Development, Implementation, and Outcomes of Post-stroke Mood Assessment Pathways: Implications for Social Workers

ABSTRACT

Depression and anxiety are prevalent in the stroke population and can have a significant impact on the patient and their family’s long-term outcomes, however the screening for these conditions does not meet best practice recommendations. To address this deficit, this study developed “Post-Stroke Mood Assessment Pathways” and embedded them into practice by utilising the PARiHS framework (for the implementation of evidence-based practice), in conjunction with staff training. The study examined the rates of mood screening, clinical interviews, and completion of interventions for stroke patients through a retrospective chart audit (n = 213) one year prior to, and one year post-implementation (n = 238) of the pathways. The data show statistically significant increased documentation around mood screening and clinical interview 95% CI [4.86, 19.88], p < .0012 and specifically, an increase in the number of patients who had a clinical interview following the introduction of the pathways 95% CI [8.05, 19.69], p < .0001.

IMPLICATIONS

• Implementing post-stroke mood assessment pathways assists in meeting the National Stroke Foundation Guidelines (2017) by providing clinically validated tools to screen for depression and anxiety in the stroke population.

• The social work profession is well positioned to incorporate post-stroke mood screen and assessments into their roles in view of their professional commitment to enhancing social and emotional functioning of patients and families.

中风人群多感到沮丧和焦虑，这对患者及其家庭有着长期影响。相关研究还无法为实践提供最好的建议。为补缺漏，本文采用PARiHS框架（为开展询证实践）并结合人员训练，提出“中风后情绪评估路径”并将其用于实践。本文通过对路径实施一年前及一年后（n=238）的回顾性图表审计（n=213），考察了中风病人的情绪筛查、临床访谈以及干预完成的比率。数据显示，关于情绪筛查及临床访谈的记载在统计上有明显的增加，尤其是引进该路径后做临床访谈的数量有了增加。

KEYWORDS:

• Stroke
• Social Work
• Screen
• Mood
• Pathway
• Depression
• Anxiety

In 2017, an estimated 55,000 Australians experienced a new or recurrent stroke and an estimated 470,000 Australians were living with the effects of stroke (Stroke Foundation, 2017). The literature demonstrates a significant number of those affected will experience depression, anxiety, or both, at some point after the event (Bennett, Thomas, Austen, Morris, & Lincoln, 2006; Kneebone, Baker, & O'Malley, 2010; Sagen et al., 2009). Depression affects 33% of patients after a stroke and is associated with reduced participation in rehabilitation, longer hospital stays, increased physical impairment, and increased mortality (Hackett, Yapa, Parag, & Anderson, 2005; Sagen et al., 2009). Anxiety affects 18–25% of stroke patients (Campbell Burton et al., 2013; Lincoln et al., 2013) and is associated with reduced participation and functional ability, social isolation, and reduced quality of life (Ahlsio, Britton, Murray, & Theorell, 1984; Astrom, 1996; Donnellan, Hickey, Hevey, & O'Neill, 2010). Additionally, changes in the patient’s mood can impact on their social network. Carers cite anxiety and depression in the person with stroke, as among the most stressful difficulties post-stroke (Haley, Roth, Kissela, Perkins, & Howard, 2011).

Determinants for depression and anxiety are multifaceted; broadly categorised they consist of (a) the biological outcomes, i.e., the severity of the stroke (Donnellan et al., 2010) and aphasia (National Stroke Foundation, 2010; Welch, 2008), (b) the individual’s psychological status, and (c) the presence, engagement, and capability of the individual’s social support structures (Hilari & Byng, 2009; Waldron, Casserly, & O'Sullivan, 2013). Depression and anxiety can persist for years, well into the chronic stage of stroke recovery (De Wit et al., 2008; Donnellan et al., 2010; Lincoln et al., 2013). The Australian National Stroke Foundation recognises the impact that depression and anxiety can have on the patient and the support network, and recommends that, “all stroke patients should be screened using a clinically validated tool”, and, that a “diagnosis should only be made following a clinical interview” (National Stroke Foundation, 2010, p. 107).

However, assessment can be complicated by the natural responses to the significant changes in the stroke survivor’s life, which can include grief, frustration, anger, and guilt (Welch, 2008). The process of undertaking a mood screen and conducting a clinical interview provides the opportunity for clinicians to engage with the affected person and family around mood and to assess if the mood indicates more debilitating mental disorders, such as depression or anxiety, are also present (Bourne, Eckhardt, Sheth, & Eskandar, 2012; National Stroke Foundation, 2010). The Australian National Stroke Audit – Rehabilitation Services Report of 2015 (Stroke Foundation, 2016) indicates that 53% of patients with a stroke who were admitted to an Australian rehabilitation unit were assessed for mood impairment.

Similar to the Australian guidelines, the 2008 United Kingdom (UK) National Clinical Guidelines for Stroke (cited in Gurr, 2011) recommends that all patients be screened for depression. However, Gurr (2011) points out that parallel deficits in screening rates are demonstrated in the UK. Gurr (2011) conducted a project to determine whether developing and implementing a post-stroke mood screening pathway improved guideline compliance. The pathway implementation saw 88.9% of patients over a six-month period screened compared with the national median of 67.8%.

