/or syntaxis. Eventually, the goal of CLT is activating such a level and with this stimulating the language production and/or comprehension. This is in contrast to communicative treatment, which is aimed at “optimizing information transfer by training compensatory strategies and use of residual skills” (De Jong-Hagelstein et al., 2012, p. 2) and

(9)

is thus not specifically aimed at restoring or reactivating underlying deficits of the language problem(s).

The MST is another approach within the cognitive linguistic therapy (CLT), as it connects directly to assessment focused on the underlying disorder(s) in language processing. In other words, the MST provides a theoretical framework for a therapy plan that taps into the underlying disorder which is primarily responsible for the symptomatology in terms of

language skills, e.g. oral production, written production, auditory and visual understanding (Feiken et al., 2020). In this way, treatment for each module is different in terms of its practice tasks, cues, strategies, and communication advice for both the PWA and his or her interlocuter. On top of that, the level of difficulty of the therapy programme is determined by linguistic variables such as wordlength, frequency and imaginability, which are factors that are of influence on the individual modules (Feiken et al., 2020). The MST focuses not only on neural reactivation, but also on the recovery mechanism called ‘neural adaption’ of the brain. In case of neural adaption, affected brain cells are compensated or ‘supported’ by other, non-affected brain cells. In some cases, neural adaption activates rearrangement, i.e. neural reorganization, of a brain function, and in other cases it activates compensation of a brain function, i.e. neural compensation. Though neural adaption can lead to positive adaptations, it can also cause negative adaptations when, for example, the non-affected hemisphere

suppresses the affected hemisphere (Geurts & Verheyden, 2014; Feiken et al., 2015), which slows down recovery of the brain and maladaptive plasticity can follow (Papathanasiou, Coppens & Ansaldo, 2013; Feiken et al., 2015).

To continue, the MST distinguishes between different therapeutic approaches, such as stimulation therapy and strategy training, which are approaches that stimulate the neural adaptation of the brain in the acute (first two weeks post-onset1_{) and rehabilitation stage (up}

to six months post-onset2_{) (Feiken et al., 2020). In short, stimulation therapy is a form of}

implicit learning as the SLP facilitates the PWA to carry out activities in exactly the same way as it was done premorbidly. That is, the SLP does not explain or teach specific strategies and the PWA learns by doing (Feiken et al., 2015). In this way, stimulation therapy appeals to the most efficient neural control, which involves the least possible neural adaptation. When the PWA is not able to carry out the task, the SLP gives cues. In this form of therapy, several modules can be addressed at once in order to carry out the language task. In contrast to

(10)

stimulation therapy, with strategy training the SLP teaches the PWA to use strategies. This makes strategy training a form of explicit learning by appealing to functions of the brain which were not previously involved in the target activity, such as other linguistic

functions/modules and/or non-linguistic cognitive functions, e.g. memory and executive functions. Specifically, the PWA learns to cue himself, contrary to stimulation therapy with which the SLP cues the PWA (Feiken et al., 2020; Feiken et al., 2015). Altogether, MST involves both therapeutic approaches and provides tasks for each module in the Speech Therapy Application (STAPP) (Feiken, 2019), which is an application available on iPad, that connects to the module specific treatment of aphasia (Feiken et al., 2020). In this application, the PWA can practice tasks that are focused on a specific module, for example the Speech Output Lexicon (SOL), which is a module of the modality concerning speech production. For this specific module, several tasks are available, for example, naming of nouns or verbs, enumerating rhyme words, etc. All tasks are provided with the options of choosing either a cue (implicit learning) or a strategy (explicit learning). In this way, the STAPP can be used in different stages of the rehabilitation of a PWA. It can be used within therapy sessions but also at home so that the PWA can practice independently from the SLP.

Nevertheless, the effectiveness of Module Specific Therapy (Feiken et al., 2020) together with the use of the Speech Therapy App (STAPP) (Feiken, 2019) has not been scientifically proven yet. Therefore, the current research will be part of the upcoming effectiveness study to the MST that will be carried out at the UMCG Center for

Rehabilitation, location Beatrixoord (Beatrixoord) in Haren, Groningen, by providing a literature research on single-case research designs that can be used to measure therapy

effectiveness. On top of that, a discussion of their applicability concerning the clinical setting of PWAs in the rehabilitation stage will be given. Specifically, throughout this thesis,

effectiveness is used to refer to “whether an intervention does more good than harm when

provided under usual circumstances of healthcare practice”, as defined by Kim (2013, p. 227).

1.4. Measuring Therapy Effectiveness Throughout the Different Stages of Aphasia

As argued by Whitworth, Webster and Howard (2014) significant improvement in performance does not automatically indicate that a specific therapy is effective for that person. That is, improvement directly resulting from therapy should be distinguished from “spontaneous recovery and non-specific effects of treatment due to such influences such as

(11)

support, social participation and encouragement” (Whitworth et al., 2014, p. 111). According to Whitworth et al. (2014), single-case study methodology allows inferences to be made about the effectiveness of a specific treatment as it has been widely used to distinguish these causes of improvement. Moreover, experimental single-case studies seem to be become more and more relevant in providing strong evidence on treatment effectiveness (Howard, Best & Nickels, 2015). On top of that, Leff and Behrmann (2008) argue that single-case studies play an important role in informing therapy concerning heterogenous disorders, such as aphasia. However, most single-case research studies in the field of therapy effectiveness concerning the treatment of aphasia have been carried out in the chronic stage (after six months post-onset3_{) (see for example, Nickels, 1992; Fink et al., 1992; Raymer, Cudworth & Haley, 2003;}

Renvall, Laine & Martin, 2005). That is, up to now, data about the effectiveness of therapy at the beginning of rehabilitation (= end of the acute stage/start of the rehabilitation stage) of PWAs are limited. Although the difference between the acute, rehabilitation and chronic stage is fundamental, it is not clearly described in the literature due to changes within the medical care and different terminology across hospitals and institutions. Therefore, the Logopedische

