LUNA treatment

This paper describes a novel discourse treatment for aphasia called, Language Underpins Narrative in Aphasia (LUNA). LUNA is a manualised, theory-based [21], codesigned [22] multi-level discourse intervention, which aims to facilitate the telling of personal stories through word, utterance (sentence) and discourse macrostructure level activities. It integrates familiar treatments—semantic feature analysis, mapping therapy, story grammar—to provide flexible metalinguistic tools for improving people’s confidence and ability to express themselves through narrative. These treatment activities are integrated through the use of a personally chosen story on which to work during treatment. LUNA is distinct from previous multi-level treatments in its form of personalisation (the focus on a story that the person has selected and wants to tell to family and friends); its explicit focus on meta-linguistic awareness (activities are aimed understanding the person’s own language profile); and meta-cognitive awareness to support self-management.

LUNA is personalised in two ways. Firstly, there is personalisation in the subject material. Participants choose stories from their own lives that they want to share with others. Secondly, there is personalisation in the linguistic content. The participant chooses the words, sentences, and macrostructure they use to tell their story in collaboration with the therapist during treatment. There is evidence that therapy outcomes are enhanced when personalised content is included [23], and that this stimulates neural re-organisation [24]. In addition, the treatment of personal stories can have broader effects. The sharing of stories may help people to express themselves, and to interact and share more with family and friends [25–27]. LUNA is a meta-cognitive [28] and meta-linguistic [29] therapy, encouraging participants to reflect on their own thinking and language; to learn about the nature of language, and the detail of their own linguistic skills and impairments; and to practise using the new skills in everyday contexts. Ultimately, this means the use of personal stories may serve to increase motivation to engage with and complete the treatment programme—of relevance to discussions of feasibility and adherence described later in the paper.

LUNA was initially devised for face-to-face delivery. However, this study coincided with the 2020 COVID-19 pandemic. All assessment and treatment procedures were therefore adapted for remote delivery using videoconferencing technology, specifically Zoom. Research has demonstrated that people with aphasia can comply with remote assessment and treatment and find such procedures acceptable [30] and that remote treatment can have positive outcomes [31–33]. Remote delivery of multi-level discourse treatment for aphasia has not previously been trialled.

This proof-of-concept study comprised a phase II randomised controlled trial, comparing remote LUNA treatment with a waitlist control. It aimed to test the feasibility of trial procedures and explore indicative outcomes from LUNA treatment. Specifically, this study aimed to:

1. Test the feasibility of a definitive trial comparing remote LUNA with a waitlist control, using the following feasibility endpoints: a) participant recruitment and retention rates; b) adherence to treatment sessions; c) counts of missing data; and d) fidelity scores for treatment delivery.
2. Explore the appropriateness of the trial outcome measures, as indicated by the level of variability of scores, missing data, and floor and ceiling effects.
3. Investigate preliminary efficacy by comparing outcomes on discourse, language, and measures of psychosocial state across participants who have and have not received the LUNA intervention.

Trial design

The study was a single-blind, waitlist, randomised, controlled, phase II, proof-of-concept, feasibility and acceptability trial of remote LUNA for people with chronic post-stroke aphasia.

This study was granted ethical approval by the City, University of London, School of Health Sciences Research Ethics Committee (ETH1920-0210) in February 2020; similarly, approval was granted in June 2020 for amendments (ETH1920-1651) following the COVID-19 national lockdowns, prior to the trial starting recruitment. The trial sponsor was City, University of London.

Participants

Twenty-eight participants were recruited to the remote LUNA study between 16/06/2020 and 06/08/2020. Twenty-eight was an intentional over-recruitment on a target of at least 24 participants, to mitigate for possible attrition. The intention was to ensure a sample size of 24 (12 treated, 12 control) following recommendations for feasibility trial sample sizes [34, 35]. Inclusion criteria were adults (18+ years); diagnosis of ischaemic or haemorrhagic stroke; and aphasia due to a stroke that occurred at least 12 months prior to recruitment. Additionally, participants were literate and fluent users of English prior to their stroke (self-reported), with adequate hearing and vision with aids or glasses (for example to see pictorial and written assessment and treatment materials). Participants were required to have access to a computer or tablet and an internet connection. They needed to be able to download and access Zoom, either independently or with the support of a friend/neighbour/family member.

