An evaluation of the validity of an aphasia friendly mood and anxiety measure for stroke patients

ABSTRACT

Introduction

People with post-stroke aphasia are at increased risk of experiencing depression and anxiety. Despite this, there is a lack of appropriate tools which can be used in routine stroke rehabilitation practice to support with the identification of patients experiencing these difficulties. The Stroke Aphasia Anxiety and Mood Screen (SAAMS) was developed by an interdisciplinary team of stroke rehabilitation professionals to improve mood and anxiety screening. The current study sought to evaluate the validity of this measure.

Methods

Twenty-one people undergoing in-patient stroke rehabilitation who experienced aphasia were administered the SAAMS, alongside the Stroke Aphasic Depression Questionnaire (SADQ-H10) and the Behavioural Outcomes of Anxiety (BOA). The inclusion criteria included patients whose aphasia prevented them from completing traditional self-report measures of mood and anxiety, but who demonstrated a reliable ability to correctly respond to yes/no closed questions in Speech and Language Therapy assessments.

Results

Both the SAAMS-anxiety (SAAMS-A) and depression (SAAMS-D) subscales achieved adequate internal consistency, and patients’ scores were correlated with observational measures of mood and anxiety. The SAAMS achieved fair classification accuracy for identifying clinical depression and anxiety.

Conclusions

Patients with aphasia provided reliable responses to questions about anxiety and depression, and this is of clinical importance. While the study identified that the SAAMS has reasonable classification accuracy, further research is required to validate the SAAMS more thoroughly before it can be used routinely in clinical practice.

KEYWORDS:

• Stroke
• aphasia
• depression
• anxiety
• screening

Introduction

Stroke patients are at increased risk of experiencing clinical anxiety and depression. Meta-analytic research has identified that 18.7% of stroke patients experience clinical anxiety (Knapp et al., 2020), 18 to 33% experience post-stroke depression (Medeiros et al., 2020) and the risk of suicide is approximately double that observed in the general population (Teasdale & Engberg, 2001). Clinical anxiety and depression are associated with lower quality of life, reduced social functioning, and negative impacts on the effectiveness of stroke rehabilitation (de Graaf et al., 2022; Donnellan et al., 2010; Paolucci et al., 2019). Similarly, a systematic review by Hilari et al. (2012) found that aphasia severity, emotional distress, communication and activity limitations, as well as medical problems and social factors, all affect the quality of life in people with aphasia. Meta-analytic research has confirmed that standard psychological and, to some degree, psychiatric interventions are effective at treating these difficulties (Ahrens et al., 2023; Medeiros et al., 2020; Wang et al., 2018). Therefore, the early identification and treatment of post stroke depression and anxiety is essential for clinical practice.

Screening for post-stroke depression and anxiety is a core aspect of stroke rehabilitation, with The National Clinical Guidelines for Stroke in the United Kingdom and Ireland (2023) explicitly recommending that all patients “should be routinely screened for anxiety and depression using standardised tools”. Measures for post stroke depression and anxiety screening should be based on an up-to-date construct, i.e., clinical depression consisting of emotional, cognitive, and physiological symptoms outlined in the Diagnostic and Statistical Manual of Mental Disorders (DSM-V); should avoid confounding variables like changes in appetite, energy levels or fatigue, all of which are commonly the result of post-stroke sequalae (Alghamdi et al., 2021; Taylor-Rowan et al., 2019); should have appropriate clinical usability (Burton & Tyson, 2015); and should include a measure of suicide and risk. The Patient Health Questionnaire (PHQ-9; Kroenke et al., 2001) is a widely used screening test for depression, with good clinical usability, which has been widely validated in stroke (Burton & Tyson, 2015). Measures of post-stroke anxiety have received comparably less research, although the Generalised Anxiety Disorders two item (GAD-2) Scale and Geriatric Anxiety Inventory (GAI) have single validation studies supporting their use (Kneebone et al., 2016; McCrory et al., 2022).