Richtlijn ‘Diagnostiek en behandeling van afasie’ (Berns et al., 2015) describes the stages

from a functional perspective; the acute stage could be described as the stage in which psychical functioning is still subject to (strong) change. This stage is followed by the

rehabilitation stage in which most recovery takes place (Bastiaanse, 2010; de Jong-Hagelstein et al., 2011). Though this recovery is determined individually as some functions can recover spontaneously well and quickly, e.g. walking, other functions enhance further by training and therapy, such as cognitive functions, of which the language functions are a part (Berns et al., 2015). Lastly, the chronic stage is characterized by that no great improvements in terms of language functions occur (Bastiaanse, 2010). However, this does not mean that there is no improvement of language functions possible, or that new skills cannot be developed, or strategies cannot be learned. Years post-onset, PWAs can still learn new skills and experience progress of language functions, which lead to better communication and a greater

self-reliance, as explained in Berns et al. (2015) (Ruiter, Kolk, & Rietveld, 2010; Meinzer,

Rodriguez, & Gonzalez-Rothi, 2012). As the stages differ fundamentally, so does the amount and rate of spontaneous recovery across the stages. That is, according to Cassidy and Cramer (2017), spontaneous recovery is often “incomplete and the recovery rates of neurological

(12)

function vary”. Factors such as “genetics, previous comorbidities, initial severity of deficits, age, stroke mechanism”, lesion size and location of the stroke and areas that are spared all influence the amount and speed of recovery (p.33). Specifically, spontaneous gains in

language function might be varying from a period of months up to years post-stroke (Cassidy & Cramer, 2017). Up till now, there is no strict consensus about until when spontaneous recovery takes place, but in general it is assumed that spontaneous recovery occurs during the first-year post-stroke, in particular during the first three months (Visch-Brink, 2006).

Following Berns et al. (2015), this period comprises the acute stage and partly the

rehabilitation stage. Because of the differences in stages, and with that recovery, effectiveness studies that measure therapy effectiveness in the chronic stage cannot be compared to studies that measure therapy effectiveness in the acute or rehabilitation stage. More specifically, the research designs that are used in the former stage, should not be used in the same way with PWAs who are still rehabilitating, since both stages differ fundamentally in terms of (spontaneous) recovery, as explained above. On top of that, factors such as ethical

responsibility and time that is available to treat the PWA at the rehabilitation center should be considered when choosing a research design to employ in the rehabilitation stage. In addition, Francis, Clark and Humphreys (2002) claim that “research into which therapy designs can be used most practically in a ‘‘real-life’’ environment is as urgently needed as research into therapy efficacy itself” (p. 256). Therefore, as a preparation on the effectiveness research that will be carried out at Beatrixoord to the effectiveness of the MST, the current research will present a literature review on types of (single-case) research designs. In this way, it is hoped to find out which single case research design or combination of types can be used best for measuring therapy effectiveness in the rehabilitation stage of aphasia. With that, a

contribution to the upcoming study at Beatrixoord for measuring the effectiveness of the MST in combination with the STAPP will be made. This will be done by describing and critically discussing several single-case research designs that are used in therapy effectiveness studies concerning the different stages (mostly chronic) of aphasia. On top of that, the designs will be put in the light of the rehabilitation stage in order to find out whether they are suitable for employing in this stage.

(13)

2. Aims and Scope of the Present Study

As most single case research designs seem to be employed in the chronic stage, little is known about applying them in the rehabilitation stage. Therefore, a literature research will be carried out, with the goal to discuss the different single case research designs in relation to the

rehabilitation stage. Specifically, the different research designs will be discussed in light of validity, feasibility and ethical responsibility concerning the rehabilitation stage of aphasia. The following research question will be answered:

• How can therapy effectiveness be validly measured in the rehabilitation stage of aphasia?

In the end, recommendations will be made concerning the use of single case research designs in the rehabilitation stage, that can be used to measure the effectiveness of the MST, as preparation on the coming effectiveness study at the UMCG Center for Rehabilitation, location Beatrixoord. Following this, a final product will be delivered by way of a ‘step-by-step plan’, using the most suitable design(s), which will be delivered to Beatrixoord.

3. (Single-Case) Research Designs

3.1. Randomized Control Trial

Although around the 1970s the randomized controlled trial (RCT) was considered to “be the design of choice for determining the efficacy of treatment for the disorder, aphasia”, recent developments in the field of cognitive neuropsychology have led to a renewed interest in using individuals as their own controls (Wertz, 1992, as cited in Franklin, 1997, p.402). This in turn has led to the development of methodologies for single-case effectiveness studies (Franklin, 1997). That is, in general, in RCTs, individuals are randomly assigned to either a treatment or a no treatment group and recovery of both groups is compared following treatment. Logically following, a treatment turns out to be effective when the treated group makes a significantly greater recovery than the untreated group. This design was initially used for medical treatments (e.g. drug trials), but it has also been used for the evaluation of speech therapy for PWAs (Franklin, 1997). However, in the latter case RCTs do usually not depict a treatment versus no treatment situation, but two treatment approaches are compared to each

(14)

other. For example, in the Rotterdamse Afasie Therapie Studie (RATS) (Visch-Brink, 2006) the effectiveness of two treatment approaches, that is, semantic therapy and phonological therapy, were investigated in the chronic stage of aphasia. 58 persons with a combined semantic and phonological disorder were included and randomly divided over two groups. In other words, the RCT was not used in a way that one group received treatment, and the other group received no treatment, but both treatment approaches were randomly assigned to both groups. It was hypothesized that semantic treatment would be more effective in terms of communication than phonological treatment. However, this hypothesis was not confirmed as both treatment approaches resulted in an improvement in terms of verbal communication (Visch-Brink, 2006). The same research design, i.e. RCT, was used in the RATS-2 study, in which CLT in comparison to communicative treatment in the acute stage of aphasia was investigated (De Jong-Hagelstein et al., 2012). In this study, 80 PWAs were included within three weeks post onset and received either CLT (semantic and/or phonological training) or communicative treatment for six months. It was found that PWAs in the acute stage did not benefit more from CLT than from a more general communicative treatment (De