The reduced Australian rates of mood screening and clinical interview may be due to reduced implementation of research findings into practice, with a corresponding lower level of knowledge and skills in the clinical setting (Bennett in Welch, 2008). Research shows that patients’ physical care needs are routinely prioritised over their psychological care (Adkins, 1993; Perry in Welch, 2008). In view of Australian compliance rates, the Australian National Stroke Audit – Rehabilitation Services Report (Stroke Foundation, 2016, p. 50) recommends a “greater focus on processes to ensure the psychological needs of all patients are assessed and appropriate support is provided during and after inpatient rehabilitation”.

An audit of the clinical records of all stroke patients admitted to the Wide Bay Hospital and Health Service (WBHHS) South between 1 October 2013 and 30 September 2014 demonstrated a 14.5% compliance rate with the Australian National Stroke Foundation’s recommendations that all patients have a mood screen and a clinical interview if low mood is identified (National Stroke Foundation, 2010). Hart and Morris (2008) suggest that mood screening compliance may be enhanced by: (a) training to increase knowledge and skills, (b) increasing awareness of guidelines, (c) support from colleagues, and (d) the integration of mood assessment into job roles including routine assessment activities. As such, healthcare social workers are ideally positioned to lead mood screening and “enhance social and emotional functioning through targeted interventions and the mobilisation of services and supports” (Australian Association of Social Workers, 2016, p. 3).

A range of theoretical and practice models are available to inform the approach to implementing evidence-based guidelines (Best & Holmes, 2010; Cooke, Ariss, Smith, & Read, 2015; Dixon-Woods, 2014; Evans & Scarborough, 2014; Graham et al., 2006) with each acknowledging that change to clinical processes and the translation of research to practice is an intricate dynamic. One framework that offers an integrated tool to assess, plan, and evaluate clinical change is the Promoting Action on Research Implementation in Health Services (PARiHS) model (Kitson, Harvey, & McCormack, 1998). The model considers three primary themes: (a) the evidence, (b) the context, and (c) the facilitators of change, and is supported by a body of literature demonstrating its effectiveness across a variety of health settings (Harvey et al., 2002; Kitson et al., 2008; McCormack et al., 2002; Rycroft-Malone, 2010; Rycroft-Malone et al., 2002; Rycroft-Malone, Harvey, et al., 2004; Rycroft-Malone, Seers, et al., 2004). In this study, the PARiHS model was chosen as the framework to develop and implement the post-stroke mood assessment pathways specific to the needs of the local population.

Methods

Aim

To develop and implement Post-Stroke Mood Assessment Pathways, which, in conjunction with staff training, would improve the rates of mood screening, clinical interviews and interventions for patients who have had a stroke. Ethical approval was received from a Queensland Health Human Research Ethics Committee (HREC/14/QPCH/174 date 14.8.2014).

Setting

The study was carried out in the WBHHS South at two hospitals from 1 October 2013 to 30 September 2015. Hospital A has 177 beds, an intensive care unit, and no rehabilitation unit. Hospital B has no intensive care unit and has 44 sub-acute beds (B1) and a dedicated 16-bed rehabilitation unit (B2). Patients are frequently admitted to Hospital A and subsequently transferred to Hospital B for their sub-acute and rehabilitation phase.

Design

The study utilised a pre- and post-test design that consisted of a retrospective audit of patient charts in the 12 months prior to the implementation of the pathways and staff training on 1 October 2014, and 12 months post-implementation. The outcome measures were: (a) the rate of mood screen completions, (b) the rate of clinical interviews, and (c) if low mood was identified, the rate and types of interventions. For the purposes of this research, the authors defined a clinical interview as “documented evidence of a conversation with the patient in regard to their mood”.

Chart Audit Sample and Data Collection Tool

Patients were followed from admission to discharge across the two hospital settings and through community health. Each discharge from a service was examined separately to determine if assessment for low mood was conducted during that episode of care. Community health patients were included in the audit only if they were specifically referred for mood follow-up or intervention. The chart audit included all patients > 18 years old admitted to Hospital A or Hospital A and B within the audit periods with a diagnosis of a stroke categorised under the following ICD 10 codes: a) subarachnoid haemorrhage i60.0–i60.9, (b) intracerebral haemorrhage i61.0–i61.9, (c) other nontraumatic intracranial haemorrhage i62.0–i62.9, (d) cerebral infarction i63.0–i63.9, (e) stroke, not specified as haemorrhage or infarction i64. Charts were excluded if the stroke patient died within four weeks of the stroke (from any cause) or was transferred out of the health service in the four weeks following the stroke.

The audit recorded: patient demographics, length of stay from time of stroke to time of discharge, mood screening assessments, the health professionals who conducted the mood screening(s), the location of the screening, and interventions for low mood.

Development of Mood Assessment Pathways

Search for Screening Tools

There was no standardised methodology within the WBHHS South for mood assessment post-stroke, i.e., no standard screening tool or mood assessment pathway. Paradiso and Robinson (1998) indicate that the optimal means for assessing for depression and anxiety is to apply the criteria in the Diagnostic and Statistical Manual (DSM-5) using the Structured Clinical Interview for DSM-5 disorders (SCID). However, this assessment requires specific staff training and the SCID-5-CV generally takes 30–120 min to administer. The SCID-5-RV Core Version takes 45–120 min, and the SCID-4-RV Enhanced Version takes 45–180 min (J. Cha, personal communication, 23 March 2017). The time requirements of this assessment tool posed an obstacle to it being implemented broadly into routine practice across the local health services. Accordingly, alternative means of screening for mood disorders that could be incorporated into a pathway needed to be considered.