While measures such as PHQ-9 and GAD-2 are useful tools to assess for post-stroke anxiety and depression, it is important to recognise that they require high levels of language ability to complete. Aphasia is a language disorder which impacts on an individual’s ability to understand and/or express themselves through verbal communication. Aphasia affects approximately 30% of stoke patients (Grönberg et al., 2022), and for many people with aphasia, traditional mood screening assessments are inaccessible. This has resulted in more than half of studies on the prevalence of depression and anxiety excluding patients with aphasia (Kouwenhoven et al., 2011), despite it affecting almost one in three patients. The limited research available suggests that aphasia is in fact a risk factor for depression, with one study reporting that 70% of people with aphasia met diagnostic criteria for depression three months post-stroke (Kauhanen et al., 2000). Another study reported that 50% met diagnostic criteria for either mild depression or major depressive disorder (Hunting Pompon et al., 2022).

Significant work has been undertaken to develop and validate aphasia-inclusive measures of mood and anxiety. van Dijk et al. (2016) completed a systematic review of aphasia-inclusive screening tests for depression. They reported that all studies had significant methodological limitations, and there was insufficient quality evidence to fully appraise the reliability and validity of each measure. There were two approaches to develop inclusive mood measures for patients with aphasia – observational measures and visual measures of mood (van Dijk et al., 2016).

Observational mood measures

Observational measures, such as the Stroke Aphasic Depression Questionnaire (SADQ; Sutcliffe & Lincoln, 1998) involves asking caregivers to complete a frequency count of target behaviours associated with depression over a one-week period (Leeds et al., 2004). In theory, this measure is relatively quick to administer, and does not require any specialist training. van Dijk and colleagues recommended the SADQ-10 item scale as having acceptable feasibility, although it has only “fair” classification accuracy (sensitivity = 0.7, specificity = 0.77; Leeds et al., 2004).

The Behavioural Outcomes of Anxiety (BOA) Scale is an anxiety specific observational scale, which is based on the SADQ format, and involves asking carers or healthcare professionals to rate the frequency with which an individual engaged in 10 target behaviours associated with anxiety (Linley‐Adams et al., 2014). The BOA has fair classification accuracy relative to scores on the Hospital Anxiety and Depression Questionnaire (HADS) (AUC = 0.75; Linley‐Adams et al., 2014).

Observational measures of mood and anxiety offer the opportunity to assess the emotional state of patients with aphasia. However, there are several important considerations. Firstly, the classification accuracy and convergent validity when compared to gold-standard measures is generally low, and in our clinical experience, many of the items on the SADQ and BOA overlap with the difficulties associated with pain, delirium and/or severe fatigue, which can make interpretation challenging. Additionally, administering a seven-day observational measure in a hospital setting poses challenges based on staffing shift patterns, for example, nursing staff do not work seven consecutive days. To accurately complete a SADQ and BOA, it involves daily ratings and recordings, which results in these measures being time consuming for staff. Engaging patients in mood screening can be a therapeutic activity, and the qualitative research suggests that people with aphasia want to be asked about their mood (Baker et al., 2020). This raises important questions about whether observational measures of mood may inadvertently reduce the likelihood of staff engaging directly with patients about their mood.

Adapted self-report measures

Adapted self-report measures often involve reducing the language requirements associated with conventional mood screening assessments, and instead using visual methods, i.e., pictures and visual analogue rating scales to assess for mood. It is important to note that to our knowledge, there are no validated aphasia-inclusive measures of clinical anxiety. The Visual Analogue Mood Scale (VAMS), VAMS-Revised (VAMS-R), and Visual Analogue Self Esteem Scale (VASES) have all been researched but are limited by having either no or low classification accuracy (Bennett et al., 2010; van Dijk et al., 2016). A computerised iteration of the VAMS, which involves the ability to manipulate facial expressions to match one’s own emotional state, the Dynamic VAMS (D-VAMS) has been developed (Barrows & Thomas, 2018; Barrows et al., 2021), and has demonstrated acceptable validity and good classification accuracy for identifying depression in a sample of people without aphasia.

While adapted self-report measures have some promise in assessing for post-stroke depression in people with aphasia, counterintuitively, they have been poorly researched in people with moderate to severe aphasia. van Dijk et al. (2016) report that most of the research has included samples with mild or no aphasia, and Vickery (2006) concluded from their study using the VASES that people with severe communication difficulties appeared to misunderstand the task, and therefore their answers were not interpretable. Additionally, an independent validation of D-VAMS found that it was not correlated with the SADQ-10 in people with aphasia (Ashaie & Cherney, 2021). Therefore, while these tests are theoretically valuable, there is a lack of clarity regarding which patients are most appropriate to receive this assessment.