Jong-Hagelstein et al., 2012). Altogether, in both studies, improvement of verbal communication in PWAs (with at least a semantic and/or phonological disorder) following a particular therapy was measured by comparing two groups of PWAs that received a different treatment. In other words, all participants within each group received the same treatment, without taking into account the underlying deficits of their language problems in detail, after which improvement was compared between groups. Nevertheless, Howard (1986) argues that randomized

controlled trials of aphasia therapy are not sensible enough to investigate a “set of

heterogeneous treatment techniques applied to a heterogeneous group of subjects” (p. 90). In other words, as PWAs are not a homogeneous population but rather a heterogenous one, it is “impossible to compare the outcome for two different patient groups” (Howard, 1986, p. 98, Franklin, 1997). Following this, even when the same treatment is given, as was done in the RATS-studies, heterogenous groups should not be compared as the underlying deficits differ from PWA to PWA, resulting in different responses to therapy. Moreover, with a

heterogenous population it is not possible to generalize results to all PWAs. Therefore, any comparisons should be made within individuals, as argued by Howard (1986) and Franklin (1997). On top of that, a more sensible approach is establishing whether a specific treatment works for one PWA, and then investigating whether that approach can be effectively applied to other PWAs with similar problems, i.e. “a single case study with replications” (Howard,

(15)

1986, p. 97). Altogether, this emphasizes the importance of the development and utilization of methodologies for single-case effectiveness studies, which are not only useful as a research tool but also as “a clinical tool to measure outcome in everyday treatment regimes” (Franklin, 1997, p. 402). For this reason, in the following paragraphs, several types of single-case research designs will be taken into account and discussed.

3.2. Withdrawal Design

According to Thompson (2006), a traditional single-case research design is the withdrawal design, also called the A-B-A-B design, in which the behaviors under study are first measured in the baseline (A phase), after which treatment is applied in the first B phase. Then, this treatment is withdrawn in the second A phase and finally reapplied in the second B phase. An important aspect of this design, that at the same time becomes problematic in treatment of aphasia, is that no treatment is given in the second A phase. That is, Thompson (2006) argues that “the goal of treatment research generally is discover methods to improve language function” (p.9). Therefore, reversal is not desirous if the treatment is successful. Similarly, reversal might not be possible as withdrawing treatment should not result in unlearning. Consequently, Thompson (2006) states that “this practice in aphasia treatment is not

recommended” (p.9). Additionally, as argued by Morgan and Morgan (2008), the A-B-A-B design becomes “problematic when changes in behavior become permanent and unlikely to return to preintervention or baseline levels, or when removal of a clinical treatment may be ethically questionable” (p.127).

However, as pointed out by Kearns (1986) (in Thompson, 2006), the withdrawal design is best suited to be used when studying behaviors that are either likely to or expected to reverse following the withdrawal of treatment. For example, Renvall et al. (2005) used the design in an anomia treatment study to investigate the effect of contextual priming (CP), which is a technique that “includes cycles of repeating and naming items in different contextual conditions”, in which items are either only semantically related, phonological related or completely unrelated (baseline condition) (p. 327). A 73-year-old right-handed male PWA (two years post-stroke onset; chronic stage) who was not receiving therapy at the time of the study was investigated. The goal of the study was to find out whether CP would have a facilitative effect on naming. It was expected to find an effect of direct interference in the semantic condition and only little and short-term improvement on the trained items. After measurement during the treatment program, no treatment was given for 1.5 months, after

(16)

which a follow-up measurement was administered. Their hypothesis could be confirmed; no long-term effects of treatment were found and a return to baseline level was shown at the 1.5 months follow up, while during treatment a significant improvement on naming was noted, indicating only direct and short-term effects of CP (Renvall et al., 2005).

3.3. Control Task Design

Franklin (1997) hypothesizes that if spontaneous recovery produces language improvement, then all aspects of language should improve at once. This is in contrast to treatment targeted at particular aspects of language, which should produce improvement in specific tasks. Consequently, when tests that “tap the processes at which treatment is targeted improve following therapy”, while other tests indicate no change, a treatment-specific effect can be demonstrated (p. 404). A design that connects to this is the control task design, which is a design that is easy to implement and therefore suitable for use in the clinic, according to Franklin (1997). In this design, the process targeted in treatment, and an unrelated, control process are tested both preceding and following treatment. Performance on the treated process will improve if treatment is effective, but there will be no or minimal improvement on the unrelated process. In this way, a treatment-specific effect will be demonstrated (Whitworth et al., 2014).

Nickels (1992) (In Franklin, 1997) studied the effectiveness of a treatment programme concerning naming and oral reading by using a control task design with TC (two years post stroke onset; chronic stage). Similar to the PWA Renvall et al. (2005) studied, no spontaneous recovery was expected because of the stage that he was in. Nickels (1992) (in Franklin, 1997) established a therapy specific effect as TC did not improve on a number of tests (e.g.,

nonword reading and auditory comprehension), on which performance was not expected to improve. Instead, TC did only improve on the language processes that were treated and showed no improvement on processes that were not targeted for treatment. Therefore, Nickels (1992) was able to rule out spontaneous recovery or a ‘charm’ effect, i.e. improvement that is due to only contact with the therapist and not specifically due to treatment (Best et al., 2011), as then TC’s performance should have enhanced on treated as well as untreated tasks, as explained by Franklin (1997). However, Franklin (1997) argued that the nonword reading task was at floor level. Therefore, the control task might not be subjected to spontaneous recovery and therefore the found significant difference could still be attributed to spontaneous recovery and/or a charm effect. This in turn implies that Nickels (1992) failed to distinguish between

(17)

these effects and a specific treatment effect, as stated by Franklin (1997). Altogether, the control task should be one which is “as likely to recover as the treated task, both in terms of what we know about patterns of recovery and in terms of degree of deficit”, in order to gain reliable results with the control task design, as claimed by Franklin (1997, p. 407). On top of that, more than one assessment should be carried out for each task due to possible variability, and the chosen assessments should be sensitive enough to demonstrate change (Franklin, 1997).

