Abstract
This longitudinal observational study examined how individual versus illness belief schema compare as predictors of post-stroke recovery. A total of 42 stroke survivors (mean age = 66.9 years/range = 29–96 years; 68% male) were involved. The primary outcome, Health-Related Quality of Life was measured using EQ-5D-5L, mood using Patient Health Questionnaire-9 and disability using Nottingham Extended Activities of Daily Living Scale. Stroke Illness Perception Questionnaire-Revised measured illness beliefs. Linear regressions showed that individual illness beliefs significantly explained more of the variance in 3-month post-stroke recovery than schema (7.4%–22.5% versus 1.9%–9.9%). Individual versus illness belief schema predict outcomes differently, but which approach predicts outcomes better remains unclear.
Keywords
Introduction
The Common Sense Model (CSM) suggests that when confronted with a threat to their health, such as the common cold, individuals experience a disequilibrium that they are motivated to resolve and do so by constructing beliefs about their illness and treatment that inform what they do to cope and feel better (Leventhal et al., 1980).
Illness beliefs have five core domains (Leventhal et al., 1980). ‘Identity’ describes perceptions about the label of the illness, for example, a cold, and associated symptoms (e.g. cough, sore throat etc.). ‘Timeline’ refers to beliefs about an acute or chronic illness duration. ‘Consequences’ refers to beliefs about the seriousness of the disease and daily impact (e.g. pain, time-off work). ‘Cure/control’ refers to perceptions about the amenability of the illness to cure/prevention/treatment. ‘Causes’ describes views about internal (e.g. genes) or external (e.g. a germ or virus) causes of the illness. There is a wealth of evidence supporting these domains in a broad range of illnesses, including cancer, hypertension and diabetes mellitus (e.g. Meyer et al., 1985). These domains have since been extended to include the following: ‘timeline-cyclical’ beliefs around an episodic nature of illness; ‘personal control’ and ‘treatment control’ beliefs around peoples’ own ability and that of their treatment (e.g. tablets) to manage their illness; ‘illness coherence’ beliefs around understanding of the illness; and ‘emotional representations’ referring to illness-related distress (Moss-Morris et al., 2002).
The CSM has been used widely for understanding treatment adherence, mood and recovery in a range of acute and chronic illness groups. This study focuses on stroke.
Stroke is an example of a condition where recovery is individual and highly variable (Lazar and Antoniello, 2008) and depends upon good patient adherence to complex long-term treatments (e.g. medication, lifestyle management and rehabilitation). Post-stroke recovery and treatment adherence can be explained in part by non-modifiable socio-demographic (e.g. age, income) and clinical variables (e.g. stroke severity and initial impairment) (Bushnell et al., 2011; Lazar and Antoniello, 2008). However, socio-demographic variables predict outcomes very weakly. In a longitudinal observational study involving 180 survivors 1 year post-stroke, O’Carroll et al. (2011) found that modifiable psychological factors (depression and treatment beliefs) explain significantly more (around 30%) of the additional variance in medication adherence.
Several systematic reviews and meta-analyses have synthesised the CSM literature to examine the relationship of individual illness belief domains with coping/treatment adherence. For example, Hagger and Orbell (2003) examined 45 studies across a range of acute and chronic conditions and demonstrated that illness beliefs are inter-related, and these so-called illness belief schema are correlated with coping and illness outcomes. These findings were recently replicated for individuals with chronic illnesses (Hagger et al., 2017). However, Aujla et al. (2016) and Brandes and Mullan (2014) both showed that when examined individually, illness belief domains predict adherence to self-management in people with acute or chronic physical illnesses very weakly (<0.13).
Illness belief schema are a key premise of the CSM; they are stored in peoples’ memory and help them to understand and respond to situations (Leventhal et al., 1980). There are now several studies utilising cluster analysis to group individuals with physical illnesses according to their illness belief schema (e.g. Medley et al., 2010; Lin and Heidrich, 2012; and Snell et al., 2014). While these studies have not involved stroke survivors, illness belief schema have consistently been shown to predict physical and psychological outcomes and coping/treatment adherence.
