Research Understanding long-term unmet needs in Australian survivors of stroke Nadine E. Andrew1, Monique Kilkenny1,2, Rebecca Naylor3, Tara Purvis1, Erin Lalor3, Natasha Moloczij2, Dominique A. Cadilhac1,2, on behalf of the National Stroke Foundation Background Limited data exist on the long-term needs of community-dwelling stroke survivors. We aimed to describe factors associated with the extent to which needs were met in Australian survivors of stroke. Method Multifaceted strategies were used to obtain a national sample. Adults 12+ months poststroke and living in the community participated. Needs were assessed over the domains of health, everyday living, work, leisure, social support, and finances. Multivariable negative-binomial and logistic regression were used. Results Seven hundred sixty-five survivors completed surveys. Most (84%) reported having needs that were not being fully met (median 4 of 20, Q1, Q3: 1, 9). Variations occurred based on age, residential location, time since stroke, and disability level. Multivariable results showed that having fatigue, cognition or emotional problems, decreasing age, and increased disability were associated with increasing numbers of needs not being fully met (P < 0·001). Factors associated with needs not being fully met were as follows: (1) greater disability (adjusted odds ratio: 3·4, 95% confidence interval: 1·9, 6·0) and fatigue problems (adjusted odds ratio: 2·0, 95% confidence interval: 1·1, 3·4) (health domain); (2) greater disability (adjusted odds ratio: 7·0, 95% confidence interval: 3·0, 17·0) and being one to two-years poststroke (adjusted odds ratio: 3·4, 95% confidence interval: 1·5, 7·8) (work domain); and (3) increased disability (adjusted odds ratio: 3·8, 95% confidence interval: 2·2, 6·5) and memory problems (adjusted odds ratio: 2·1, 95% confidence interval: 1·0, 4·2) (leisure domain). Conclusion The extent to which long-term needs were met was influenced by a variety of factors, particularly age, disability levels, and residential location. Changes need to be made to the way and extent to which survivors are supported following stroke. Key words: community care, outcomes, stroke care, stroke, treatment, unmet needs
Correspondence: Nadine E. Andrew*, Translational Public Health Unit, Stroke & Ageing Research Centre, Southern Clinical School, Monash University, Clayton, Melbourne, Vic. 3168, Australia. E-mail: [email protected] 1 Translational Public Health and Evaluation Unit, Stroke & Ageing Research, Southern Clinical School, Monash University, Melbourne, Victoria, Australia 2 Stroke: Public Health, Florey Institute of Neurosciences and Mental Health, Heidelberg, Victoria, Australia 3 Programs, National Stroke Foundation, Melbourne, Victoria, Australia Received: 17 February 2014; Accepted: 20 May 2014; Published online 15 July 2014 Conflict of interest: None declared. Funding: This project was commissioned and funded by the National Stroke Foundation. The Trust Company contributed funds via a grant to the NSF. Dominique A. Cadilhac is supported by an NHMRC/National Heart Foundation postdoctoral fellowship, grant number: 610302 and 1063761, and Nadine E. Andrew is supported by an NHMRC postdoctoral fellowship, grant number: 1072053. DOI: 10.1111/ijs.12325
106
Vol 9, October 2014, 106–112
Introduction Stroke is a leading cause of long-term adult disability (1). In Australia, 90% of stroke survivors live at home (2), and approximately 65% are dependent on others to help them with everyday activities (1). Quality of life two and five-years after stroke has been reported by many stroke survivors as poor (3,4), and many experience activity limitation, restricted social participation, and psychological issues such as anxiety and depression many years after their stroke (5,6). Long-term problems are often a consequence of survivors struggling to come to terms with their new condition and integrating back into the community (7). In particular, mood changes have been frequently reported and found to be associated with loss of independence, inability to resume previously enjoyed activities, and an inability to fulfill previous roles (8–11). Stroke survivors have also reported that health systems are not responsive to their changing needs and that there is a lack of long-term re-assessment of their needs (12). Despite the growing numbers of stroke survivors and the recognized need to support survivors and their carers living in the community, there is little research in this area. The authors of recent literature reviews have concluded that research in this area lacks a systematic approach to problem identification, has a poor evidence base, and is not underpinned by sound theoretical concepts (13–15). In 2010, the United Kingdom Stroke Association released a report which was used to identify the extent to which the long-term needs of 799 community-dwelling stroke survivors were being met (16). The UK survey highlighted that the longterm health needs of stroke survivors were not being met and that the consequences of stroke were impacting many facets of survivors’ lives. However, the extent to which these results reflect the needs of stroke survivors in other countries and impact on domains other than health is unknown. Commissioned by the National Stroke Foundation (NSF), we aimed to describe, from the perspectives of Australian community-based stroke survivors: (1) levels of long-term needs not being fully met (minimum one-year poststroke); (2) the greatest areas where needs were not fully met; (3) factors associated with the increased numbers of long-term needs not being fully met; and (4) factors associated with the proportion of stroke survivors that reported needs not being fully met in the domains of health, everyday living, work, leisure, social support, and finances.