3.4. Item-Specific Design

With an item-specific design, “treatment is preceded by testing of performance on items to be treated” and items that will not be treated (Whitworth et al., 2014, p. 112). In other words, significant progress can be seen on treated items, and no or minimal progress on untreated items if the treatment is effective, except for when generalization occurs across items

(Thompson, 2006). More specifically, prior to therapy, two sets of items are assessed (usually words). After this, one-half of the items will be treated, while the other half will remain untreated. Then, if treatment has worked, the treated items will improve, while the untreated items will not change. Spontaneous recovery can be eliminated as such an effect will be seen through improvement in both treated and untreated items, which then will improve equally (Franklin, 1997).

Fink et al. (1992) used such a design in an effectiveness research to verb retrieval in a PWA (8 years post-onset; chronic stage) with a severe agrammatic Broca’s aphasia. The effectiveness of two treatment approaches were examined, i.e. Direct Verb Training (DVT) and Verb Priming (VP). That is, DVT involves the subject to be trained to name the verb that describes a shown action and to report the agent and the theme of that action consequently. In the end, the subject has to construct a sentence that incorporates the information while

describing the picture. In short, by this approach, the subject is made aware of the target verb and its noun phrases that fill thematic roles (Fink et al., 1992). With VP, the subject is being asked to repeat a sentence that incorporates the target verb. After this, the subject has to describe a picture that is designed to elicit the same target (primed) verb (Fink et al., 1992). Fink et al. (1992) hypothesized that the verb retrieval deficit, that is rooted in impaired access to lexical phonology, “leads to the prediction that training on a particular verb should increase access to that verb but not generalize to use of other verbs” (p. 270). In other words, no generalization was expected to other items besides the trained verbs, which in turn makes the

(18)

item-specific design a suited design for this study. A set of untrained items was used as a control set in order to study the effectiveness of the DVT. As a result, it was indeed found that the DVT procedure caused improvement only on the trained items, but not on untrained items, confirming that no generalization had taken place and the treatment had worked for the

treated items only. In another study by Raymer et al. (2003), the patterns of spelling generalization resulting from treatment in a person with severe dysgraphia (2 years post-onset; chronic stage), NM, was examined. NM received training on two sets of items, and treatment effects were monitored with sets of untrained items. In short, Raymer et al. (2003) used this design in order to be able to consider “the possible mechanisms of his spelling improvements following training”, by examining the patterns of spelling generalization in untrained words (p. 621). That is, by investigating these patterns, it was, among other things, concluded that treatment effects that affected the orthographic output lexicon are assumed to result in little generalization to untrained whole-word spellings (Raymer et al., 2003).

3.5. Multiple Baseline Design

According to Thompson (2006) “the multiple baseline strategy is the most commonly used and is likely the best suited for most studies of aphasia” (p.11). In this design, treatment is preceded and followed by assessing the targeted process of treatment and a series of other processes. These other processes are either unrelated to the targeted process and are not expected to improve by treatment or related and show generalized effects of treatment (Whitworth et al., 2014). When using this design while the PWA is not in a period of spontaneous recovery, no improvement will be visible through repeated testing without intervention, i.e. the pre-treatment baseline measures. That is, with the multiple baseline design, spontaneous recovery can be measured by administering the same task multiple times, as explained by Franklin (1997). By way of illustration, if spontaneous recovery is not

occurring, then the “pre-treatment baseline measures will not differ significantly from each other, and any significant difference in the test score following therapy” will demonstrate a treatment effect (Franklin, 1997, p. 407). In contrast, the pre-therapy baseline will differ significantly when spontaneous recovery does occur. In this case, a line can be fitted to the baseline measures in order to predict the development of further spontaneous recovery (Franklin, 1997). To illustrate this, see figure 2.

(19)

Figure 2. A multiple baseline design where spontaneous recovery is still occurring (Franklin, 1997, p. 408).

Nevertheless, it should be kept in mind that the fitted line will be a broad estimate of expected spontaneous recovery, seen that the rate of spontaneous recovery declines over time.

Therefore, any improvement over and above this line measured post-therapy, can be assumed to be caused by the treatment (Franklin, 1997). In other words, the multiple baseline design requires to establish a stable performance before therapy commences, by the administration of several repeated pre-therapy tests. In this way, by using the multiple baseline design,

spontaneous recovery can either be ruled out or its rate can be estimated. Moreover, as argued by Franklin (1997), assuming that improvement caused by different treatment approaches is measured by using different tasks, the design is suitable for measuring the effect of several treatment approaches at once.

The design was employed by Morris et al. (1996) in order to investigate the effect of lip-reading information on the performance on several tasks of speech discrimination in a PWA with multiple language deficits (13 months post-onset; chronic stage). No difference was found between two administrations of three pre-therapy tests, indicating a stable performance. After therapy was applied, improvement was only found on the tests of

phoneme discrimination (i.e. the targeted process) and the other two tests remained stable and did not change in comparison to the pre-therapy administrations. Therefore, as argued by

(20)

Morris (1996) this indicated that the effects were not a result of spontaneous recovery but were a direct result of the intervention.