A search of electronic databases (PsychoINFO, CINAHL, EMBASE and Medline) from their inception to December 2013 using the keywords: cerebrovascular accident or CVA OR stroke and screen* OR pathway OR tool OR scale OR measure* OR question* AND mood OR depression OR anxiety was undertaken. All searches were limited to English language and human studies. Reference lists of selected articles and reviews were searched (n = 267). The title, abstracts and then full texts were reviewed by two researchers to identify reports with validated post-stroke tools and pathways where mood was assessed in post-stroke patients, and where referral or treatment was identified.

Shortlisting of Screening Tools and Pathways

The clinical utility of both the screening tools and pathways were considered important parameters. They needed to be suitable for use in acute, sub-acute, inpatient rehabilitation and community rehabilitation settings, and be effective and efficient. Tools and pathways that required specific trained personnel or took excessive time were excluded. In addition, tools and pathways that were perceived by the researchers to be confusing were excluded. Screening tools needed to: (a) assess for both depression and anxiety, (b) be used in the presence of cognitive and communication difficulties, (c) have been validated in the stroke population (Bennett et al., 2006), and (d) be suitable for all >18 years. A shortlisting conducted by the steering committee produced a total of 12 nonduplicate screening tools. The screening tools identified as being the most valid and efficient were: HADS-A, PHQ-9, SADQ-10 or SADQ-H10 (for hospital patients) and the BOA, the clinical utility of which had been evidenced by Kneebone, Neffgen, and Pettyfer (2012).

The pathway search produced a total of 18 pathways. However, Kneebone et al. (2012) had already developed pathways incorporating the tools (apart from the SADQ-H10) identified in this study’s search. Kneebone et al.’s (2012) pathways were designed for use in the community and assess for both anxiety and depression in patients with and without communication challenges. The authors adapted these pathways to suit the WBHHS by including the SADQ-H10 (specifically for inpatients) and ensuring the pathway reflected appropriate local referral processes (e.g., refer to Mental Health Services or a social worker). Table 1 describes the final screening tools that were considered as applicable to the needs of the WBHHS, Table 2 describes the final screening tools that were not considered applicable, and Table 3 lists the final pathways that were considered to inform the development of the context specific pathways.

Table 1 Mood Assessment Screening Tools Considered for Inclusion into the Mood Assessment Pathway

Pathway	Context of use	Valid in stroke, efficient sensitivity	Benefits/issues	Included in pathways analysis
Hospital Anxiety Depression Scale—Anxiety (HADS-A)	It can be used for all adults. Initially developed for use in non-psychiatric hospitals however further research demonstrates it is reliable and valid in other settings (Snaith & Zigmond, 1994) Self-reported tool.	The sensitivity of the HADS-A is higher than the DASS-A (Dahm, Wong, & Ponsford, 2013). A cut-off of six recommended for stroke patients (Kneebone et al., 2012). Burton and Tyson (2015b) systematic review advises the HADS was the most effective tool to identify anxiety.	There is a copyright on the HADS-A resulting in a cost to implement it.	This tool was chosen for the pathway due to its proven validity in the stroke population and its efficiency (Aben, Verhey, Lousberg, Lodder, & Honig, 2002)
Patient Health Questionnaire-9 (PHQ-9)	Recommended as a clinically valid tool for stroke patients by the National Stroke Foundation (2010). This tool can be used regardless of age, gender, or ethnicity. It taps the nine diagnostic symptom criteria that define DSM-IV major depressive disorder (Lowe, Kroenke, Herzog, & Grafe, 2004). Self-reported tool.	It is quick to administer and has adequate reliability, validity, sensitivity, and specificity in stroke samples (Lowe et al., 2004). Good sensitivity for detecting severe depression and recommended as effective tool in stroke patients (Burton & Tyson, 2015b)	It can be used for patients without significant cognitive or communications problems (Kneebone et al., 2012). Freely available	This tool was chosen for the pathway due to its proven validity in the stroke population and its efficiency
Behaviour Outcomes for Anxiety (BOA)	Specifically designed to detect and monitor anxiety post-stroke for people with communication challenges. The BOA is an observational rated instrument that is completed by someone who knows the patient well (Linley-Adams, Morris, & Kneebone, 2014).	Eccles, Morris, and Kneebone (2017) suggest the BOA has adequate sensitivity and specificity in the stroke population.	It is the only tool screening for anxiety that has been developed for stroke patients who experience communication or cognitive problems. Limited research to date as the tool is relatively new. Freely available	This tool was chosen for the pathway as it is the only tool to specifically assess anxiety for stroke patients with significant communication or cognitive challenges.
Stroke Aphasic Depression Questionnaire-10 or H10 for hospital patients (SADQ-10 or SADQ-H10)	Specifically designed for stroke patients with significant communication challenges. It can be used for all age groups and is an observer rating scale that is completed by the primary nurse, key worker, or clinician who knows the patient best.	Kneebone et al. (2010) recommend a cut-off is six or more on the SADQ-H10 and 14 on the SADQ-10 (Kneebone et al., 2012). The clinician should assess suicidal ideation if patient scores six or more (if possible given communication challenges). Good psychometric and clinical utility for major depression and any depressive disorder (Burton & Tyson, 2015b).	This tool is recommended to screen for depression in stroke populations for patients who have cognitive and/or communication problems that preclude administration of the PHQ-9 (Kneebone et al., 2012). Freely available	This tool was chosen as it has the most research to support its use in detecting depression in stroke patients with significant communication challenges (Burton & Tyson, 2015b)