In addition to unclear validity for use with people with aphasia and a lack of any scales to assess post-stroke anxiety, adapted self-report mood measures typically only explore emotional difficulties of depression, do not assess the cognitive or physiological sequalae of depression (see DSM-V) or any questions about suicide risk. Overall, despite people with aphasia experiencing a disproportionate prevalence of depression and anxiety, there are significant limitations associated with the available mood and anxiety screening tests, and new approaches are therefore required.

Stroke Aphasia Anxiety and Mood Measure (SAAMS)

SAAMS was developed by an interdisciplinary team of speech and language therapists (SLT), a neuropsychologist and a psychiatric nurse. The goal was to develop an adapted self-report mood measure which mapped clearly onto standard speech and language therapy assessment practices, assessed the whole construct of depression and generalised anxiety, assesses suicide risk, and most importantly, is accessible for patients with aphasia. This was developed as a clinical tool to inform clinical practice, rather than as part of a research study.

The development team started by using the items from the PHQ-9 and GAD-7, which assess the constructs of clinical depression and generalised anxiety, and map onto DSM-V criteria. Each item was reviewed by the panel with consideration for how it could be adapted for people with aphasia. Firstly, the wording on all questions was amended to simplify the question, remove double negatives, and make questions more concrete with key words highlighted in bold. Questions which could not be adapted were removed, e.g., “not being able to control worry”. Abstract information, such as understanding time and frequency were removed, and a simplified response chart was developed, whereby the words “yes”, “maybe” and “no” were written down and presented to patients, who could communicate their answer through pointing, gesture (nodding or shaking head and hand) or verbalising an answer. As such, the structure of the questions was changed from a frequency count of depression and anxiety symptomology to questions posed in the present tense, i.e., “do you feel depressed?”. Scores on each item ranged from 0 (absence of symptom) to 2 (presence of symptom), with a score of 1 being given for a “maybe” response. The information was presented visually, with key words in bold and an accompanying Boardmaker® picture to support with comprehension. And finally, reverse scoring of select items was included, to avoid perseverated responses from patients, e.g., consistently selecting “no” on each item without considering the question. The final set of items were selected based on consensus agreement by the team. The measure was used with several patients prior to the creation of a clinical database, to ensure it was feasible for use in clinical practice.

The objective of this paper is to provisionally evaluate the suitability and convergent validity of the SAAMS in patients with aphasia, and to determine whether it could be a useful screening tool for clinical practice on a stroke rehabilitation unit.

Method

Study design and participants

This study involved the retrospective analysis of an anonymised clinical database from a stroke rehabilitation hospital, and as such, it is a practice-based cross-sectional evaluation. This study received full ethical approval by the research ethics committee from the University of Plymouth. The database was used to collect patients’ mood screening scores for the purpose of the Sentinel Stroke National Audit Programme (SSNAP) in the United Kingdom. Patients were admitted onto the stroke rehabilitation unit for a period of inpatient rehabilitation. Patients who were unable to complete the standard mood measures (PHQ-9 and GAD-7) were identified by the multidisciplinary team, and instead were administered observational measures (SADQ and BOA). All patients with communication difficulties were assessed by the Speech and Language Therapists (SLT) working on the ward. Patients who were assessed by the SLTs as having a reliable yes/no response to concrete and in-context communication were then administered the SAAMS by either a Clinical Neuropsychologist or an Assistant Psychologist. The SAAMS was administered within one week of the SADQ and BOA. Assessment with the SAAMS was not predicated on aphasia severity, and no aphasia severity tool was used to classify patients. Given that patients were deemed by the ward’s multidisciplinary team as being unable to complete the standard mood measures but demonstrated an ability to engage reliably answer yes/no questions, it is likely that the patient’s aphasia severity was in the moderate to severe range, although this was not formally established. For some patients with more complex language difficulties, a joint session with an SLT was carried out. The SAAMS was administered using the standardised format for all patients.