3.6. Cross-Over Design

The cross-over design is an “extremely robust single-case design well suited to dealing with the problem of having controls for spontaneous recovery” (Grayson, Hilton & Franklin, 1997, p.271). Moreover, according to Lum (2005), the cross-over design is “popular in single case-studies” as it avoids the ethical problem of the withdrawal design, i.e. a no-treatment

condition. In a cross-over design, first one process will be treated and then the treatment of another process will follow. The design requires the assessment of performance on both tasks for three times: prior to treatment A, after treatment A, and after treatment B (Franklin, 1997). More specifically, according to Whitworth et al. (2014), two phases of treatment (A and B) are preceded and followed by the testing of two processes. Process 1 is treated initially and followed by the treatment for process 2. Consequently, if the treatment is effective, then significant improvement will be found in process 1, following the first phase of treatment, and significant improvement will be found in process 2, following the second phase. In this way, it can be established whether treatment A is better or comparable to treatment B, by

comparing the pre-test results with the reassessment (Lum, 2005). In other words, if both treatments are effective, this will be seen through a better performance on the task assessing process 1 after applying treatment A, and vice versa for process 2. In short, the design deals with the impossibility of achieving a stable baseline and allows to compare the recovery rates of different processes with themselves, serving as a control measurement.

Grayson et al. (1997) employed the cross-over design by using four tests in order to examine individual aspects of language processing in a PWA during the first six months post-onset. The tests were repeated after each phase of treatment (e.g. semantic, both semantic and auditory, and sentence therapy respectively). In other words, the four tests were used as control tasks for each other. Consequently, it was argued that with the design it was possible to detect any improvement in “the selected language processes regardless of whether the subject was receiving direct therapeutic intervention” (Grayson et al., 1997, p. 263).

Therefore, it was hypothesized that, if there would be no improvement in any area except the one that is receiving treatment, then that progress might be a direct result of the treatment and not due to spontaneous recovery (Grayson et al., 1997). In line with this, it was expected that if the recovery would be accountable in terms of spontaneous recovery, then a gradual

(21)

improvement would be seen in certain areas. Following the results of Grayson et al. (1997), improvement was indeed visible on the spoken and written word-to-picture matching tests following a period of semantic therapy, in contrast to the minimal pair test or the TROG (Bishop, 1982) (i.e. a sentence comprehension test), on which there was no significant improvement after the first stage of therapy. Moreover, there was a significant improvement on the minimal pairs test following the auditory and semantic phase of therapy, while there was still no improvement on the TROG, supporting the former finding. Interestingly, significant improvement on the TROG became visible only from the third phase of therapy, which was focused on processing “an increased number of information-carrying words” (Grayson et al., 1997, p. 271). Altogether, Grayson et al. (1997) were able to demonstrate that the found significant improvements were not entirely accountable regarding spontaneous recovery, but were due to treatment, confirming their hypothesis and expectations. Moreover, as argued by Grayson et al. (1997), the design of their study did not only allow to demonstrate therapy specific improvements, it also permitted the therapy to be focused, i.e. the therapy was given in phases directed at different language processes, which makes the therapy programme clear and structured for both the PWA and the clinician.

3.7. Multiple Baseline Across Behaviors Design

The last design to be presented is a variant of the multiple baseline design that is combined with elements of the cross-over design, that is, the multiple baseline across behaviors design. It is a frequently used alternative to the withdrawal design and essentially is a series of A-B designs in which baseline data are collected on at least one behavior of the participant. However, in contrast to the withdrawal design, the multiple baseline design across behaviors does not require returning to baseline levels of responding in order to demonstrate internal validity (Thompson, 2006). By way of illustration, in the A phase, baseline data are collected on two or more (independent) behaviors. Then treatment is applied to one behavior during the B phase, while the A phase is continued for the other behaviors under study. When a

treatment effect is determined for the first behavior, that is, the process that is being treated, the treatment is extended to the second behavior, and so on, until all behaviors have been treated. In this way, a specific treatment effect can be established when changes in the treated process occur only when treatment is applied for that process (i.e. the B-phase), while the baseline performance of untreated processes remains stable until treated, which in turn indicates experimental control, according to Thompson (2006). Furthermore, processes, i.e.

(22)

the language behaviors, must be functionally independent and be sensitive to the treatment under investigation (Thompson, 2006, Francis et al., 2002). If processes are related to each other, generalization may occur across the different behaviors. As Thompson (2006) argues, a generalization effect in this design is a “disaster experimentally”, as experimental control is then lost (p.10). Consequently, it will then be unknown which treatment caused the

improvement of a particular process (Thompson, 2006).

Besides the multiple baseline across behaviors design, there are also the multiple baseline across subjects and the multiple baseline across settings/situations design. The multiple baseline across settings/situations is similar to the multiple baseline across behaviors design. However, it differs in that the treatment or intervention is applied to a single behavior of a participant across different and independent environmental conditions (Murphy & Bryan, 1980). The multiple baseline across subjects design requires data from two or more

participants with exactly the same disorder who are exposed to similar environmental conditions (Murphy & Bryan, 1980). The intervention will begin at different points for each participant and the baseline phases will be of unequal lengths (Morgan & Morgan, 2008).

Francis et al. (2002) used the multiple baseline across behaviors design to investigate the effect of circumlocution-induced naming (CIN) on persons with pure anomia, which are word finding difficulties in the absence of semantic impairment. CIN involves a technique which requires the PWA to describe and talk around a picture as long as necessary to get to the name of it, without cues from the SLP (Francis et al., 2002). The PWA (MB) Francis et al. (2002) studied, a person with anomia who had impaired access to the SOL, was in the

rehabilitation stage as the intervention started at 3 months post-onset. It was found that MB improved on naming and that this improvement generalized to untreated items following the CIN therapy, which was focused on the access to the SOL. However, Francis et al. (2002) argued that at least some of the progress MB made, was caused by spontaneous recovery as she was only 2-3 months post-onset. In despite of the baseline measurements they obtained, it was argued that different cognitive abilities do not automatically recover with the same speed. Therefore, Francis et al. (2002) argued that their design might be somewhat weak to establish a treatment specific effect in a stage in which spontaneous recovery still occurs. In other words, though the extended baseline that Francis et al. (2002) did have, they argued that MB’s anomia might be the most benefiting from spontaneous recovery as MB was spending much time on trying to speak every day apart from the therapy she was receiving. However, no improvement on word finding was found a month after the first assessment, while other

(23)

domains, e.g. speech comprehension, and verbal memory, did improve. This indicated that other domains than spoken language benefited from spontaneous recovery, despite MB practicing more with spoken language. On top of that, improvement was only visible after therapy and did not persevere after treatment was ended (Francis et al., 2002). Altogether, Francis et al. (2002) still stated that spontaneous recovery “may have had some influence on the amount of generalization” (p. 256).