Participants were excluded if they were receiving speech and language therapy elsewhere or participating in any other aphasia treatment research project for the duration of the study. Usual stroke supports, such as voluntary sector support groups, could proceed. Although many of these support services were curtailed due to COVID-19, some moved online. Participants with severe aphasia, as defined as a score of 7 or less on the Frenchay Aphasia Screening Test (FAST) [36], were excluded. This criterion was applied because remote LUNA was designed for people with some verbal output. It was also judged that people with severe aphasia would struggle to manage remote delivery. Participants were also excluded if they had a secondary cognitive diagnosis such as dementia. This was established via self-report and/or the confirmation of the referring group co-ordinator and/or by expert clinical judgment of research project staff. Screening and recruitment were completed by experienced SLT members of the research team (authors KS and MC).

Participants were a volunteer sample recruited by advertising the study through UK-based stroke support groups, signposting people to the dedicated project website or self-referral. Self-referrals were accepted from anywhere in the United Kingdom. All recruitment, assessment, treatment, and interview sessions were conducted online using Zoom. All participants gave written consent. All participant information sheets and consent forms were made accessible to people with aphasia following evidence-based recommendations [37]. Recruitment began on 16 June 2020 (first screening) and data collection finished on 28 April 2021 (final assessment).

Intervention

The LUNA treatment is specified in the TIDIER checklist (see S1 Checklist). Before treatment started, participants were supported to choose two personal narrative monologues to share. They were given about a week to consider their choice and then both narratives were elicited at the beginning of the first assessment session, under controlled conditions, following a set procedure. Participants then decided which of the two narratives they wished to work on in treatment sessions. This choice was shared with therapists, and the chosen narrative was transcribed, analysed and deconstructed to identify potential treatment targets, ahead of the first treatment session.

Remote LUNA comprised 20 hours of treatment, 2 sessions per week of 60 minutes each, for 10 weeks. A set-up week preceded treatment, where the SLT and participant met for an hour to agree on goals–the deconstructed narrative was used as a basis for this discussion. This resulted in an intervention lasting 11 weeks, consisting of 21 hours of treatment in total. In week 2–11, the chosen personal narrative was progressively re-built through integrated word, phrase, clause, multi-clause, and discourse macrostructure treatment activities. Treatment targeted three language levels: word (wks 1–4); utterance (weeks 5–7); and macrostructure (weeks 8–10).

All sessions were delivered over Zoom. One session per week was delivered by a qualified SLT and the other session was delivered by an assistant: a student SLT (SSLT). Both the SLT and SSLT followed the treatment manual and received guidance via remote supervision. Linked ‘challenge tasks’ promoted generalisation outside of treatment sessions.

A team of three experienced SLTs and twelve SSLTs delivered remote LUNA to the 28 participants. Alongside guidance from the treatment manual, SLTs received six days of remote training across a three-week period prior to implementing treatment, in addition to weekly remote group supervision from a clinical linguist (author LD) throughout the trial. SSLTs received fourteen hours of remote training and received a mixture of 1:1, paired and group supervisions remotely throughout the trial. Each participant worked with the same SLT and SSLT for treatment for the duration of the study (there were different screening and Assessor SLTs for recruitment and assessment–see below).

Feasibility outcomes

Feasibility of remote LUNA was tested in terms of participant recruitment and retention, adherence throughout the study, and missing data. To inform a future trial, treatment fidelity, appropriateness of outcome measures and estimated sample size were also explored. Six feasibility endpoints were outlined:

1. Feasibility of recruitment and retention to the trial: Data comprised of counts/ proportion of those who expressed interest, were screened and deemed eligible, those who consented, attrition and reasons for attrition if known.
2. Adherence: data comprised of number/proportion of treatment sessions attended as scheduled, and percentage completion of the LUNA treatment programme; reasons for non-attendance.
3. Missing Data: Data comprised of attrition rates and counts of other missing data.
4. Assessment of treatment fidelity through ratings of provider adherence to the LUNA manual/essential elements; reliability of the rating procedure was checked and whether scores were affected by the treatment provider, treatment level or group allocation.
5. Appropriateness of outcome measures: indicated by the level of variability of scores, missing data, and floor and ceiling effects.
6. Estimate of sample size for a future trial: based on a preliminary power calculation using WAB-R effect sizes.