Methods Survey design and development The survey was designed based on a review of the UK Stroke Survivors Needs Survey (16). Many of the health questions in our survey were consistent with the UK version. The survey was © 2014 World Stroke Organization
N. E. Andrew et al. adapted for the Australian context and restructured so that needs could be assessed across additional domains such as leisure, work, everyday needs, social support, and finances. Input was provided from an expert Advisory group established and chaired by the NSF and consisting of representatives from Monash University, the UK research team, general practice, stroke research, and consumers. Demographic variables and questions regarding participants’ current health status related to stroke were collected. Open-ended questions were also provided at the end of each section. Quantitative data were collected using structured questions with Likert scale of responses. The survey was pilot tested with a sample of volunteers from the NSF StrokeConnect group (n = 15) which is an NSF membership program for stroke survivors and carers. The final survey contained 58 survivor questions. A carers section was also included and contained questions such as carer demographics, the impact that taking on a carer role had on relationships leisure and work, and the extent to which their support needs were being met (data not reported). A copy of the questionnaire can be obtained by e-mail from the NSF (admin@ strokefoundation.com.au). Recruitment and data collection Participants were recruited over a 12-month period from December 2011 to December 2012. Eligibility criteria were: (1) aged 18 years or over; (2) living in the community; (3) a clinical diagnosis of stroke; (4) had their first stroke at least one-year prior to survey completion; and (5) able to complete a survey with or without
Research help from a friend, family member, or carer. Stroke survivors were encouraged to complete the survey themselves. However, proxy responses were accepted so as not to exclude survivors with severe disability. A two-phase, multifaceted, cross-sectional sampling strategy was used to recruit a national sample of stroke survivors. Phase one utilized available recruitment sources. These were: • the Australian Stroke Clinical Registry (AuSCR – www.auscr .com.au), a collaborative national quality clinical registry. Stroke survivors from participating hospitals are recruited using an optoff consent process. Registrants who elected to be part of further research studies at their 90+ day follow-up interview were invited to participate in the survey • NSF Stroke Connect members, Stroke Connect online members, and Stroke Support group members (www.strokefoundation .com.au/afterstroke/get-help). These are support programs for stroke survivors, carers, friends, and family, and • the NSF website which provided a link to an online version of the survey At the completion of phase one, an interim analysis was performed to identify gaps in our recruitment coverage based on area of residence (Fig. 1). Hospitals were purposefully selected to ensure a national sample and rural/urban coverage. Hospital staff sent surveys to previously admitted patients. The number of surveys distributed per hospital was based on the number of annual stroke admissions for that hospital and the number of completed surveys needed to achieve representative coverage
Fig. 1 Recruitment strategy flow diagram. © 2014 World Stroke Organization
Vol 9, October 2014, 106–112
107
Research for that region. Indirect recruitment strategies such as snowballing and public advertising campaigns were also utilized. To maximize participation, the survey could be completed online, using a paper-based version, or by telephone interview. For the direct recruitment strategies, paper-based surveys were sent to eligible participants, but options for online or telephone completion were provided. The NSF staff provided surveys to organizations involved in recruitment, completed surveys were scanned by staff at the Florey Institute of Neurosciences and Mental Health, and data analysis was performed by the Monash University research team. Ethics The project was approved by the Monash University Human Research Ethics Committee (approval number: CF11/2579 – 2011001510). Individual ethics approval was obtained from the HREC responsible for each of the hospitals that participated in the project (n = 18). And approval was granted by the AuSCR Research Task Group and Management Committee. Definitions We used the definition of ‘unmet’ needs described by McKevitt et al. (17) as‘something or help from someone that would help you to overcome some of the effects of your stroke and resulting difficulties’. Responses to the question ‘have you had enough help with. . . .’ were: (a) yes definitely; (b) to some extent; (c) no I did not get enough help; (d) I did not want help; or (e) I did not have difficulties. A need was defined as ‘fully met’ if the response to the question was ‘(a) yes definitely’, and defined as ‘not fully met’ if the response was ‘(b) yes, to some extent’ or ‘(c) no, I did not get enough help’. The denominator for all proportions calculated only included those with needs in the area being described. The responses ‘I did not want help’ and ‘I did not have any problems with’ indicated that the participant did not have a need in that area. Postcodes were mapped to the Accessibility/Remoteness Index for Australia (ARIA) (18). The ARIA is used to calculate geographic remoteness with reference to accessibility to 201 service centers based on road distance. Each postcode is given an ARIA value which is grouped into: (1) major cities; (2) inner regional; (3) outer regional; (4) remote; and (5) very remote. Categories 3, 4, and 5 were combined and defined as remote to provide sufficient numbers for sub-group analyses. For the logistic regression models, they were dichotomized into major cities and regional/ remote areas. Needing help with activities of daily living (ADLs) was used to define level of disability, and age was divided into those less than 65 years or 65 years or over. Analyses Descriptive statistics were used. The chi-squared statistical test for differences in proportions was used to compare groups and the Wilcoxon–Mann–Whitney rank sum test was used to compare continuous variables. Multivariable negative-binomial regression was used to assess for factors associated with the number of needs not being fully met. Logistic regression was used to assess for factors associated with the proportion of participants that reported having needs that were not fully met across each of the six domains. Independent variables were selected based on statis-
108
Vol 9, October 2014, 106–112
N. E. Andrew et al. tical significance of P < 0·1 from initial univariable analyses. In models where large numbers of variables were associated with the outcome, backwards stepwise regression was used to obtain the final models. All models were tested for collinearity and a condition index of 10–15 was considered acceptable (19). Age was modeled as a continuous variable for the negative binomial regression, but dichotomized for the logistic regression analyses for ease of interpretation. A P value of
Correspondence: Nadine E. Andrew*, Translational Public Health Unit, Stroke & Ageing Research Centre, Southern Clinical School, Monash University, Clayton, Melbourne, Vic. 3168, Australia. E-mail: [email protected] 1 Translational Public Health and Evaluation Unit, Stroke & Ageing Research, Southern Clinical School, Monash University, Melbourne, Victoria, Australia 2 Stroke: Public Health, Florey Institute of Neurosciences and Mental Health, Heidelberg, Victoria, Australia 3 Programs, National Stroke Foundation, Melbourne, Victoria, Australia Received: 17 February 2014; Accepted: 20 May 2014; Published online 15 July 2014 Conflict of interest: None declared. Funding: This project was commissioned and funded by the National Stroke Foundation. The Trust Company contributed funds via a grant to the NSF. Dominique A. Cadilhac is supported by an NHMRC/National Heart Foundation postdoctoral fellowship, grant number: 610302 and 1063761, and Nadine E. Andrew is supported by an NHMRC postdoctoral fellowship, grant number: 1072053. DOI: 10.1111/ijs.12325
106
Vol 9, October 2014, 106–112
Introduction Stroke is a leading cause of long-term adult disability (1). In Australia, 90% of stroke survivors live at home (2), and approximately 65% are dependent on others to help them with everyday activities (1). Quality of life two and five-years after stroke has been reported by many stroke survivors as poor (3,4), and many experience activity limitation, restricted social participation, and psychological issues such as anxiety and depression many years after their stroke (5,6). Long-term problems are often a consequence of survivors struggling to come to terms with their new condition and integrating back into the community (7). In particular, mood changes have been frequently reported and found to