Table 2 Mood Assessment Screening Tools Not Considered for Inclusion into the Mood Assessment Pathway

Pathway	Context of use	Valid in stroke, efficient sensitivity	Benefits/issues	Not included in pathways analysis
Hamilton Depression Rating Scale	This is a clinically valid tool (Aben et al., 2002) that is recommended by the National Stroke Foundation (2010).	It has similar predictive validity as HADS and BECK, however requires trained staff to administer and is more time consuming (Aben et al., 2002).	Training required to implement tool (Aben et al., 2002). Takes 15–20 min to administer (Sharp, 2015). Freely available.	This screen was excluded from our pathway due to length and training requirements (Aben et al., 2002).
Geriatric Depression Scale 15-item (GDS)	The GDS is clinically valid in the stroke population for over-65s. It is recommended by the National Stroke Foundation (2010). There has been limited research conducted in < 65 years population, therefore researchers excluded this tool on the pathway as we ideally wanted a valid tool that could be used for all ages.	The researchers acknowledge that this is a valuable tool for patients with hearing impairments, expressive dysphasia, or mild cognitive difficulties. Some of the user rated scales have quite complicated response options that can be confusing for the user (Cinamon, Finch, Miller, Higgins, & Mayo, 2011). In such cases the GDS could be used as an alternative. (Roger & Johnson-Greene, 2009) recommend a cut off of 3 in the stroke population (usual cut off is 6). The GDS has high sensitivity and internal consistency in the elderly stroke patient cohort (Agrell & Dehlin, 1989).	The yes/no answers as opposed to scaling questions a good alternative for those with hearing or mild cognitive impairments. Freely available.	Not included in pathway as it targets > 65 years of age. The researchers do recommend this to be considered as a complimentary tool for its specific target group.
Brief Assessment Schedule for Depression Cards (BASDEC)	Recommended for use in the over 65 age group in noisy environments that lack privacy, such as hospital wards (Kneebone et al., 2010). It is a valid tool in the over 65 years of age stroke population (Healey, Kneebone, Carroll, & Anderson, 2008).	Acceptable internal consistency and reliability in older stroke survivors in an inpatient rehabilitation setting (Healey et al., 2008).	Tool unavailable therefore unable to be included.	The BASDEC cards were not chosen because the authors wanted to be able to use the tools across all ages. Additionally, unable to obtain them from the publisher at the time of writing.
Depression Intensity Scale Circles (DISCs)	The DISCs can be used for under and over 65 year olds if there is a major cognitive or communication difficulty (Kneebone et al., 2010).	Burton and Tyson (2015b) state the DISCs did not have adequate sensitivity and specificity in detecting any depression or major depression.	There is no suicide question, therefore Kneebone et al. (2010) recommend using the BASDEC suicide question. Freely available	Excluded due to lack of sensitivity and specificity in the stroke population.
Depression and Anxiety Stress Scale (DASS)	There has been limited research in using the DASS with the stroke population (Ownsworth, Little, Turner, Hawkes, & Shum, 2008).	Sukantarat, Williamson, and Brett (2007) conducted a comparative study between the DASS and HADS for patients who had experienced a recent critical illness. Both DASS and HADS emerged as reliable tools for detecting anxiety and depression in this patient group, however Dahm et al. (2013) found that the sensitivity of the HADS-A is higher than the DASS-A	Inadequate research to support its validity in the stroke population Freely available	The DASS was not chosen for the pathway due to the limited research conducted specifically with the stroke population
Hospital Anxiety Depression Scale –Depression (HADS –D)	It can be used for all adults. Initially developed for use in non-psychiatric hospitals, however further research demonstrates it is reliable and valid in other settings (Snaith & Zigmond, 1994).	Research suggests the PHQ-9 a superior tool for screening Post-Stroke Depression as it taps into the nine diagnostic symptom criteria that define DSM-IV major depressive disorder (Kneebone et al., 2012).	There is no suicide question in the HADS-D, which the researchers felt was important.	The HADS-D was not chosen for the pathway as the PHQ-9 had greater research to support its use in the stroke population.
Generalised Anxiety Disorder-7 (GAD-7)	The GAD-7 is a brief (7 item) tool to assess anxiety in the general population (Spitzer, Kroenke, Williams, & Löwe, 2006). The questions are scaling questions, therefore patient requires reasonable language and cognition to follow.	The GAD-7 has not been specifically evaluated in the stroke population (Eccles et al., 2017).	Requires further evaluation in regard to its effectiveness in the stroke population (Eccles et al., 2017).	Inadequate research using this tool in the stroke population existed at time of implementation.