Treatment fidelity

Remote LUNA yielded a total of 560 hours of Zoom-recorded treatment sessions (28 participants x 20 sessions), and a sample of 10% of sessions (56hrs) was selected for review. Sample selection was stratified (by authors LD and MC)–it was organised to ensure that a range of providers and treatment levels were sampled but was otherwise random (i.e. done without reference to session content or participant details). Treatment fidelity (TF) was assessed by evaluating providers’ adherence to the treatment manual (as determined by SLT student raters) using a TF checklist of essential elements of LUNA.

The checklist was developed iteratively with the research team, co-designers with aphasia, co-designer SLTs, LUNA therapists, and research students. The final checklist (Table 1) comprised 12 items. These were used by all treatment providers, during the treatment phase, as a self-reflective tool after completing sessions. The same checklist was then used post-treatment phase by SLT student raters to evaluate providers’ adherence to the treatment manual. Two of these students were not part of the team that delivered intervention, and two were. The latter two students did not evaluate their own sessions and so all four students were considered unbiased raters.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. LUNA treatment fidelity checklist items.

https://doi.org/10.1371/journal.pone.0304385.t001

In the post-treatment phase, SLT student raters evaluated fidelity by marking each of the 12 items as either present or absent [38] including additional qualitative notes to justify their decisions. Fifty-six hours (10%) of treatment sessions were viewed by four research students. These raters received training (4 hours) which comprised group and independent viewing and discussion, and independent benchmarking. Training was carried out on six representative sessions, selected to include: word, sentence, discourse macrostructure treatment activities; and delivery by SLTs and SSLTs. Percent agreement on benchmarked sessions was 72% (26/36 items) with most discrepancies on items 8, 10 and 11. These were discussed, with refinements added to the checklist.

Fourteen sessions were allocated to each of the four raters (total 56 sessions) and assessed independently. Eight (8) of 56 sessions were subjected to intra-rater reliability checks with ratings separated by a period of at least 1 month, and a further eight (8) of 56 sessions subjected to inter-rater reliability checks. Reliability was determined by calculating percentage agreement with agreement interpreted as high if >70% [39].

Clinical outcome measures

Participants completed assessments at three time points: T1 (weeks 1 & 2), T2 (weeks 13 & 14) and T3 (weeks 25 & 26). Only efficacy outcomes at T2 are reported here, to enable a comparison of treated (i.e. Immediate treatment group) and untreated (i.e Delayed treatment group) participants. Participants in the Immediate group received LUNA treatment between T1 and T2. Participants in the Delayed group received treatment between T2 and T3, but their efficacy results are not reported in this paper. Participants were recruited to the study in two waves to allow for appropriate staffing.

Feasibility findings from all three timepoints are presented for completeness. For the preliminary efficacy evaluation, we report clinical outcomes from T1 and T2 only, comparing the experimental (Immediate) group who had received treatment at this point to the control (Delayed) group who had not yet received treatment.

At each timepoint, assessment was completed by LUNA Assessors (n = 2) who were qualified SLTs who were kept blinded to participant treatment group allocation throughout the study. Assessment processes were adapted for online delivery and manualised. Assessors undertook this development work across a 6-week period prior to assessing participants, also using this time to undertake training and practice remote assessment with 3 people with aphasia who were part of the LUNA PPI Advisory group. For the discourse analysis, Assessors were given weekly training over a two-month period (including training with a clinical linguist and self-directed exercises). In addition, they received regular supervision from an SLT (author MC) during the assessment phases and from a clinical linguist (author LD) during the narrative assessment phase.

Personal narratives measure (LUNA discourse protocol)

Participants produced two personal narratives at each assessment point, which were recorded, transcribed, and analysed by according to the LUNA Research Discourse Analysis Protocol. Several discourse metrics were calculated from the analysis (see S1 Appendix), with the selection made during codesign session with the SLTs and guided by: use of a measure in the systematic review [13], the psychometric properties [40, 41] of the measures, and the appropriateness of the measure for measuring change after LUNA treatment. A novel measure ‘narrative words’ was designed by the research team which, while similar to Correct Information Units (CIUs), was intended to be more clinically feasible as an analysis. Number of narrative words was proposed as the primary clinical outcome measures. Other discourse metrics included: number of CIUs, percentage of CIUs, and number of CIUs/minute (following the protocol of Nicholas and Brookshire, 1993); number of narrative words, percentage of narrative words, and number of narrative words/minute; number of complete utterances and percentage of complete utterances; number of multiclause utterances and percentage of multi-clause utterances; predicate argument structure (PAS) score; a Story Grammar score; and a count of the number of clear reference chains (see S1 Appendix).