In summary, illness beliefs outlined by the CSM are inter-related and held in schema. When examined individually, illness beliefs predict outcomes, such as adherence, weakly. We argued that the low effect sizes to have emerged in our earlier review were a result of examining the predictive utility of individual CSM components rather than of the model as a whole. Therefore, we posit that individual and illness belief schema are likely to differ in their prediction of outcomes, where schema may have greater predictive power.
A previous study has examined this hypothesis and showed that individual and illness belief schema perform comparably when predicting outcomes in people with type 2 diabetes mellitus (Skinner et al., 2011). However, this study did not include all illness belief domains. Specifically, treatment control beliefs and emotional representations were excluded, both of which are important predictors of post-stroke outcomes after stroke (e.g. mood (Twiddy et al., 2012) and medication adherence (O’Carroll et al., 2011)). In addition, the authors did not report on statistical comparisons between regression models containing individual versus illness belief schema (i.e. clusters), for example, goodness of fit, using the change in R2 and likelihood ratio test. However, by carrying out these comparisons, one would be able to discern how well variables (or groups of variables) in each regression model perform and therefore judge which approach (individual versus illness belief schema) better predicts (or explains more of the variability) in outcome(s). To our knowledge, this analysis has not yet been undertaken.
This study has three aims. First, to examine whether stroke survivors can be grouped according to different illness belief schema. Second, to conduct a preliminary enquiry into how individual versus illness belief schema may compare as predictors of 3-month post-stroke recovery (Health-Related Quality of Life (HRQL), mood and disability). Third, to explore how much of the additional variability in post-stroke recovery is explained by individual versus illness belief schema, over and above that explained by socio-demographic and clinical variables.
Methods
Design and participants
A longitudinal observational design was employed and reported in accordance with the appropriate guidelines. Data were collected at baseline (after study enrolment) and follow-up (3 months post-stroke). Participants were recruited from consecutive admissions to acute stroke and rehabilitation wards and from outpatient clinics in one hospital in Nottingham, UK. Inclusion criteria were as follows: adults (>18 years); confirmed diagnosis of acute stroke (within 8 weeks); sufficient (English) language and cognitive ability to participate, assessed through liaison with the clinical care team and review of medical notes; and written informed consent. We excluded individuals involved in psychological intervention studies. Ethical approval was granted by the National Research Ethics Service (NRES) Committee East Midlands (UK) – Leicester (13/EM/0392).
We initially collected socio-demographic information, including age, sex, ethnicity, education/employment and social deprivation (using the English Indices of Multiple Deprivation (IMD) (Department for Communities and Local Government, 2015)). Medical history, including stroke severity (National Institutes of Health Stroke Scale (NIHSS) (Duncan et al., 2005)), disability (Modified Rankin Scale (Rankin, 1957) and Barthel Index (Mahoney and Barthel, 1965), and comorbidities, was also gathered. In addition, we collected clinical information (blood pressure, blood glucose, cholesterol and body mass index (BMI)); family history (first-degree relative); and lifestyle data (smoking status, alcohol consumption, physical activity and diet). Participants also completed the study measures (elaborated below) at baseline and follow-up. Most of the data were self-reported. Clinician-reported data (e.g. stroke severity, blood tests and BMI) are collected routinely locally and were therefore extracted from patients’ medical records.
Eligibility was assessed in 1085 patients over a 12-month period. In all, 88 were eligible. The main reasons for non-eligibility were as follows: non-stroke diagnosis (N = 249), stroke onset over 8 weeks before (N = 186) and enrolled in other research studies (N = 141). In all, 38 people declined and 50 consented, and attrition was 16 per cent. Therefore, we had baseline data from 46 and follow-up data from 42 participants. In brief, average age was 66.9 years (standard deviation (SD) = 14.5 years; range 29–96 years), with 68 per cent male gender and 98 per cent White British ethnicity. IMD scores indicated that participants resided on average among the most deprived 70 per cent of areas in England. The majority of participants reported a non-severe first stroke (78%) and experienced low levels of disability and mild depressive symptomatology.