Table 3 Mood Assessment Pathways Considered for Inclusion into the WBHHS Post-stroke Mood Assessment Pathways

Pathway name, (Weighting), Authors	Ease of use	Training required	Clinical utility	Time taken to administer (minutes)	Validity in stroke population yes/no	Score/10
Screening for depression and anxiety after stroke: Developing protocols for use in the community. (Weighting)	(2)	(2)	(1.5)	(2)	(2)	(9.5)
Kneebone et al. (2012)	Streamlined	HADS-A: Nil PHQ-9: Nil BOA: Nil SADQ 10: Nil	Straight forward	HADS-A <5 PHQ-9 < 5 BOA <5 SADQ-10 < 5	HADS-A: yes PHQ-9: yes BOA: yes SADQ-10: yes
Screening for depression after stroke: Developing protocols for the occupational therapist (Weighting)	(1)	(1.5)	(0.5)	(2)	(1.5)	(6.5)
Kneebone et al. (2010)	More complex/more variables	DISCS: Nil SADQ-H10: Nil BASDEC: minimal (unable to obtain) HADS: Nil	Unable to obtain the BASDEC from publisher at time of writing	DISCS <5 SADQ-H10 < 5 BASDEC 3–4 HADS <5	DISCS: yes* SADQ-H10: yes BASDEC: yes HADS: yes
Mood Pathway—Stroke Services (Weighting)	(1)	(1.5)	(0)	(2)	(1.5)	(6.5)
The Royal Bournemouth and Christchurch Hospitals NHS Foundation Trust (Clark & Taylor, 2010)	More complex/more variables	DISCS: Nil BADEC: (unable to obtain) SADQ-H10: Nil HADS: Nil	Unable to obtain the BASDEC from publisher at time of writing	DISCS <5 BASDEC 3–4 SADQ-H10 < 5 HADS <5	DISCS: yes* BASDEC: yes SADQ-H10: yes HADS: yes
G-Master (Weighting)	(2)	(1.5)	(2)	(2)	(1)	(8.5)
Greater Manchester Assessment of Stroke Rehabilitation (Burton, Tyson, & McGovern, 2013)	Streamlined	SADQ-H10: Nil DISCS: Nil PHQ-9: Nil GAD-7: Nil**	Straight forward	SADQ-H10 < 5 DISCS <5 PHQ-9 < 5 GAD -7 < 5**	SADQ-H10: yes DISCS: yes* PHQ-9: yes GAD 7: not specifically evaluated in the stroke population

Notes. Weighting Scale 0 = absent, 0.5 = low, 1 = medium, 1.5 = medium high, 2 = high

*valid with stroke patients though did not have adequate sensitivity and specificity in detecting any depression or major depression (Burton & Tyson, 2015b). **Authors unable to obtain published information. Tested in author’s research.

As aphasia affects 23–38% of people who experience a stroke (Dickey et al., 2010; Engelter et al., 2006), the authors followed the work of Kneebone et al. (2012) by developing two tools, one that assessed patients with cognitive and communication difficulties (Figure 1), and the other that assessed patients without those difficulties (Figure 2).

Figure 1 Post-stroke mood assessment pathway for patients without communication or cognitive difficulties (Adapted with permission from Kneebone et al., 2012)

Figure 2 Post-stroke mood assessment pathway for patients with communication or cognitive difficulties (Adapted with permission from Kneebone et al., 2012)

After obtaining and reviewing the literature for this project, a systematic review by Burton and Tyson (2015a) identified a range of alternative mood screening tools that have varying levels of applicability in the context of stroke. The Burton and Tyson review (2015a) further supported the tools Kneebone et al. (2012) used in their pathways, however the BOA was not included in the Burton and Tyson review (2015a). At that time, the BOA was a reasonably new tool and may not have had been clinically verified for the rigour of systematic review analysis. In this project, the BOA was included as it was the only screening tool to assess anxiety in stroke patients with significant cognitive and communication difficulties.

Development and Implementation of Staff Training

The pathways were constructed by authors PMcL and RT and agreed upon by a multidisciplinary team who were identified as having stakeholder involvement in the management and care of stroke patients. The team consisted of speech therapy, occupational therapy, social work, and mental health services, and nursing and medical teams. The multidisciplinary team utilised the PARiHS framework (Rycroft-Malone, 2010) to develop a structured approach to the implementation of the mood screening pathways. Table 4 outlines the specific activities, personnel responsible, and their goals in line with the PARiHS framework.

Table 4 The PARiHS Framework Applied to the Target Groups, the Activities, the Responsibilities and Goals of the Research

Target group	Activities	Health professional	Goal	PARiHS Framework
Stroke patients in the Wide Bay South Hospital and Health Service	Discussion of research project Consider aim and purpose Identify Australian PSMAPs	SW rehab SW community	Determine whether implementing a PSMAP would improve rates of mood screens in conjunction with clinical interview for patients who have had a stroke	E, C, F
SW rehab SW community	Literature review	SW rehab SW community	Identify clinically validated screening tools for stroke population	E, F
Stroke patients: Pre-implementation (2013–2014) Post-implementation (2014–2015)	Develop audit tool Chart audit WBHHS (South)	SW rehab SW community Allied health volunteers	Identify: Completion of mood screens Clinical interviews Interventions	E, F
SWs and OT	Stakeholders meeting	SW rehab SW community	Identify the health professionals who would conduct the PSMAP	E, C, F
OT acute SW sub-acute/rehab SW community	Staff training	SW rehab SW community	Complete survey Increase knowledge and skills Create readiness for change	E, C, F
OT and SWs in acute, sub-acute, rehab and community	Implement the PSMAP pathway	OT and SWs	Quality improvement Identify patients at risk of depression and/or anxiety	C, F
Speech therapists	Consultation re patients with aphasia/communication challenges	OT and SWs	Determine most effective screening tool to be used	E, F
Patient (in hospital) with their consent	Complete PSMAP in conjunction with clinical interview MDT input	OT acute SW sub-acute/rehab All MDT Mental health liaison	Identify patients at risk of depression/anxiety Intervention	E, C, F
Family/carer (for inpatients)	Attend family meeting	SW and MDT	Support patient and family/carers Discharge planning	E, C, F
Patients community with their consent	Complete PSMAP and/or review and repeat mood screen Ongoing intervention	SW community	SW support Intervention	E, C, F
Stroke patients in the Wide Bay Hospital and Health Service (South)	Evaluation	SW rehab SW community	To determine if implementing a PSMAP improved rates of mood screens in conjunction with clinical interview Improve psychological wellbeing of patients	E, C