The Western Aphasia Battery-Revised (WAB-R) [42]

The WAB-R is a performance-based outcome measure assessing speaking, auditory comprehension, naming, and repetition across four sections. It classifies aphasia type and generates an aphasia severity score between 0–100, the Aphasia Quotient (AQ), where a score of 0–25 is very severe, 26–50 is severe, 51–75 is moderate, 76+ is mild. A cut-off score 93.8 and above is considered "normal or nonaphasic" (pg. 91, [42]). The AQ score was used in the analysis. It was standardised on people with aphasia (n = 150) and controls (n = 59) [43]. Internal consistency and interrater reliability are good [44]. It is internationally used as part of the core outcome set for aphasia trials [12] and has been validated for remote online delivery [45].

The Communicative Participation Item Bank (CPIB)–General short form [46]

The CPIB is a 10-item patient-reported outcome measure (PROM). Patients rate the level of interference caused by their condition for each item, on a 4-point scale. Items ask, for example, how much the condition interferes with communicating with people known to the person with aphasia, with people not known to them, when giving someone detailed information, and when communicating as part of a small group. Scores are converted to a summary score which ranges between 0–30 where a high score is favourable, representing little interference from the health condition. The summary score was used in the analysis. The measure was designed for community-dwelling adults with spasmodic dysphonia but was adapted for aphasia with a representative sample. The short form is appropriate and valid for people with aphasia [47].

The Communication Confidence Rating Scale for Aphasia (CCRSA) [48, 49]

The CCRSA is a 10-item PROM. Patients rate their confidence in communicating in different contexts on a scale of 0–100. Scores are converted to derive a total score of between 10–40, where 40 represents feeling very confident in communicating. The total was used in the analysis. It is the only communication confidence measure in the field and is increasingly used in treatment studies. It was validated on 47 people with aphasia from different treatment settings. The psychometric properties of sensitivity to change and reliability (inter- and intra-) remain to be established [49].

The Assessment for Living with Aphasia (ALA) [50]

The ALA is a 45-item pictographic PROM assessing aphasia-related quality of life and was developed by an internationally leading aphasia charity in Canada. Questions cover four domains relating to living with aphasia (language impairment, participation, personal factors, and environmental factors) and there is a summary question relating to the overall impact of aphasia. The total scores of these 5 items are divided by 37 to create a single mean score. The mean score ranges between 0–4, where 4 represents a perception of good quality of life. The mean score was used in the analysis. Acceptable construct validity and reliability have been established [51].

Visual Analogue Mood Scales (VAMS)–Sad [52]

Following feedback from LUNA advisors with aphasia (and supported by the research team), a single item mood measure, the Visual Analogue Mood Scales (VAMS) Sad scale was added to all testing time points. Scores range between 0–100, with 100 representing a maximal level of sadness and zero representing a minimal level (or absence) of that mood. It has been used successfully in aphasia studies [53, 54] and takes three minutes to complete. It is accessible and appropriate to be used with stroke survivors who have aphasia [55]. Content validity [52, 56] and test re-test reliability [57] have been established.

Randomisation

Stratified random sampling was used. After T1, participants were classified into two groups: group (i) ‘mild’ and group (ii) ‘moderate’ aphasia severity based on WAB AQ score. Randomisation was carried out by a research team member (author NB) who was blinded to severity by use of the group labels (i) and (ii), and who was also blinded to screening and assessment results and had no knowledge of the participants. Participants were randomised to the immediate or delayed condition by the following method: for each group (i/ii, i.e. mild/moderate), participant numbers were written on identical pieces of paper which were then folded in half; these were placed in a box and shaken, then pulled out in a random order; in alternating fashion, each number was allocated to the Immediate group or the Delayed group.

Blinding

Limited members of the research team were aware of participant treatment group allocation (Immediate/Delayed). These members were the joint principal investigators, project manager, treating SLTs and SSLTs, and the qualitative researcher. Other members of the research team (n = 6) were kept blinded to group allocation. This included, most importantly, the Assessors (n = 2) who were qualified SLTs kept blinded to group allocation throughout the study–this included them agreeing to delete their social media accounts for the duration of the trial in order to remove the risk of unblinding in that context.