Measures
The International Classification of Functioning (ICF) framework for health and disability highlights important outcomes to assess in post-stroke recovery: ‘impairments’ – problems or loss in body function; ‘activities’ – performance of a task or action; and ‘participation’ – involvement in a life situation (World Health Organization (WHO), 2001). We selected our outcomes according to these domains. The choice of measures was guided by a series of papers (Salter et al., 2005a; Salter et al., 2005b; and Salter et al., 2005c) that synthesised the most commonly used instruments across each ICF domain and critically reviewed their psychometric and administrative properties.
ICF participation: HRQL
We measured HRQL using EQ-5D-5L (Brooks, 1996). The first part, termed the ‘Descriptive System’, comprises five items assessing Mobility, Self-Care, Usual Activities, Pain/Discomfort and Anxiety/Depression domains of HRQL. Possible responses were ‘no problems’, ‘slight problems’, ‘moderate problems’, ‘severe problems’, to ‘unable’. The second part, termed the ‘Your health today’ Visual Analogue Scale (VAS), comprises a picture of a thermometer with a scale ranging from 0 (worst) to 100 (best), and respondents mark whereabouts on this scale they perceive their health to be on that day. Scores are converted to a percentage, where high scores represent the best health imaginable (Van Hout et al., 2012). Responses on the ‘Descriptive System’ are considered to be health states, which are converted to a single index value, and interpreted as follows: 0 = dead, 1 = full health and <0 = worse than death (Van Hout et al., 2012). The EQ-5D-5L is valid for stroke survivors (Golicki et al., 2015) and showed good internal consistency in our study (Cronbach’s alpha (α) = 0.84).
ICF impairments: mood
Post-stroke mood was measured using the Patient Health Questionnaire (PHQ)-9. PHQ-9 includes nine items measuring the frequency of depressive symptomatology, according to nine of the main criteria for major depressive disorder defined by the American Psychiatric Association Diagnostic and Statistical Manual of Mental Disorders (DSM; Kroenke et al., 2001). Respondents recall over a period of 2 weeks, the frequency with which they have been bothered by various symptoms (e.g. ‘Little interest or pleasure in doing things’). Possible responses were ‘Not at all’, ‘Several days’, ‘More than half the days’, to ‘Nearly every day’. A total PHQ-9 score was obtained by summing scores on the individual items (Kroenke et al., 2001). High scores indicated a greater severity of depressive symptomatology. PHQ-9 has recently been advocated as a promising approach for measuring mood among stroke survivors (Meader et al., 2014). Internal consistency was good in this study (α = 0.72).
ICF activities: disability
The Nottingham Extended Activities of Daily Living Scale provided a stroke-specific measure of instrumental activities of daily living after stroke (i.e. disability; Nouri and Lincoln, 1987). It comprises 22 items covering everyday activities that respondents have performed in the ‘last few weeks’ (e.g. ‘walk around outside’, ‘climb stairs’). Possible responses were ‘Not at all’, ‘With help’, ‘On your own with difficulty’, to ‘On your own’. A total score was obtained by summing scores on the individual items, where high scores represent greater independence/recovery from post-stroke disability (Nouri and Lincoln, 1987). This instrument is valid and reliable for stroke survivors (Nouri and Lincoln, 1987) and had good internal consistency in our study (α = 0.93).