4. Applicability of Single-Case Research Designs in the

Rehabilitation Stage of Aphasia

Howard et al. (2015) provide an evaluation of methods in single case study designs

concerning interventions in the domain of speech pathology by focusing on the treatment of aphasia in particular. However, the set of design principles that they present is not focused on the rehabilitation stage specifically. Therefore, in the following sections, the above-described types of single-case research designs will be discussed in light of their advantages and

disadvantages in relation to their applicability in the rehabilitation stage. Important factors considering the rehabilitation stage, such as ethical responsibility, e.g. periods without treatment, the ability to rule out spontaneous recovery and time that is available for

(additional) assessment and/or therapy in a rehabilitation center, will be included. On top of that, it will be shortly discussed whether the applicable designs are also suitable for measuring the effect of the different treatment approaches within the Module Specific Therapy, that is, stimulation therapy and strategy training, as explained in the introduction of this thesis. In the end, recommendations will be made on using single case research designs in order to measure the effectiveness of the Module Specific Therapy (Feiken et al., 2020) in the upcoming effectiveness research.

4.1. Withdrawal Design

To begin with, a general disadvantage of the withdrawal design, also called the A-B-A-B design as used by Renvall et al. (2005), is that no treatment is given in the second A-phase. This design is therefore considered to be unsuitable and not recommended for using with PWAs in general, as any improvement caused by therapy is usually not desirous to be reversed (Thompson, 2006). On top of that, removal of any treatment might be ethically

(24)

irresponsible (Morgan & Morgan, 2008). Seen in this light, employing the design in the rehabilitation stage (i.e. the stage in which most recovery takes place (Bastiaanse, 2011)) might not be responsible. In short, removal of any treatment in this stage will be ethically irresponsible, as the possibility to recover by treatment will then be taken away. Therefore, the design is more suitable when it is expected that improvement only occurs at the moment of treatment and a reversal occurs when no treatment is given, as argued by Kearns (1986), and when a PWA is in the chronic stage in particular, as in Renvall et al. (2005). That is, the PWA studied by Renvall et al. (2005) was in the chronic stage of aphasia and was not

receiving treatment at that moment. Therefore, refraining the PWA from therapy might not be considered to be unethical. On top of that, assuming that the PWA will not make much

progress in the chronic stage, refraining the PWA from therapy is not likely to cause harm. This is in contrast to the rehabilitation stage, in which the PWA should receive treatment as in this stage most recovery (by treatment) is likely to occur (Bastiaanse, 2010; de

Jong-Hagelstein et al., 2011). Altogether, the withdrawal design is unsuitable for measuring the effectiveness of a specific treatment in the rehabilitation stage of aphasia. On top of that, the design is more suitable to use in the chronic stage when the PWA does not receive treatment, or when a return to baseline level is expected when withdrawing treatment, due to the nature of the treatment task. In the end, when applying the MST in the rehabilitation stage, a return to baseline level is not desired since the goal of treatment is to improve a specific language skill regardless of whether stimulation or strategy therapy is given. That is, for both

approaches within the MST, any improvement caused by therapy is not desirous to be reversed. On top of that, when measuring the effectiveness of the MST, there should not be periods of no treatment since treatment should be given in this stage, as explained above. This in turn makes the withdrawal design unsuitable to measure the effectiveness of the MST in the rehabilitation stage.

4.2. Control Task Design

In a control task design, the process targeted in treatment and an unrelated control process are tested preceding and following treatment. This design is easy to use and not time-consuming at all, but choosing the right control task is fundamental, so that any effectiveness of therapy can be established validly. However, selecting an appropriate control task could be difficult. That is, as Franklin (1997) claims, “for this design to work, it makes the assumption that all areas of language will recover at the same rate” (p. 406). Nevertheless, different cognitive

(25)

abilities do not automatically recover at the same rate (Francis et al., 2002), so this will also not be the case with aphasia. For example, in general, comprehension improves faster and more thoroughly than production does, following group studies, according to Basso (1989) (In Franklin, 1997). In other words, when two tasks from different modalities, both with a

differing rate of recovery, are selected to be compared, any significant recovery can be due to spontaneous recovery. By way of illustration, when treatment for comprehension is contrasted with a control task which requires word finding, any significant difference could still be explained in terms of spontaneous recovery, as argued by Franklin (1997). On top of that, the control task should not be very much more impaired than the task to be treated, as this would also give misleading results. By way of explanation, “even though spontaneous recovery is occurring”, the particular aspect of language may be so very impaired that it is not responsive to this recovery, i.e. at floor level, and thus any treatment effect could still be attributed to spontaneous recovery. Likewise, the control task should not be at ceiling level, as also then any improvement is not likely to be statistically significant (Franklin, 1997, p. 406). By way of illustration, when a control task is at ceiling level, performance is not subjected to

spontaneous recovery. If the targeted process is then measured to be improved after therapy and performance on the control task does not enhance, it cannot be established whether the improvement is caused by therapy or by spontaneous recovery, given that the control task did not improve. Altogether, the control task should tap a process which is likely to recover at the same rate as the treated process and is neither at floor level nor ceiling level. Only when the language processes that are not treated do not improve in contrast to the targeted language process(es), a therapy effect can be established (Franklin, 1997). Therefore, control tasks should be chosen carefully. Furthermore, according to Cardell and Chenery (1999), in order to measure possible generalization effects to other processes by treatment, it is necessary to include control tasks that evaluate both “functions related to the target behavior” that are expected to improve due to treatment and “functions unrelated to the target behavior” that are not expected to be influenced (p. 371).