Recruitment sessions were organised by the project manager and one principal investigator, and assessors had no access to participant files or details that would reveal group allocation. Remote working, imposed by COVID-19, also ensured that assessors had minimal contact with the unblinded members of the research team, beyond formal supervision with the Principal Investigators. Participants were instructed not to reveal their group allocation to assessors during assessment sessions. A log was kept of any instances of unblinding and near misses, with the reason for the unblinding.

Analyses

Regarding feasibility, analyses were descriptive to ascertain feasibility endpoints such as recruitment and attrition. Adherence, in terms of sessions delivered as scheduled and participants’ completion of the treatment programme, was recorded as a percentage of sessions. With respect to treatment fidelity, a score was calculated for each item as a percentage of items marked as present and interpreted as high if 80–100% and low if 50% or lower (scores 51%-79% being medium) [58]. Fidelity findings were also examined in relation to treatment provider (SLT/SSLT), treatment level (word, sentence, discourse) and group (immediate, delayed).

Regarding clinical outcomes, a between-group comparison analysis was carried out. ANCOVA compared assessment scores of both groups at T2 (when the Immediate group had received treatment, but the Delayed group had not) in measures of discourse, language, and psychosocial state, using T1 scores as a covariate. These analyses were exploratory, examining whether the treatment showed promise of efficacy. An indication of treatment promise would be seen in a significant group effect favouring the immediate condition and/or effect size (partial eta squared: η2 ~ 0.01 = small effect; η2 ~ 0.06 = medium effect; η2 ~ 0.14 = large effect). Preliminary power calculations were conducted based on the effect sizes of the standardised language measure (WAB-R) to determine sample size for a future clinical efficacy trial of LUNA.

Participants

Twenty-eight (28) people with aphasia were recruited to the trial in a two-month period between 16 June 2020 and 6 August 2020. Fifty-eight (58) people expressed an interest, 40 people were screened using the FAST, and 28 were randomised (Fig 1).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Participant flow diagram.

https://doi.org/10.1371/journal.pone.0304385.g001

Participants were on average ~60 years old, ranging from 34–83 years (See Table 2). They were predominantly from a White British ethnic group, university educated, and had held highly skilled positions in their working lives as measured by the Standard Occupational Classification [59]. All participants had English as their primary language with more than half the sample using more than one language but only three participants described an advanced ability in other languages. Participants came from two of the four UK countries and from seven of the nine regions in England, representing a large geographical spread. There were no participants from Wales and Northern Ireland, or from the West Midlands or the North East of England. See Table 2 for participant characteristics. Participants were on average 55 months post-stroke (range 14–181 months) and largely balanced between mild and moderate aphasia severity.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Participant characteristics at baseline (T1).

https://doi.org/10.1371/journal.pone.0304385.t002

Feasibility outcomes

a) Participant recruitment and retention.

The remote LUNA study recruited 28 participants, which was 100% (28/28) of the target sample size. In brief, 48% (28/58) of those who expressed an interest, and 85% (28/33) of those who were eligible consented to participate in the trial (Table 3). See Fig 1 for further detail about reasons for exclusion at each stage. Four participants withdrew from the study due to ill health: three from the Immediate group following T2 testing, and one from the Delayed group after treatment but before T3 testing. Therefore, retention was 86% (24/28).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Participant recruitment and retention.

https://doi.org/10.1371/journal.pone.0304385.t003

Clinical outcomes

Clinical outcomes were measures of discourse from personal narratives. Descriptive statistics are presented for the discourse measures in Table 4 and for the measures of language and psychological state in Table 5. At T1 (pre-treatment) there were no significant differences between groups (all p values > 0.3).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. Means and standard deviations for discourse measures.

https://doi.org/10.1371/journal.pone.0304385.t004

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 5. Means and standard deviations for measures of language and psychological state.

https://doi.org/10.1371/journal.pone.0304385.t005

Preliminary efficacy data.