Notes. PSMAP—Post-Stroke Mood Assessment Pathway; SW—social worker; OT—occupational therapist; MDT—multidisciplinary team. E—Evidence; C—Context; F—Facilitation

Identification of Barriers and Enablers

The key professions identified by the multidisciplinary team as appropriate, responsible, and accountable for the conduct of mood assessments in the context of the health services were: occupational therapy—acute; social work—sub-acute; social work—rehabilitation; and social work—community. All staff in those categories working closely with stroke patients in the service area were included in the staff training (n = four social workers and one occupational therapist). Social workers were deemed to be the preferred staff member to complete the assessments as one of their core roles within the hospital environment is to improve psychosocial outcomes in relation to depression (Australian Association of Social Workers, 2016). Due to the staff mix in this regional health service, occupational therapy was chosen to conduct the assessments in the acute context. A pre-training survey of the group’s knowledge and practice elicited responses from three out of five key professionals. Of those responses, all considered mood assessment post-stroke as part of their role, however only one routinely conducted that assessment. Barriers to assessment were identified as: (a) knowledge regarding validated screening tools, (b) time, and (c) environmental issues including privacy on wards.

Staff Training

The purpose of the training was to enable therapists to acquire the knowledge and skills to competently undertake mood screens. The training was delivered over one three-hour period on 1 October 2014 by authors PMcL and RT and involved the use of discussion, a PowerPoint presentation, and role play. The training commenced with a trigger point discussion on: the rates of depression and anxiety in stroke patients, the National Stroke Foundation Guidelines, and an introduction to Pathway A (Figure 1) and Pathway B (Figure 2) and their mood screening tools.

Case studies were then provided and therapists completed role plays assessing each case as per the appropriate pathway. The roles plays were conducted in pairs taking turns to follow the pathways and undertake mood screens. The authors were present to provide assistance and clarification. During the training, and with support from a speech therapist, the authors encouraged the therapists to explore a range of ways to directly communicate with patients who had communication challenges and to use carer-rated scales and tools only when those direct methods were not viable. Following the training, the pathways were provided to the teams and staff responsible for implementing them and were distributed to the ward areas. There was no training follow-up, however, therapists were verbally encouraged to use the pathways and their use was routinely discussed in team meetings.

Procedures

The chart audit was conducted by authors PMcL and RT with the assistance of colleagues who were trained in the use of the data collection tool. Complete verification of data was conducted by authors PMcL, RT, and AR. AR cleansed the data in Excel and imported it into SPSS and conducted the data analysis. The primary analysis was the proportion change using the “N-1” Chi-squared test as recommended by Campbell (2007) and Richardson (2011). The confidence interval was calculated according to the recommended method by Altman, Machin, Bryant, and Gardner (2000).

Findings

The findings from the 2013–2014 and the 2014–2015 survey periods are presented in Tables 5–8. Table 5 shows that n = 213 patient charts were audited in the pre-implementation period, and n = 238 in the post-implementation period (11.7% increase). There was an increase in discharges from Hospital A in the second period (n = 10), however there were a similar number of transfers from Hospital A to Hospital B in the two periods (n = 71 versus n = 74). Sub-acute discharges were similar in the two periods (n = 20 versus n = 19) and there was a small increase in the number of transfers from sub-acute to rehabilitation in the second period (n = 6). There was a larger increase in the number of patients referred for social worker community health follow-up (mood assessment or intervention) post-rehabilitation discharge in the second period compared with the first (n = 10).

Table 5 Chart Audit Pre- and Post-implementation of Post-stroke Mood Assessment Pathways and Staff Training

Variable	Pre-implementation n =	Post-implementation n =
Charts audited (Hospital A discharges and transfers, Hospital B sub-acute discharges and rehabilitation discharges, Community Health discharges)	213	238
Hospital A: admissions	136	149
Hospital A: discharges (home or to Community Health)	65	75
Hospital A: transfers to subacute in Hospital B	71	74
Hospital B: direct admissions into sub-acute	3	5
Hospital B: discharges from sub-acute (home or to Community Health)	20	19
Hospital B: transfers from sub-acute to rehabilitation—discharge from rehabilitation	54	60
Hospital A: Community Health admissions post-discharge	2	0
Hospital B: Community Health admissions post-discharge	1	10

Table 6 Patient Demographics and Length of Stay by Service

Variable	Pre-intervention		Post-intervention		p
Charts audited N =	213		238
Males n = (%)	116 (54.5)		126 (53)
Age, mean (SD)	70 (12)		71 (11)		.49
Females n = (%)	97 (45.5%)		112 (47)
Age, mean (SD)	74 (13)		77 (11)		.07
	Male	Female	Male	Female
Days in Hospital A mean (SD)	7 (5)	6 (4)	6 (4)	6 (4)
Days in Hospital B (sub-acute and rehabilitation) mean (SD)#	29 (24)	37 (42)	29 (21)	32 (25)