be associated with loss of independence, inability to resume previously enjoyed activities, and an inability to fulfill previous roles (8–11). Stroke survivors have also reported that health systems are not responsive to their changing needs and that there is a lack of long-term re-assessment of their needs (12). Despite the growing numbers of stroke survivors and the recognized need to support survivors and their carers living in the community, there is little research in this area. The authors of recent literature reviews have concluded that research in this area lacks a systematic approach to problem identification, has a poor evidence base, and is not underpinned by sound theoretical concepts (13–15). In 2010, the United Kingdom Stroke Association released a report which was used to identify the extent to which the long-term needs of 799 community-dwelling stroke survivors were being met (16). The UK survey highlighted that the longterm health needs of stroke survivors were not being met and that the consequences of stroke were impacting many facets of survivors’ lives. However, the extent to which these results reflect the needs of stroke survivors in other countries and impact on domains other than health is unknown. Commissioned by the National Stroke Foundation (NSF), we aimed to describe, from the perspectives of Australian community-based stroke survivors: (1) levels of long-term needs not being fully met (minimum one-year poststroke); (2) the greatest areas where needs were not fully met; (3) factors associated with the increased numbers of long-term needs not being fully met; and (4) factors associated with the proportion of stroke survivors that reported needs not being fully met in the domains of health, everyday living, work, leisure, social support, and finances.
Methods Survey design and development The survey was designed based on a review of the UK Stroke Survivors Needs Survey (16). Many of the health questions in our survey were consistent with the UK version. The survey was © 2014 World Stroke Organization
N. E. Andrew et al. adapted for the Australian context and restructured so that needs could be assessed across additional domains such as leisure, work, everyday needs, social support, and finances. Input was provided from an expert Advisory group established and chaired by the NSF and consisting of representatives from Monash University, the UK research team, general practice, stroke research, and consumers. Demographic variables and questions regarding participants’ current health status related to stroke were collected. Open-ended questions were also provided at the end of each section. Quantitative data were collected using structured questions with Likert scale of responses. The survey was pilot tested with a sample of volunteers from the NSF StrokeConnect group (n = 15) which is an NSF membership program for stroke survivors and carers. The final survey contained 58 survivor questions. A carers section was also included and contained questions such as carer demographics, the impact that taking on a carer role had on relationships leisure and work, and the extent to which their support needs were being met (data not reported). A copy of the questionnaire can be obtained by e-mail from the NSF (admin@ strokefoundation.com.au). Recruitment and data collection Participants were recruited over a 12-month period from December 2011 to December 2012. Eligibility criteria were: (1) aged 18 years or over; (2) living in the community; (3) a clinical diagnosis of stroke; (4) had their first stroke at least one-year prior to survey completion; and (5) able to complete a survey with or without
Research help from a friend, family member, or carer. Stroke survivors were encouraged to complete the survey themselves. However, proxy responses were accepted so as not to exclude survivors with severe disability. A two-phase, multifaceted, cross-sectional sampling strategy was used to recruit a national sample of stroke survivors. Phase one utilized available recruitment sources. These were: • the Australian Stroke Clinical Registry (AuSCR – www.auscr .com.au), a collaborative national quality clinical registry. Stroke survivors from participating hospitals are recruited using an optoff consent process. Registrants who elected to be part of further research studies at their 90+ day follow-up interview were invited to participate in the survey • NSF Stroke Connect members, Stroke Connect online members, and Stroke Support group members (www.strokefoundation .com.au/afterstroke/get-help). These are support programs for stroke survivors, carers, friends, and family, and • the NSF website which provided a link to an online version of the survey At the completion of phase one, an interim analysis was performed to identify gaps in our recruitment coverage based on area of residence (Fig. 1). Hospitals were purposefully selected to ensure a national sample and rural/urban coverage. Hospital staff sent surveys to previously admitted patients. The number of surveys distributed per hospital was based on the number of annual stroke admissions for that hospital and the number of completed surveys needed to achieve representative coverage
Fig. 1 Recruitment strategy flow diagram. © 2014 World Stroke Organization
Vol 9, October 2014, 106–112
107
Research for that region. Indirect recruitment strategies such as snowballing and public advertising campaigns were also utilized. To maximize participation, the survey could be completed online, using a paper-based version, or by telephone interview. For the direct recruitment strategies, paper-based surveys were sent to eligible participants, but options for online or telephone completion were provided. The NSF staff provided surveys to organizations involved in recruitment, completed surveys were scanned by staff at the Florey Institute of Neurosciences and Mental Health, and data analysis was performed by the Monash University research team. Ethics The project was approved by the Monash University Human Research Ethics Committee (approval number: CF11/2579 – 2011001510). Individual ethics approval was obtained from the HREC responsible for each of the hospitals that participated in the project (n = 18). And approval was granted by the AuSCR Research Task Group and Management Committee. Definitions We used the definition of ‘unmet’ needs described by McKevitt et al. (17) as‘something or help from someone that would help you to overcome some of the effects of your stroke and resulting difficulties’. Responses to the question ‘have you had enough help with. . . .’ were: (a) yes definitely; (b) to some extent; (c) no I did not get enough help; (d) I did not want help; or (e) I did not have difficulties. A need was defined as ‘fully met’ if the response to the question was ‘(a) yes definitely’, and defined as ‘not fully met’ if the response was ‘(b) yes, to some extent’ or ‘(c) no, I did not get enough help’. The denominator for all proportions calculated only included those with needs in the area being described. The responses ‘I did not want help’ and ‘I did not have any problems with’ indicated that the participant did not have a need in that area. Postcodes were mapped to the Accessibility/Remoteness Index for Australia (ARIA) (18). The ARIA is used to calculate geographic remoteness with reference to accessibility to 201 service centers based on road distance. Each postcode is given an ARIA value which is grouped into: (1) major cities; (2) inner regional; (3) outer regional; (4) remote; and (5) very remote. Categories 3, 4, and 5 were combined and defined as remote to provide sufficient numbers for sub-group analyses. For the logistic regression models, they were dichotomized into major cities and regional/ remote areas. Needing help with activities of daily living (ADLs) was used to define level of disability, and age was divided into those less than 65 years or 65 years or over. Analyses Descriptive statistics were used. The chi-squared statistical test for differences in proportions was used to compare groups and the Wilcoxon–Mann–Whitney rank sum test was used to compare continuous variables. Multivariable negative-binomial regression was used to assess for factors associated with the number of needs not being fully met. Logistic regression was used to assess for factors associated with the proportion of participants that reported having needs that were not fully met across each of the six domains. Independent variables were selected based on statis-
108
Vol 9, October 2014, 106–112
N. E. Andrew et al. tical significance of P < 0·1 from initial univariable analyses. In models where large numbers of variables were associated with the outcome, backwards stepwise regression was used to obtain the final models. All models were tested for collinearity and a condition index of 10–15 was considered acceptable (19). Age was modeled as a continuous variable for the negative binomial regression, but dichotomized for the logistic regression analyses for ease of interpretation. A P value of