The SAAMS was trialled with 23 patients. However, two patients were unable to complete it due to fatigue, and therefore their results have been excluded. Data from a total of 21 participants were included in this evaluation. The data comprises of 9 males and 12 females with an age range of 52–86 years old (M = 76.8 years; SD = 9.1), who were assessed during a two-year period (January 2021 to January 2023). One participant had a self-reported ethnicity of Middle Eastern, whereas the remaining 20 participants were White British. Participants’ stroke characteristics are presented in Table 1. The most common diagnosis was a left middle cerebral artery stroke. Most patients experienced receptive and/or expressive aphasia, and some patients, e.g., a patient with a right posterior cerebral artery infarct, experienced a “cognitive communication” difficulty, rather than aphasia. Given that this was a practice-based evaluation, a deliberate sampling procedure was not used, and therefore there is no record of the number of people with aphasia who did not have a reliable yes/no. Similarly, some patients fulfilled the criteria of being unable to complete the standard mood measures but having been assessed as having a reliable yes/no communication in an SLT assessment, were not administered the SAAMS due to staffing availability. The median length of time between stroke and mood screen was 42 days (IQR = 30, 70 days), and therefore the mood screen was completed in the post-acute stage.

Table 1. Stroke characteristics.

Stroke	Total	Percentage
Left MCA	8	38%
Right MCA	4	19%
Left ICH	3	14%
Multi-territory	2	9%
Right ICH	1	5%
Left ACA	1	5%
Right PCA	1	5%
Left PCA	1	5%

MCA = middle cerebral artery, ICH = intracerebral haemorrhage, ACA = anterior cerebral artery, PCA = posterior cerebral artery.

Materials

Stroke Aphasic Depression Questionnaire – Hospital Version (SADQ-H10)

The SADQ-H10 is a 10-item observational measure of depression for use with stroke patients with aphasia (Lincoln et al., 2000). Scores range from 0 to 30, with higher scores suggesting greater levels of behaviours associated with depression. A cut off score of 6/7 (i.e., 6 and below is a negative result, and 7 and above represent a positive result) is associated with a sensitivity of 0.68 and specificity of 0.79, although the authors caution against using a single cut off score due to low classification accuracy (Hacker et al., 2010). Conversely, a score of three or lower had adequate psychometric properties to rule out depression (sensitivity = 0.825), and a score of greater than seven had adequate psychometric properties to rule in depression (specificity = 0.858) (Hacker et al., 2010). The SADQ-H10 is thought to take 3–4 minutes to complete (Sutcliffe & Lincoln, 1998). At the site of this evaluation, due to staff fluctuations, it was not possible for one person to rate the SADQ-H10 based on seven days of observation because staff typically only worked for four or five consecutive days spanning 8–12 hours of the day. Therefore, The SADQ-H10 and BOA were completed at the end of each nursing shift by the patient’s nurse (night, day, and evening) for a period of seven days. The frequency of the target behaviours was tallied at the end of the week, and a total SADQ-H10 and BOA score was derived.

Behavioural outcomes of anxiety

The BOA is a 10-item observational measure of anxiety, which has been developed for use with stroke patients with aphasia (Linley‐Adams et al., 2014). Scores also range from 0–30. A cut off score of 16/17 is associated with good sensitivity (0.85) and specificity (0.85) (Eccles et al., 2017). The BOA was administered in this study using the same procedure as the SADQ-H10.

SAAMS

The SAAMS is made up of a depression (SAAMS-D) and anxiety (SAAMS-A) scale. Each item was scored on a scale from 0–2, with 0 reflecting the absence of the target symptom, 1 being given for a “maybe” response, and 2 denoting the endorsement of the target depression or anxiety symptom. Scores on each scale were summed to give a total score. The SAAMS-D consists of 8 items, with scores ranging from 0 to 16. The SAAMS-A is made up of five items, with scores ranging from 0 to 10. Higher scores represent higher endorsement of depression and anxiety symptoms. Prior to administration, all patients had been assessed as having a reliable yes/no, and therefore they had previous exposure to the assessment paradigm of the SAAMS. Patients were asked if they agreed to be asked about their feelings. No additional instructions were provided prior to administration of the SAAMS. The SAAMS took approximately 5 to 15 minutes to administer, depending on the patient’s communication ability. The wording and visual aids in the SAAMS are standardised, and responses are recorded on the record form. Please see Supplementary Information for a copy of the SAAMS.