Illness beliefs
Illness beliefs were assessed using a stroke-specific version of the IPQ-R (Stroke IPQ-R; Aujla et al., 2018). This measures nine domains of illness beliefs. In total, 21 items assessed illness identity. Yes/no responses for each item (e.g. ‘Weakness or paralysis in arm or leg’) were summed to obtain a total score for stroke-related symptom burden. A total of 66 items assessed the remainder of domains (timeline acute–chronic, consequences, timeline-cyclical, personal control, treatment control, illness coherence, emotional representations and causal beliefs). Possible responses were ‘strongly disagree’, ‘disagree’, ‘I don’t know’, ‘agree’ and ‘strongly agree’ (Moss-Morris et al., 2002). ‘The effects of my stroke will last for a short time’ is an example question for the timeline acute-chronic domain. The remainder of questions are summarised in the work by Aujla et al. (2018). Scores for each sub-scale were generated by summating the scores for individual items. Items with negative wording were reversed. High scores on the illness identity, timeline acute–chronic, timeline-cyclical, consequences and emotional representations sub-scales indicated more negative beliefs about stroke, while those on the personal control, treatment control and illness coherence sub-scales represented more positive beliefs (Moss-Morris et al., 2002). Internal consistency for all except the treatment control sub-scale (α = 0.42) was good (α ⩾ 0.77) in our study.
Statistical analysis
The primary outcome was 3-month post-stroke HRQL – arguably the best way to elicit how well survivors are recovering from the perspective of patients themselves (Deshpande et al., 2011). The secondary outcomes were mood and disability. Sample size was informed by findings from existing CSM stroke studies (e.g. O’Carroll et al., 2011 and Twiddy et al., 2012). Based on detecting a correlation of 0.4 between illness beliefs and markers of post-stroke recovery (e.g. mood), power = 80 per cent, α = 0.05 and 20 per cent attrition, we calculated a minimum sample of 55 participants.
The analyses were conducted in STATA 13 (StataCorp LP, College Station, TX, USA). In view of our low participant numbers, our analyses should be considered exploratory. Statistical significance was assessed at the 5 per cent level (p < 0.05). A Bonferroni adjustment corrected for multiple testing. We assessed normality using histograms and skewness/kurtosis tests. Logarithmic (base 10) transformations were attempted for skewed variables, without fruition. The mean and SD (or where data were non-normal, median and interquartile range) were calculated for all continuous variables and frequency/percentages for categorical variables. A series of non-parametric Wilcoxon signed rank tests (or parametric paired sample t tests for normal data) examined differences in our outcomes between baseline and follow-up.
The remaining analyses were undertaken in two stages. Causal beliefs (16 separate attributions) were excluded in view of our modest sample size. First, we performed hierarchical agglomerative cluster analysis, the optimal method for clustering illness beliefs data (Clatworthy et al., 2007). Illness belief scores were standardised to Z scores prior to clustering to ensure that all IPQ-R sub-scale scores had equal weight (Clatworthy et al., 2007). Clusters were formed based on similarities between participants on scores for each IPQ-R sub-scale, which were assessed using the Euclidean distance (i.e. the actual distance between two points if measured with a ruler). The optimum number of clusters was visually determined on a dendrogram (hierarchical tree-like structure). Clusters were combined using Ward’s method; the most common linkage rule used for illness beliefs data (Clatworthy et al., 2007). Participants were subsequently partitioned into ‘k’ clusters using k-medians, a partitioning method suitable for non-normal data. Cluster characteristics were summarised using the mean/SD (or median/interquartile range for non-normal data). A series of one-way analysis of variance (ANOVA) plus Scheffe post hoc comparisons (or Kruskal–Wallis for non-normal variables) examined between-cluster differences.
Second, multiple linear regression models were fitted. Socio-demographic and clinical variables were selected based on prior evidence. The analysis initially examined individual illness belief domains (not schema). The order of entry was socio-demographic variables at the first step, clinical variables at the second step and illness beliefs at the third step, as per O’Carroll et al. (2011). Models were compared using the change in R2 and likelihood ratio test, which provided a chi-square p value to assess the statistical significance of variables added at each step. The analyses were subsequently repeated for illness belief schema (step 4).