Note. # Days in Hospital B unable to be split

Table 7 Post-stroke Mood Assessment by Service Location and Profession

Screening &/or interview and health professional	Mood assessment and service location	Audit 2013–2014 n =	Year 2014–2015 n =	Proportion change %	95% CI [LL, UL]	p
Charts audited		213	238
Formal mood screen used	Yes	8	42	13.85	[8.05, 19.69]	< .0001**
Evidence of discussion in notes only (i.e. no screening tool used)	Yes	23	21	1.99	[−3.79, 7.93]	.47
Total evidence of mood discussion	Yes	31	63	11.85	[4.10, 19.38]	.002**
Was low mood identified?	No	12	33	15.78	[−7.6, 36.8]	.15
	Yes	19	30	15.7	[−7.7, 36.73]	.15
Screen &/or interview	Hospital A no screen, no interview	124	147	7.5	[2.06, 13.7]	.003**
	Hospital A screen only	0	0	0	0	0
	Hospital A interview only	12	2	7.5	[2.11, 13.6]	.003**
	Hospital A screen and interview	0	0	0	0	0
	Hospital B sub-acute no screen, no interview	20	19	0	[−17.6, 16.8]	–
	Hospital B sub-acute screen only	0	0	0	0	0
	Hospital B sub-acute interview only	0	0	0	0	0
	Hospital B sub-acute screen and interview	0	0	0	0	0
	Hospital B rehab no screen, no interview	37	7	56.9	[39.1, 70.66]	<.0001**
	Hospital B rehab screen only	1	0	1.8	[−4.41, -9.8]	.29
	Hospital B rehab interview only	10	16	11.1	[−5.91, 27.2]	.16
	Hospital B rehab screen and interview	6	37	50.5	[32.78, 64.58]	<.0001**
Screen &/or interview in community	No community screen or interview	1	2	13.3	[−34.88, −73.16]	.64
	Community interview only	1	3	3.3	[−44.61, 65.12]	.91
	Community screen and interview	1	5	16.7	[−48.54, 61.78]	.62
Health profession who documented mood or discussion	Social worker	21	53	14.1	[−4.9, 35.07]	.0.11
	Doctor	2	0	6.6	[−1.52, 21.98]	0.04*
	Occupational therapist	6	0	20	[6.46, 38.56]	.0003**
	Alcohol Tobacco Other Drugs (ATODs)	1	1	1.72	[−5.92, 15.67]	0.59
	Community Team	0	9	14.2	[0.40, 25.29]	.03*
	Mental Health	0	0	0	0	0

Notes. * indicates p < .05, ** indicates p < .001

LL and UL indicate the lower and upper limits of a confidence interval respectively

Table 8 Interventions for Low Mood

Intervention	Audit year
	2013-2014 n =	2014-2015 n =	Proportion change %	95% CI [LL, UL]	Difference in proportions p
Low mood identified	19	30
Counselling	7	25	46.5	[15.81, 69.81]	.001**
Mental Health	3	13	27.6	[−2.1, 50.46]	.04*
Carer Support	0	0	0	0	0
Refer psychologist	0	2	6.6	[−11.97, 21.98]	.25
Increase social support	1	1	1.9	[−12.86, 22.88]	.74
GP follow-up	2	12	29.5	[1.11, 50.9]	.027*
Refer community SW	3	9	14.22	[−14.04, 37.24]	.26
Antidepressants	3	13	27.6	[−12.1, 50.4]	.46
Sunshine	0	4	13.3	[−6.77, 30.68]	.1
Education	0	4	13.3	[−6.77, 30.68]	.1
Diversional therapy	0	2	6.6	[−11.97, 21.98]	.25
Watchful waiting	0	1	3.3	[−14.63, 17.1]	.42
Pt declined intervention	1	0	5.2	[−7.43, 25.93]	.21
No intervention	9	1	44	[17.22, 67.99]	.0002**

Notes. * indicates p < .05, ** indicates p < .001

LL and UL indicate the lower and upper limits of a confidence interval respectively

Demographic and episode of care information for the two audit periods is reported in Table 6. The data show a higher number of males in both periods with a lower mean age on presentation in comparison with females.

The evidence of mood screening is detailed in Table 7. The table shows an overall statistically significant increase in documentation around mood screening and clinical interview in the second period compared to the first 95% CI [4.10, 19.38], p < .002, specifically consisting of an increase in the number of patients who had a clinical interview 95% CI [8.05, 19.69], p < .0001. Paradoxically, in the second period there was a significant increase in the number of patients who had no screen or interview in Hospital A (acute service) 95% CI [2.06, 13.7], p < .003 and a significant decrease in the number of patients in Hospital A who had an interview only 95% CI [2.11, 13.6], p < .003. The data show a continued absence of screening and clinical interview in the sub-acute service. Conversely, the rehabilitation service showed a significant increase in patients with a screen and interview 95% CI [32.78, 64.58], p < .0001.

Table 8 presents the interventions for low mood and shows that in the pre-implementation period, 47% of patients identified with low mood received no intervention, however in the post-implementation period, only one patient identified with low mood did not have an intervention (p = .0002). Compared with the pre-implementation period, the significant interventions for identified low mood in the post-implementation period were: counselling 95% CI [15.82, 69.81], p < .001, mental health 95% CI [−2.1, 50.46], p < .04, and GP follow-up 95% CI [1.11, 50.9], p < .027.