For the chronic stage, i.e. the stage in which no spontaneous recovery is expected, the control task design is suitable for ruling out a charm effect, which is improvement caused by pure contact and attention from the practitioner(s) and other persons in the environment of the PWA. That is, the control task is then not necessary for ruling out spontaneous recovery but mainly for eliminating a charm effect. However, in the rehabilitation stage, spontaneous recovery is assumed to be occurring much more, and therefore should be ruled out in order to

(26)

establish a treatment effect. Given this, the control task design seems suitable for this stage. However, one major drawback of this approach in the rehabilitation stage is that selecting a control task that assesses a process that recovers at the same rate as the targeted process is extremely difficult. On top of that, given that the control task should neither be at ceiling level nor at floor level, this reduces the ‘options’ even more. Therefore, the selection of a suitable control task could be difficult and rather time-consuming. However, when there is actually an appropriate control task, the control task design is particularly suitable in the rehabilitation stage, given the ability to rule out spontaneous recovery, which is the case in a stage in which the PWA is being subjected to such recovery. In terms of the two different approaches within the MST, i.e. stimulation therapy and strategy training, the control task design seems more suitable for measuring the effect of strategy training than for stimulation therapy. By way of explanation, with stimulation therapy, the SLP cues the PWA by appealing for several language modules. This is in contrast to strategy training, with which it is possible to teach a strategy concerning only one module. Therefore, choosing the right control task within stimulation therapy is difficult or even impossible, as the control task should then cover several modules that are involved in a specific language skill.

4.3. Item-Specific Design

Prior to therapy, two sets of items are assessed with an item-specific design. After this, one-half of the items will be treated, while the other one-half will remain untreated. Then, if treatment has worked, the treated items will improve, while the untreated items will not change

(Franklin, 1997). However, while the design is easy to use and also a well-controlled one, Franklin (1997) argues that in most cases, improvement on also untrained items is desired in order to establish a treatment to be functional for the PWA. Therefore, this design does not seem suitable for most studies concerning aphasia. Studies for which the design might be appropriate are studies concerning dysgraphia and acquired dyslexia where the effectiveness of treatment in relation to re-learning phoneme-grapheme/grapheme-phoneme conversion rules are investigated, as was done in Raymer et al. (2003). That is, in these disorders, the application of several spelling conventions and the recognition of letters respectively are lost (Bastiaanse, 2011) and should be learned separately. Given this, by assuming that the re-learning of one conversion rule does not lead to the mastering of another conversion rule, the item-specific design might be more suitable to investigate whether a treatment is effective for re-learning such rules. Moreover, in terms of strategy training within the MST, generalization

(27)

to other, untrained items is expected. By way of explanation, when the PWA is taught to cue himself it is both desired and expected that the PWA can apply this strategy also to untrained items, making the item-specific design even more unsuitable for measuring a therapy effect of this approach. Moreover, the design will not be suitable for the MST as within this

framework, several treatment methods are often used interchangeably, which makes it difficult to keep track on which items have been treated and which have not been treated. On top of that, within the STAPP, it is not possible to distinguish treated from untreated items in an easy way, making the design not suitable for the upcoming effectiveness research. All things considered, except for specific cases in which no generalization to other (untrained)

items is expected or specific spelling conventions or recognition of letters are re-learned, the

item-specific design will not be suitable for determining therapy effectiveness, regardless of the stage a PWA is in.

4.4. Multiple Baseline (Across Behaviors) Design

In both the multiple baseline design and the multiple baseline across behaviors design, an extended baseline is gained before commencing therapy in order to establish a stable performance on a specific language process. Specifically, the former design allows to determine the effect of treatment that is focused on a specific process or module, by

monitoring that and other untreated, independent processes. More importantly, the latter, i.e. the multiple baseline across behaviors design, allows to determine the effect of different treatment approaches focused on independent different processes/modules, which comes in handy when measuring the effectiveness of the MST, which involves module-specific

therapy. Nevertheless, while the multiple baseline across behaviors design allows to deal with spontaneous recovery by obtaining a stable baseline prior to the therapy, the design is rather time-consuming, given the several repeated pre-therapy tests that should be administered first. Furthermore, the idea behind this kind of design is “that spontaneous recovery does not occur in spurts”, but in a more gradual line (Franklin, 1997, p. 409). However, as argued by Francis (1997), little is known about the details of recovery within PWAs. In other words, it cannot be simply assumed that spontaneous recovery occurs gradually. Moreover, since several tests are administered before therapy commences, there is also danger of a so-called ‘practice effect’, instead of an effect due to spontaneous recovery. That is, the PWA might improve without therapy as he or she becomes familiar with the tests that are administrated during the baseline measurements. Likewise, the design does not allow to distinguish between a therapy specific

(28)

effect and a charm effect (Franklin, 1997).

In light of the rehabilitation stage, the multiple baseline across behaviors design does not seem to be suitable for this stage in terms of ruling out spontaneous recovery, when considering Francis et al. (2002). That is, although Francis et al. (2002) obtained an extended baseline of a PWA who was still in the rehabilitation stage, it was indicated that the design they used was somewhat weak to establish a therapy specific effect in this stage. That is, it was expected that MB’s anomia might be the most benefiting from spontaneous recovery as MB was spending much time on trying to speak every day, apart from the therapy she was receiving. However, this was not found, as instead, other processes were found to be

improved, which probably benefited more from spontaneous recovery (Francis et al., 2002). In other words, these speculations and uncertainties indicate that distinguishing between spontaneous recovery and specific treatment effects is extremely difficult in the rehabilitation stage. That is, even when baseline measurements are obtained, it is still hard to establish a therapy effect. In addition, Francis et al. (2002) argue that, in order to prove therapy specific effects of CIN and with that its usefulness, it is necessary to study PWAs with more chronic anomia as well. This in turn implies that in this stage, the multiple baseline design would be more effective in comparison to the rehabilitation stage, as more stable baseline

measurements can be obtained in the former.