Results
Descriptive analyses
Our analyses used complete cases. Following Bonferroni adjustment, neither HRQL, mood nor disability differed significantly between baseline and follow-up (Appendix Table A). In terms of illness beliefs, participants reported few symptoms of their stroke at baseline and significantly fewer at follow-up (p < 0.001). The majority perceived their stroke to be chronic, with fluctuating effects, greatly impacting upon their lives and causing considerable distress. Most participants believed in their own ability and that of their treatment to control the effects of their stroke, though seemed to have an unsatisfactory understanding of their stroke. Participants tended to attribute their stroke to ‘Ageing’ or ‘Chance or bad luck’ rather than many of the modifiable risk factors for stroke (e.g. ‘High cholesterol’ and ‘High blood pressure’.)
Identification of illness belief schema
Three clusters emerged (see Appendix Figure A) that were named as follows: Cluster 1 – ‘Low Adjusters’; Cluster 2 – ‘Moderate Adjusters’; and Cluster 3 – ‘High Adjusters’. Each cluster was characterised by specific illness belief schema (Appendix Table B). ‘Low Adjusters’ perceived their stroke to be chronic, with fluctuating effects, associated with a lot of symptoms, serious consequences and considerable distress. ‘Moderate Adjusters’ also perceived their stroke to be chronic, with serious consequences, associated with some symptoms and distress, but believed their stroke had non-fluctuating effects. ‘High Adjusters’ perceived their stroke as an acute condition with non-fluctuating effects, associated with few symptoms, less serious consequences and little distress. All participants strongly believed in their own ability (personal control) and that of their treatment (treatment control) to control the effects of their stroke but had a less than satisfactory understanding of stroke (illness coherence). Despite statistical non-significance, these domains were retained in the schema characterising each group of stroke survivors.
The demographic and clinical characteristics of participants within each cluster are summarised in Appendix Table C. Three non-statistically significant trends were identified. First, ‘High Adjusters’ were older (mean = 72.1, SD = 15.1) than ‘Moderate Adjusters’ (mean = 69.5, SD = 12.3) and ‘Low Adjusters’ (mean = 57.2, SD = 13.5). Second, ‘High Adjusters’ were more likely to be retired, university educated, have a history of stroke or transient ischaemic attack and a first-degree relative with stroke. Third, ‘Low Adjusters’ were more likely to have had a severe stroke, history of depression/anxiety, be an ex-smoker, drink more alcohol and exercise less. ‘Low Adjusters’ were also more likely to have worse HRQL, mood and disability compared to the other clusters.
Prediction of post-stroke recovery: individual versus illness belief schema
Findings from the multiple linear regression models for each outcome are reported in this section and summarised in Table 1. The primary outcome, HRQL, will be considered first.
Summary of goodness of fit statistics for HRQL, mood and disability at each model step in.
CI: confidence interval; HRQL; Health-Related Quality of Life; PHQ-9: Patient Health Questionnaire-9; VAS: Visual Analogue Scale.
Likelihood ratio test.
Bonferroni-adjusted significance level (p < 0.02).
Compared to Low Adjusters group.
HRQL
Baseline socio-demographic and clinical variables significantly explained 51.4 per cent of the overall variance in 3-month EQ-5D-5L Descriptive System Index score and EQ-5D-5L ‘Your health today’ VAS score. When individual illness beliefs were entered, none emerged as significant independent predictors of HRQL. However, when examining the change in R2, individual illness beliefs significantly explained a further 17.5 per cent of the overall variance in 3-month EQ-5D-5L Descriptive System Index score, over and above that explained by baseline socio-demographic and clinical variables. When illness belief schema were entered, none of the clusters independently predicted HRQL. Inclusion of clusters in the model significantly explained only a further 0.60 and 5.7 per cent of the overall variance in 3-month EQ-5D-5L Descriptive System Index score and ‘Your health today’ VAS score, respectively, over and above that explained by baseline socio-demographic and clinical variables.