Discussion

The National Stroke Foundation (2010) recommends that following a stroke, all patients receive a mood screen and if low mood is detected, a clinical interview. In this project, the aim was to develop and implement Post-Stroke Mood Assessment Pathways which, in conjunction with staff training, would improve the rates of mood screening, clinical interviews and interventions for patients who have had a stroke. While the findings of the study demonstrate that the implementation of the mood pathways and staff training increased compliance from 14.5% to 26.4%, clearly, there is room for further improvement.

The development and implementation of the pathways was centred around staff training, the planning and conduct of which was based on the PARiHS framework (Kitson et al., 1998). This framework proved useful in initially deconstructing and then establishing the roles and responsibilities of the stakeholder team around the elements of the pathway implementation. The pathways did not direct therapists to specific interventions, instead they supported therapists in conducting a solution-focused conversation with patients around what would be helpful to them in gaining control over recovery. This is recognised as an important factor in the psychological care of stroke patients (Welch, 2008). Accordingly, during the staff training, therapists were encouraged to work with the patient, family, and multidisciplinary team to create an individual plan to address the patient’s needs.

Lincoln, Kneebone, Macniven, and Morris (2011) indicate that mood assessment should be conducted four weeks post-stroke. In the context of a regional health service with a specific stroke rehabilitation service, this timeline would position patients who have experienced a severe stroke (and who are still in hospital) in the sub-acute or rehabilitation unit of Hospital B. Those patients with a less acute stroke would have been discharged from Hospital A or B and would be within the community.

Overall, the implementation of the pathway and staff training did increase screening, interviews, and interventions for post-stroke patients whose ultimate discharge was from the rehabilitation unit, however there was no improvement in the number of screens and interventions for patients discharged from either the acute or sub-acute setting.

In total there were n = 182 (85%) and n = 175 (73%) missed mood assessment episodes in the two audit periods. In both years, patients discharged home from the acute service were 91% and 98% (respectively) unlikely to receive a mood assessment. This finding suggests that stroke in these patients resulted in relatively minor impairment and they were discharged rapidly (i.e., prior to the four-week post-stoke assessment period). Similarly, the data show that no sub-acute patient (in either period) received a mood assessment. The sub-acute patients were more likely to have experienced a catastrophic stroke and be transferred to a nursing home prior to the four-week post-stoke assessment period. It can be argued that patients with a minor stroke or catastrophic stroke will experience depression and anxiety, however these groups had minimal assessments conducted and thus, the incidence of low mood in these cohorts was impossible to assess. These findings are of concern as there is no evidence to suggest that patients experiencing a minor or a catastrophic stroke don’t require screening—all post-stroke patients are at risk of depression, or depression and anxiety. It is possible that the patients discharged from the acute and sub-acute areas were screened either by their general practitioners or in the nursing home, however this research did not follow up those discharge paths.

The stroke patients most likely to respond to rehabilitation were placed in the rehabilitation unit where there is a larger volume of resources and equally, an extended contact time with specialised social work and stroke care teams. Accordingly, it was the rehabilitation patients who were more likely to have a mood assessment, with the assessment rate rising from 31% in the first year to 88% in the second year. This cohort also had the greatest increase in interventions for low mood.

Limitations

The authors relied on a secondary data source (medical records) to determine whether assessments, clinical interviews and interventions were conducted. There may have been more clinical interviews and interventions around mood that were not documented. However, for those patients where a mood assessment was conducted, the authors noted that the quality of the clinical interview was much richer in the second year, however this was not separated in the research findings. Patients diagnosed with a Transient Ischemic Attack (TIA) were not included, however this may be worth considering in the future.

Conclusion and Recommendations

Depression and anxiety is prevalent in the stroke population and clinicians have a responsibility to ensure they are screening and assessing for mood changes post-stroke. Clinical handover of patients who have not been assessed in the hospital environment is a serious gap in service continuity and places potentially vulnerable patients into the nursing home and community sector without adequate follow-up. Therapists need to allow the stroke survivor time to talk through these feelings during the recovery stage. While it is normal to experience a range of emotions post-stroke, care needs to be taken not to normalise depression and significant feelings of anxiety. Both can become chronic issues if not addressed and affect long-term outcomes for stroke survivors and their family (Lincoln et al., 2011).

When implementing these pathways, no added resources were available to increase staffing to ensure all patients were screened according to the pathways. In addition, staff training was conducted once at the implementation of the pathways. This research demonstrates that by incorporating evidence-based mood screening for stroke patients into practice, health care social workers can increase the rate of screening and mood assessments. However, the findings suggest that even when the tools and pathways are present, unless mood assessment is an embedded component of the social worker role, it is unlikely to be conducted.

This research recommends that: (a) staff be specialised in stroke care, (b) mood assessment be embedded into the performance indicators of social workers who work with stroke patients, and (c) ongoing training and support be provided for the clinicians who are conducting the assessments. These recommendations will further the attainment of the National Screening Guidelines (National Stroke Foundation, 2010).

Acknowledgements

Wide Bay Hospital and Health Service staff for their support of this research, and Amanda Kratzmann for formatting of Pathways.

Disclosure Statement

No potential conflict of interest was reported by the authors.

ORCID

Angela Ratsch http://orcid.org/0000-0003-0806-6293