Analysis plan

This study sought to establish whether the SAAMS has sufficient validity to warrant future research into its clinical usefulness. To achieve this, this study sought to establish its internal consistency, convergent validity, and classification accuracy. Internal consistency was calculated using Cronbach’s alpha and Guttman’s λ2 (Peters, 2014). Secondly, a series of Spearman’s correlations were used to determine the convergent validity against the SADQ and BOA. Finally, the classification accuracy was assessed using ROC curves. Depression and anxiety were categorised using scores on SADQ and BOA as the criterion variable, and the sensitivity and specificity of the SAAMS subscales were calculated. The statistical analysis for this evaluation was conducted using IBM SPSS statistical software (version 25).

Results

Descriptive statistics

Descriptive statistics were computed for the mood variables (see Table 2). Of note, the SAAMS-D demonstrated adequate skewness and kurtosis, and was normally distributed (Shapiro-Wilk = 0.915, p = 0.69). Similarly, the SAAMS-A achieved adequate skewness and kurtosis, and was normally distributed (Shapiro-Wilk = 0.927, p = 0.12). In total, 48% (10/21) of the sample scored above the cut-off for depression on SADQ-H10, and 19% (4/21) scored above the cut-off for clinical anxiety on BOA.

Table 2. Descriptive statistics on aphasia-inclusive mood measures.

	SADQ-H10	BOA	SAAMS-D8	SAAMS-D6	SAAMS-A5	SAAMS-A4
Mean	8.19	9.29	5.43	3.76	4.05	3.00
SD	5.83	6.08	3.28	2.84	2.85	2.57
Median	7	8	4	3	4	2
Min: Max	0:23	2:21	1:12	0:9	0:10	0:8
Skewness	0.82	0.85	0.64	0.44	0.76	0.74
Kurtosis	0.55	−0.16	−0.71	−1.11	0.10	−0.44

SD = standard deviation, SADQ-H10 = Stroke Aphasic Depression Questionnaire – Hospital version, BOA = Behavioural Outcomes of Anxiety, SAAMS = Stroke Aphasia Anxiety and Mood Screen.

SAAMS-D

The most highly endorsed difficulty on the SAAMS-D was feeling depressed (M = 1.0), followed by difficulty with sleep (M = 0.75), feeling bad about yourself (M = 0.75), and difficulty concentrating (M = 0.7). Having “thoughts about suicide or harming yourself” were rarely endorsed (M = 0.1). The test score reliability of SAAMS-D was assessed. The SAAMS-D had a Cronbach’s alpha of 0.66 and Guttman’s λ₂ = 0.69. Item level analysis revealed that the “feeling tired” item was poorly correlated with the overall construct (r = 0.20).

A Pearson’s correlation was conducted between SAAMS-D and the SADQ-H10. A small positive correlation was observed (r = 0.47, p = 0.03). Patients who were classified as “depressed” on SADQ scored higher on SAAMS-D (M = 6.91, SD = 3.65) than those classified as “not depressed” (M = 4.5, SD = 2.38), although this difference was not statistically significant (Mann-Whitney U = 13.0, Z = −1.18, p = 0.24).

The classification accuracy of the SAAMS-D was calculated. The classification accuracy of the SAAMS-D relative to SADQ-H10 was acceptable (n = 15; AUC = 0.71), with a cut off score of 2/3 (2 or low associated with “not depressed”, and 3 and above associated with “depressed”) associated with good sensitivity and 5/6 associated with reasonable specificity (see Table 3).

Table 3. Classification accuracy of SAAMS-D.

Cut off Score	Sensitivity	Specificity
1/2	1.00	0.00
2/3	0.82	0.25
3/4	0.82	0.50
4/5	0.64	0.50
5/6	0.55	0.75
6/7	0.55	1.00

SAAMS-A

Similarly, feeling restless was the most endorsed item on the SAAMS-A (M = 1.05) followed by feeling afraid (M = 0.9), “irritability” and “difficulty relaxing” (both M = 0.76). The test reliability of SAAMS-A was assessed. The SAAMS-A had a Cronbach’s alpha of 0.70 and Guttman’s λ₂ = 0.73.