Mood
Baseline socio-demographic and clinical variables significantly explained 47.7 per cent of the overall variance in 3-month post-stroke mood. When individual illness beliefs were entered, only timeline-cyclical beliefs were a significant independent predictor of mood (β = 0.71, 95% confidence interval (CI) = 0.08−1.34), controlling for socio-demographic and clinical variables. When illness belief schema were entered, ‘Moderate Adjusters’ had significantly better mood at 3 months than ‘Low Adjusters’ (β = −4.74, 95% CI = −9.31−(−0.17)). In examining the change in R2, individual versus illness belief schema significantly explained an additional 22.8 and 9.9 per cent of the overall variance in 3-month mood, respectively, over and above that explained by baseline socio-demographic and clinical variables.
Disability
Baseline socio-demographic and clinical variables significantly explained 79.8 per cent of the overall variance in our other secondary outcome, 3-month post-stroke disability. Neither individual nor illness belief schema significantly predicted post-stroke disability. When examining the change in R2, individual versus illness belief schema significantly explained only an additional 7.4 and 1.9 per cent of the overall variance in 3-month disability, respectively, over and above that explained by baseline socio-demographic and clinical variables.
Discussion
Our study identified three distinct illness belief schema: ‘Low Adjusters’, ‘Moderate Adjusters’ and ‘High Adjusters’. In contrast to ‘High Adjusters’, ‘Low Adjusters’ perceived their stroke to be chronic, with fluctuating effects, associated with a lot of symptoms, serious consequences and considerable distress. We showed that when individual illness beliefs were examined in a series of multiple linear regression models, they significantly explained more of the variance in 3-month post-stroke recovery (HRQL, mood and disability) than illness belief schema, over and above that explained by baseline socio-demographic and clinical variables.
The evidence to date has consistently shown that individual and illness belief schema predict coping/treatment adherence and recovery in a range of conditions, including stroke (Hagger and Orbell, 2003; Lin and Heidrich, 2012; Medley et al., 2010; Snell et al., 2014; Twiddy et al., 2012). Our findings supported this premise. We too found that timeline-cyclical beliefs significantly predict mood 3 months after stroke (consistent with the work by Twiddy et al. (2012)), and that post-stroke mood is significantly better among ‘Moderate’ than ‘Low Adjusters’ (consistent with the work by Snell et al. (2014)). Our schema also comprised many of the illness beliefs (i.e. identity, chronicity and consequence beliefs, and emotional representations) that have emerged in prior studies.
However, unlike earlier papers (e.g. Medley et al., 2010), we did not find a statistically significant role for personal and treatment control beliefs in any of our clusters. Nor did these domains significantly predict post-stroke HRQL, mood or disability. There may be three reasons for these findings. First, our participants generally demonstrated strong personal and treatment control beliefs (i.e. ceiling effect). Second, the treatment control sub-scale of the Stroke IPQ-R had poor internal consistency. Third, the study was limited by a modest sample size, meaning a high risk of type 2 error.
Nonetheless, our study is the first to show that there are differences in the amount of variability in post-stroke recovery explained by individual versus illness belief schema. This analysis was of theoretical relevance given the literature on inter-relatedness of illness beliefs (Hagger et al., 2017; Hagger and Orbell, 2003), existence of illness belief schema (Leventhal et al., 1980) and growing interest in the use of cluster analysis to examine whether illness belief schema predict outcomes. However, given that the traditional approach to analysing illness beliefs has been to consider them individually (i.e. the same manner in which they have been measured), it was important to formally examine which approach most usefully predicts outcomes. Contrary to earlier findings by Skinner et al. (2011), our study showed that if the amount of variability explained by these two approaches is the only consideration, individual illness beliefs have the potential to outperform illness belief schema in the prediction of post-stroke outcomes.