Furthermore, according to Grayson et al. (1997) in contrast to the chronic stage, in the subacute stage, i.e. “the early days post-CVA”, there are many extraneous factors that should be considered. That is, the family’s acceptance or adjustment to the new situation and

condition are of great significance during this period, resulting in that it does not always seem possible “to achieve strong baselines and tight controls in the clinical setting” (Grayson et al., 1997, p.274). This claim might also apply to the rehabilitation stage, as in this stage, language therapy is crucial in order to gain the most possible recovery. Therefore, achieving strong baselines and tight controls might stand in the way of therapy that must be given in this critical period. Moreover, according to Franklin (1997), to get a significant effect through multiple testing events, a large number of items may be necessary. This in turn might make the design less attractive for clinical use (Franklin, 1997). In addition to that, Francis et al. (2002) emphasizes that “obtaining several baselines to demonstrate the stability of the disorder takes up valuable time in wards where PWAs are discharged rapidly, and the therapist is unable to follow them up” (p. 256). In other words, it is not always possible to

(29)

start with collecting baseline data, given the little time that therapists have for each PWA in a clinical setting, e.g. a rehabilitation center.

Moreover, periods of no treatment are extremely difficult to justify in this stage, which is the case when obtaining baselines (Francis et al., 2002). Moreover, with a multiple baseline design the PWA receives treatment for one particular process in fixed blocks during, for example, five or six weeks. These long blocks are necessary so that significant improvement can be inferred from a significant difference between the baseline measurements and

performance during and after treatment. This in turn means that other processes are not trained during a specific block. Based on clinical experience, this might not be ethically responsible since in the rehabilitation stage a PWA often prefers to train several different language skills as the PWA would like to improve, for example, both speaking and writing at once. In the chronic stage this might not be the case, since in this stage the PWA already received much treatment and might therefore be less vulnerable to receiving treatment in blocks focused on only one particular process/module.

Moreover, in light of measuring the effectiveness of the MST at Beatrixoord, the other two variants of the multiple baseline design, i.e. the multiple baseline across

settings/situations design and the multiple baseline across subjects design are not suitable as well. This is due to the fact that the effectiveness of the MST will only be examined in PWAs rehabilitating at Beatrixoord for now, therefore, other settings/rehabilitation centers will not be included in the research. On top of that, recruiting PWAs with the same language problems will not be possible, due to the heterogenous aspect of aphasia, making also the multiple baseline across subjects design unsuitable for the effectiveness research.

Altogether, although the multiple baseline across behaviors design allows for studying the effect of module-specific therapy perfectly (e.g. MST), as it is possible to apply treatments focused on different modules consecutively, the design does not seem suitable for the

rehabilitation stage. That is, in the rehabilitation stage, obtaining baseline measurements is not always possible due to the need of early intervention and the little time that is available for a PWA in a rehabilitation center. Furthermore, a PWA in this stage is more likely to prefer training several processes at once or in quick succession instead of in blocks covering a few weeks. On top of that, gaining a stable baseline is extremely difficult in this stage given that spontaneous recovery is not assumed to occur gradually. Lastly, these disadvantages of the design regarding the rehabilitation stage apply to both approaches within the MST, i.e. either stimulation therapy or strategy training. By way of explanation, for both approaches the

(30)

extended baseline and the long blocks of therapy will be necessary when using the multiple baseline across behaviors design. Seen in this light, there is no difference between the two approaches in terms of the applicability of the design. In other words, the multiple baseline across behaviors design is unsuitable for measuring therapy effectiveness of the MST, regardless of the chosen approach.

4.5. Cross-Over Design

Grayson et al. (1997) did not use a multiple baseline design, due to the need of early

intervention. Instead, the effectiveness of therapy was measured by using a cross-over design, as described earlier. A great advantage of this design is that the assessment of one language process functions as control task for the other processes, and vice versa. In other words, when treatment focused on a specific process is given, then it is expected that the other processes do not improve. In this way, it is possible to differ between a treatment specific effect and an effect caused by spontaneous recovery (Franklin, 1997; Lum, 2005). On top of that, recovery rates of individual processes can be compared with themselves to indicate improvement. In this way, the cross-over design deals with the impossibility of achieving a stable baseline in contrast to the multiple baseline (across behaviors) design (Grayson et al., 1997). Still, it is possible to focus on different aspects of language processing by giving the PWA treatment that is focused on different processes/modules, similar to the multiple baseline design across behaviors. Likewise, the processes/modules should be functionally independent to retain experimental control. Furthermore, by employing the cross-over design it is easier to explain the reasons and aims for therapy to both the PWA and his relatives. On top of that, progress can be tracked precisely by conducting tests at the start and the end of each therapy phase, which in turn makes the process concrete to show the PWA, according to Grayson et al. (1997). Seen in the light of the rehabilitation stage, this aspect of the cross-over design is beneficial, given that the PWA is in a stage that involves both emotional stress and physical exhaustion as the PWA has to get used to his or her new situation. Moreover, this design is suitable as no baseline measurements are necessary in order to distinguish treatment specific effects from spontaneous recovery. Altogether, these are all advantages that make the design suitable for an effectiveness study in the rehabilitation stage, not only to demonstrate therapy specific effects, but also in terms of ethical aspects. Additionally, the cross-over design is suitable for measuring therapy effectiveness of both stimulation therapy and strategy training of the MST, since it allows to accurately track the PWA’s improvement by administering

Measuring Therapy Effectiveness in the Rehabilitation Stage of Aphasia