Due to the feasibility design of this study, it was intentionally underpowered for definitive efficacy testing. However, clinical outcomes were analysed to investigate preliminary efficacy using ANCOVAs to ascertain differences between Immediate and Delayed groups for each outcome measure at Time 2, controlling for Time 1 [60]. The results indicate that LUNA shows preliminary efficacy with 50% of measures (9/18) showing medium or large effect sizes (bolded in Table 6) for group differences at Time 2 once Time 1 was controlled for. Medium effect sizes were noted for all levels of discourse (number of narrative words, CIUs, complete and multi-clause utterances, clear reference chains), language (WAB-R AQ), and psychosocial state (VAMS). Large effect sizes were noted for one discourse level—% complete and % multi-clause utterances–and were also significantly different even with low power, indicating a proportionate increase in these narrative structures.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 6. Between group differences with effect sizes for each measure at T2.

https://doi.org/10.1371/journal.pone.0304385.t006

We additionally ran a non-parametric Wilcoxon on all the measures that the ANCOVA showed as having a **large effect size. We found that all narrative variables improve for the active group, and none for the control group. For WAB and VAMS the Wilcoxon results are not significant for either group (as per the ANCOVA). Note that the parametric effect sizes (as shown in the table) are needed here because calculating effect size cannot be reliably done for non-parametric analysis. Correlational analysis was additionally undertaken to explore which factors were associated with optimum response to the LUNA treatment, but there were no convincing patterns of predictors that would inform future studies or practice. See S1 File, for the detail.

Safety

Adverse Events were logged and are reported by participant. Four participants (4/28, 14%) had a new health event. Two participants broke bones, one participant’s health deteriorated, and one participant had a further haemorrhage, a known risk factor within the stroke population [62]. These were unrelated to trial activity. Reports of distress were recorded by session and nine episodes were recorded across the 308 sessions in the trial (9/308, 3%). Episodes of distress were connected to the activities of the trial e.g., a participant became upset when asked to reflect on the impact of aphasia on their lives in the ALA assessment, and one episode was due to distress that the trial was finishing. Episodes were managed in accordance with an established protocol, and in discussion with the project manager.

Limitations

Some limitations are noted. Firstly, the sample recruited to this remote feasibility trial is not typical of the wider stroke and aphasia population and future studies should aim to recruit a more representative sample. With a mean age of 60 years, this sample was younger than both a national sample, a mean age of 78 years [80] and a London sample of 68.9 years [81]. Additionally, both London and national samples have more ethnic diversity reporting 56% and 95.7% white participants respectively, compared to the 100% white sample in the remote LUNA study [81, 82]. Secondly, measuring change in spoken discourse is a challenging undertaking, as there are numerous metrics used in the research field and their psychometric properties are generally not well established [11, 83]. To address this problem, this study employed: (1) traditional discourse metrics used in many research studies e.g., number of CIUs; (2) discourse metrics with proven psychometric properties of reliability and validity [40, 41, 83]; and (3) a novel word-level metric of narrative words intended to act as a comparator for CIUs to explore the possibility that it would be more clinically feasible. Further analysis not reported here does not support the notion that the narrative words measure is a straightforward alternative to CIUs, and further research is needed with any novel measures subjected to traditional psychometric testing. Thirdly, the LUNA Discourse Analysis Protocol was created for this study and has some, not insignificant, assessor burden with analysis of each narrative at each time point taking approximately three hours. However, this represents the time for the research version of the LUNA discourse analysis protocol and the intention is to reduce this protocol in the future for clinical implementation. Finally, most of the clinical outcome measures were not validated for online delivery, except for WAB-R which has demonstrated equivalence [84], but differences in outcomes between face-to-face and online delivery of the Boston Naming Test demonstrate this cannot be assumed [85].

Future implications

This study’s findings meet the set criteria for progression to definitive trial testing, in the context of remote treatment delivery. LUNA was co-designed as a face-to-face intervention but delivered online due to the COVID-19 pandemic, and the positive feasibility findings presented here are of remote LUNA. Future studies could consider a similar study of face-to-face delivery, compare face-to-face with remote delivery, or co-design a hybrid delivery model.

LUNA appears to have potential clinical value because of its multi-level language focus, personalised narrative approach, and emphasis on metalinguistic and metacognitive skills which translate well towards self-management during and following treatment. The original co-design of LUNA with providers and recipients of SLT [22] also strengthens LUNA’s applicability and relevance to the treatment of people with chronic aphasia in UK clinical settings. This approach serves as a good example for the development of further interventions seeking to embed co-design, salience, and authentic, functional language change.