Three notable problems exist with our analysis. The first is sample size, which limited statistical power. A second issue is generalisability of our findings. There is a need for further research to be undertaken with larger and more diverse samples of stroke survivors. As illness belief schema are different for different illnesses, comparable work in other conditions should also be carried out to see whether consistent findings emerge. The third relates to the cluster analysis approach, which is inherently a data reduction technique. However, dichotomising data in this way is likely to result in a loss of sensitivity (Skinner et al., 2011). Therefore, it is unsurprising that less of the variability in outcomes was explained by clusters rather than individual illness beliefs. Other, more sensitive, ways of analysing these data (currently unclear) are worth exploring in future research.
It is apparent though, at least in the context of stroke, that non-modifiable clinical variables still explain the largest amount of variability in post-stroke recovery (Lazar and Antoniello, 2008). However, the fact that our study showed that illness beliefs (individual or schema) explain even small to moderate amounts of variability in post-stroke recovery suggests that there is still scope to intervene and modify maladaptive beliefs to improve outcomes, as has been considered in belief-based intervention studies to date (e.g. O’Carroll et al., 2013).
Strengths and limitations
A strength of our work is that it is the first to examine whether individual or illness belief schema differ in their prediction of post-stroke recovery. In addition, our study was conducted rigorously; employed validated instruments (e.g. EQ-5D-5L); measured multiple aspects of post-stroke recovery, as recommended by the ICF framework; and utilised robust statistical methods.
Despite these strengths, our sample size was modest. This limited statistical power means that our results should be interpreted cautiously. In addition, our sample comprised individuals with a non-severe stroke, mild residual disability and strong personal and treatment control beliefs, which may be non-representative of survivors in the acute phase of stroke. This issue has also affected prior CSM stroke studies. Therefore, we recommend that future researchers examine ways in which a more diverse and representative sample of stroke survivors can be engaged in illness belief studies moving forward.
Implications and conclusion
We have demonstrated that individual versus illness belief schema differ in the amount of variability in post-stroke recovery explained. However, we cannot advocate which is the best approach for dealing with illness beliefs data henceforth. This remains an unanswered but important question to clarify as it could help to shape the way in which future belief-based interventions are developed (i.e. continue addressing individual maladaptive illness beliefs or begin dealing more holistically with maladaptive illness belief schema).
Supplemental Material
Figure_A_APPENDIX – Supplemental material for Do individual versus illness belief schema differ in the prediction of post-stroke recovery?
Supplemental material, Figure_A_APPENDIX for Do individual versus illness belief schema differ in the prediction of post-stroke recovery? by Navneet Aujla, Marion Walker, Nikola Sprigg and Kavita Vedhara in Journal of Health Psychology
Supplemental Material
Table_A_APPENDIX_FINAL – Supplemental material for Do individual versus illness belief schema differ in the prediction of post-stroke recovery?
Supplemental material, Table_A_APPENDIX_FINAL for Do individual versus illness belief schema differ in the prediction of post-stroke recovery? by Navneet Aujla, Marion Walker, Nikola Sprigg and Kavita Vedhara in Journal of Health Psychology
Supplemental Material
Table_B_APPENDIX_FINAL – Supplemental material for Do individual versus illness belief schema differ in the prediction of post-stroke recovery?
Supplemental material, Table_B_APPENDIX_FINAL for Do individual versus illness belief schema differ in the prediction of post-stroke recovery? by Navneet Aujla, Marion Walker, Nikola Sprigg and Kavita Vedhara in Journal of Health Psychology
Supplemental Material
Table_C_APPENDIX_FINAL – Supplemental material for Do individual versus illness belief schema differ in the prediction of post-stroke recovery?
Supplemental material, Table_C_APPENDIX_FINAL for Do individual versus illness belief schema differ in the prediction of post-stroke recovery? by Navneet Aujla, Marion Walker, Nikola Sprigg and Kavita Vedhara in Journal of Health Psychology
Footnotes
Acknowledgements
The authors would like to thank the participants of this research for giving up their time to be involved.
Declaration of conflicting interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was undertaken as part of a PhD Studentship awarded to N.Aujla by the University of Nottingham.
Supplemental Material
Supplemental material for this article is available online.
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