A Pearson’s correlation was conducted between SAAMS-A and BOA. A small positive correlation was observed (r = 0.41), although this was not statistically significant (p = 0.06). The classification accuracy of the SAAMS-A was evaluated. Anxiety status was classified using a cut off of 16/17 on BOA, which resulted in 17 “not anxious” and 4 “anxious” classifications. The classification accuracy of the SAAMS-A relative to the BOA was fair (AUC = 0.75), with a cut off score of 3/4 achieving optimal sensitivity, and 5/6 achieving adequate specificity (see Table 4).

Table 4. Classification accuracy of SAAMS-A.

Cut off Score	Sensitivity	Specificity
0/1	01.00	0.22
1/2	1.00	0.28
2/3	1.00	0.41
3/4	1.00	0.59
4/5	0.50	0.71
5/6	0.50	0.84

Discussion

Despite people with aphasia being at increased risk of experiencing depression and anxiety following a stroke, there is a lack of aphasia-inclusive tools for assessing mood and anxiety. Conversely, many of the proposed aphasia friendly tools which have been developed have not been validated in people with moderate or severe aphasia (van Dijk et al., 2016), and there are significant concerns about their validity (Vickery, 2006). The SAAMS was developed by an interdisciplinary team of speech and language therapists, a neuropsychologist and a psychiatric nurse, to address many of the limitations to practice. Importantly, the assessment format mimics the assessment of reliable yes/no communication, which is routinely used by speech and language therapists, and therefore there is a clear process to determine which patients are suitable and unsuitable for this assessment, and patients with aphasia will already have had some exposure to the task. Furthermore, it samples the whole construct of depression and anxiety rather than only aspects of it, it includes a question on suicide, which is essential to practice, and the assessment questions and response format is informed by best practice in speech and language therapy for how to maximise effective communication with people with aphasia.

The current study sought to evaluate the internal consistency, convergent validity and classification accuracy of the SAAMS in people with post-acute stroke with significant aphasia, who were unable to complete conventional mood screening assessments. Firstly, the SAAMS was well-tolerated by all patients, and of the two people who were unable to complete it, this was attributable to fatigue rather than aphasia.

The SAAMS-D achieved adequate internal consistency and correlated with the SADQ total score which again that patients who are often excluded from being asked about how they feel provided information which was broadly consistent with observational measures of mood. The SAAMS-D achieved reasonable classification accuracy, and the use of an upper and lower cut off score may be useful in practice, with scores of less than 2 ruling out depression, and scores of greater than 5, ruling it in.

Similarly, the SAAMS-A also achieved good internal consistency. Scores on SAAMS-A were correlated with BOA, and although this was not statistically significant, this is likely a reflection of the lack of statistical power associated with the small sample. The SAAMS-A had good classification accuracy using an upper cut off score of five and above to rule in anxiety, and a lower cut off score of zero or one to rule it out.

Methodological considerations

It is important to adopt a critical perspective in interpreting the results of this study. While the inclusion of people with aphasia in the validation of an instrument is a rarity in the literature, the current study has significant threats of bias which should be considered. Firstly, the data is derived from clinical practice rather than a research trial, and therefore no sampling approach was used and there is an absence of a measure of aphasia severity to help situation the sample. The inclusion criteria of patients who were deemed unable to complete standard mood screening assessments and demonstrated a reliable yes/no in a speech and language therapy assessment may readily translate into stroke rehabilitation practice, the absence of clear sampling and aphasia staging limits our understanding of who this measure is appropriate for and increases the risk of bias. Secondly, this study had a small sample size, which poses a threat to internal and external validity, and indeed the lack of correlation between SAAMS-A and BOA is mostly likely a type 2 error given the magnitude of the correlation coefficient (0.41) observed. Furthermore, in typical validation studies, the test and criterion measures are administered in the same sitting to avoid the impact of confounding factors. In this evaluation, the observational measures were administered over a seven-day period, whereas the SAAMS was administered within one week of this. This variability in administration time may result in increased measurement error.

Another significant limitation in this study is the use of the BOA and SADQ-H10 to classify patients as being clinically depressed and anxious. While both measures have some validity, they have not been well-validated in people with significant aphasia, and their classification accuracy, especially the SADQ-H10, has a high rate of error (Hacker et al., 2010), and therefore it is likely that at least some participants were misclassified. While the SADQ-H10 and BOA are the “best available” measures of classifying depression and anxiety in this population, they are by no means “gold standard”, and therefore the classification accuracy statistics produced in this article should be treated with caution.

It is important to reflect on how SAAMS measure was developed. The SAAMS was developed with clinical practice in mind. This means that it did not go through many of the steps normally required for a research study, i.e., including patient and public involvement (PPI) to co-design the assessment and an appropriate power analysis to guide the recruitment target. Additionally, the SAAMS altered the establish mood screening paradigm of asking people to rate the frequency of their symptomology over a time. In contrast, the SAAMS has removed the frequency count because of the abstract nature of this, and instead asks present tense yes/no question about someone’s emotional wellbeing. This means that it is difficult to establish whether an individuals scores reflect their mood in that moment, or whether it reflects a more enduring emotional state characteristic of mental health difficulties. A retest reliability analysis is needed to determine the stability of the SAAMS results, and it may be advisable to repeat the administration of the SAMMS on several occasions over time, to establish the stability of emotional difficulties.

Given that people with aphasia were able to reliably engage with the SAAMS and the scale achieved reasonable classification accuracy, a full research trial of SAAMS is warranted. Full validation of SAAMS would require PPI involvement to finalise the design of the measure, followed by an adequately powered trial using stroke patients without aphasia, so that the SAAMS could be compared against a gold standard mood measure for stroke patients to determine the most appropriate cut off score. Following this, the SAAMS should be re-evaluated in people with aphasia, which would include a controlled sample, with aphasia staging (i.e., mild, moderate, severe) and measure the retest and inter-rater reliability and the factor structure of the measure.

Practical applications

The findings of this study provide some evidence that people with aphasia who have a reliable yes/no communication, can provide reliable results on a screening tool of anxiety and depression, and as such, this study provides provisional evidence of the validity of this measure in helping the multidisciplinary team to understand the needs of patients with significant aphasia. On this basis alone, the SAAMS is a useful tool. Additionally, unlike observational measures, the SAAMS includes a question about suicide and self-harm, which can help clinicians to assess risk. In its current format, clinicians are required to ask follow-up questions to determine whether a “yes” response refers to suicide, self-harm or both. In future iterations of SAAMS, this will be separated into two separate questions. Given the lack of sampling procedure, the limitations associated with using the BOA and SADQ-H10 as criterion variables, and the lack of retest reliability, the authors of this paper believe that it is premature to recommend that the cut off scores for clinical depression and anxiety outlined in this article are adopted in clinical practice. Instead, the results of this study suggest that the SAAMS is worthy of further research using a controlled sampling procedure, aphasia staging, and a larger sample, to fully validate the use of SAAMS for clinical practice.

Conclusions

The SAAMS is a novel measure of clinical depression and anxiety which has been developed for post-stroke patients with aphasia by an interdisciplinary team of stroke rehabilitation professionals. Given the overlap between the SAAMS and speech and language therapy practice, it offers important benefits over previously developed aphasia-inclusive screening tools. The results of this evaluation suggest that people with aphasia, with a reliable method of yes/no communication, are able to reliably rate their anxiety and depression. While this study identified that the SAAMS-D and SAAMS-A both had reasonable classification accuracy for diagnosing depression and anxiety, further research is needs before this tool can be recommended as a screening test for use in standard clinical practice.

Author contributions

Donnchadh Murphy: conception, data collection, analysis, writing, editing, approval and accountability.

Jess Hourston: data collection, analysis, writing, editing, approval and accountability.

Emma Freeman: conception, editing, approval and accountability.

Rosanna Morris-Haynes: conception, editing, approval and accountability.

Nicky Hawker: conception, editing, approval and accountability.

Statement of ethics

This study protocol was reviewed and approved by the Research Ethics Committee at University of Plymouth.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

The data used in this article was derived from an anonymised clinical database. As such, the data used in this article will not be made available.

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/02687038.2024.2361512.

Additional information

Funding

The author(s) reported there is no funding associated with the work featured in this article.