Running head: Patient activation in stroke rehabilitation Patient activation during the first 6 months after the start of stroke rehabilitation

Objective – To examine patient activation from the start of stroke rehabilitation and its course up until the 6 month follow-up. Design – Inception cohort study with a follow-up of 6 months. Setting – Multidisciplinary rehabilitation facility. Participants – 478 stroke patients who received inpatient or outpatient rehabilitation with a median age of 63.0 years (inter quartile range (IQR) 56.0-70.0 years) with 308 (64.2%) being male. The study was completed by 439 (91.8%) patients. Interventions – Not applicable. Main Outcome Measures – Patient activation was measured with the Patient Activation Measure (PAM, score 0-100, 4 levels, where a higher score and level denotes more patient activation). The PAM was measured at the start of the rehabilitation (baseline), and 3 and 6 months thereafter, and analysed using the multivariate mixed model analysis. Results – At baseline, the mean PAM score was 60.2 (SD 14.3), with the number of patients in PAM level 1, 2, 3 and 4 being 76 (17.8%), 85 (19.9%), 177 (41.4%) and 90 (21.0%), respectively. The multivariate mixed-model analysis demonstrated that the PAM score increased over time (baseline 60.2 (Standard Deviation (SD) 14.3) versus 3 months 60.7 (SD 14.8) versus 6 months 61.9 (SD 18.0), p 0.007). Between baseline and 6 months, 122 (41.4%) patients remained at the same PAM level, 105 (35.6%) patients increased, and 68 (23.1%) decreased. At all time points, >35% of patients were in level 1 or 2. Conclusion – PAM scores increased slightly over time from the start of rehabilitation up to the 6 month follow-up. However, more than a third of patients remained at low levels (i.e. level 1 and 2) of patient activation, which indicates that specific interventions during rehabilitation to increase patient activation might be of value.

Stroke is a common health problem worldwide, leading 50% of patients to develop a chronic condition with a combination of motor, communication, cognitive, or emotional limitations. [1][2][3][4][5] In patients with chronic conditions, such as stroke, self-management is of great importance. 6 Self-management refers to the strategies, decisions and activities individuals take to manage a long-term health condition. 7 Specifically in stroke patients, three sub-domains of selfmanagement strategies can be distinguished; focusing on prevention of a secondary stroke, adherence to exercises, and enhancement of participation and activities of daily living. 6 A review has shown that adding training for these self-management strategies during stroke rehabilitation can improve activities of daily living and independence. 8 In order to use self-management strategies, patient activation is a prerequisite. 9 Patient activation is defined as one's role in the care process and having the knowledge, skills, and confidence to manage one's health and healthcare. A review demonstrated that patients with chronic conditions who are more activated have better health outcomes and better care experiences than those who are less activated. However, patients with stroke were not included. 10 Until now there is only one questionnaire that measures patient activation: the Patient Activation Measure (PAM). 11 The PAM distinguishes passive patients who experience no influence on their health status from active patients who do experience this influence.
Although having a sufficient level of activation is important for stroke patients, research on this topic in stroke patients is scarce. To our knowledge, there are only a few studies done in community-based 12,13,14,15 or hospital-based 16 stroke patients. These studies show different levels of patient activation, varying between a level where patients are disengaged and overwhelmed 16 to a level where patients are maintaining behaviours and pushing further 12 .
Increasing patient activation during stroke rehabilitation is not explicitly included in stroke rehabilitation guidelines as a treatment goal. 17,18 Consequently, stroke rehabilitation is mainly aimed at improving limitations after stroke and is not specifically aimed at increasing patient activation. 19 We therefore hypothesized that patient activation does not improve or only slightly improve during and after stroke rehabilitation. Therefore, the aim of this prospective observational study is to examine patient activation at the start of the rehabilitation, and the course of patient activation up until the 6 month follow-up.

Study design
This study was part of the Stroke Cohort Outcomes of Rehabilitation (SCORE) study, a cohort study in a rehabilitation facility, which started in March 2014 and ended in December 2019. This study has been described extensively elsewhere. 20 The protocol of the study is registered in The Netherlands Trial Register. This study is reported according to the STrengthening the Reporting of OBservational studies in Epidemiology (STROBE) guidelines. 21

Study population
Consecutive stroke patients who received inpatient or outpatient multidisciplinary rehabilitation were invited by the rehabilitation physician to participate in the SCORE study when they: 1) were 18 years or older, 2) had a first or recurrent stroke less than 6 months prior, 3) had no psychiatric disorder or dementia and 4) were able to complete questionnaires in Dutch. After patients were checked for their eligibility, were willing to participate and provided written informed consent, they were included in the study.

Procedure
The protocol of the study was approved by the Medical Ethics Board of the Leiden University Medical Center (NL465321.058.13).
Patients filled in questionnaires on paper or online, depending on their preference. When there was no response within 10 days, patients were contacted by telephone or e-mail, with a maximum of two reminders.

Assessments
At the start of the rehabilitation (i.e. baseline) baseline characteristics and patient reported outcome measures (PROMs) were collected.

Baseline characteristics
Age, sex and type of stroke (ischemic/haemorrhagic) were extracted from the patients' medical file. A questionnaire was used to assess the level of education (6-point scale which was split into 3 categories according to the Dutch system, i.e. low, medium, high), living situation (married or living with a partner), and paid work before stroke and the number of comorbidities (by the Dutch study on Life Situation Questionnaire 22 ). Questions about lifestyle prior to stroke included smoking (≥ 1 cigarette per day), alcohol (≥2 glasses per day), and physical activity (30 minutes of moderate to intensive daily physical activity.
A nurse assessed the Barthel Index at baseline only in patients receiving inpatient rehabilitation. This is a measure of functional independence with a score ranging from 0 to 20, where higher scores indicate more functional independence. 23
The Stroke Impact Scale (SIS) 25 was used to measure self-reported impact of stroke on the domains mobility, communication, memory and thinking, and mood and emotions.
Summative scores for each domain range from 0-100, where higher scores indicate better functionality.

Patient activation
Patient activation was assessed at baseline, at 3 months and at 6 months follow-up by means of the PAM. 11 This generic measure consists of 13 items, with ratings on a 5-point Likert-type scale (disagree strongly, disagree, agree, agree strongly and N/A). Total scores range from 0-100, where higher scores denote higher patient activation. 9 The PAM score can be divided in four progressively higher activation levels. Patients at level 1 (score 0.0-47.0) may not yet understand that their role is important. Patients at level 2 (score 47.1-55.1) lack confidence and knowledge to take action. Patients at level 3 (55. 2-72.4) are beginning to take action, whereas patients at level 4 (72.5-100) are proactive about health and take action to perform many recommended health behaviours. 26 The Dutch version of the PAM has shown adequate psychometric properties in people with a chronic illness. 27 In persons with neurological conditions (not stroke patients) the PAM was found to have good internal reliability and to be valid for research purposes. 28

Statistical analyses
Data were analysed with SPSS Statistics for Windows, Version 22.0 (IBM Corp, Armonk, NY, USA). Data were presented descriptively. A p-value of 0.05 was considered statistically significant.
To analyse whether there were differences in baseline characteristics between patients with paired measurements on the PAM at baseline and 6 months and patients without paired measurements, Mann-Whitney U-tests, Fisher's Exact tests, and Chi 2 tests were used, where appropriate. The same tests were used to compare all patients included in the current analyses and patients who were excluded in the current analyses (because they did not complete the PAM or were outliers).
Baseline characteristics of patients at the 4 PAM levels were compared using Chi 2 tests and Kruskal Wallis tests. Post-hoc tests with Bonferroni correction to correct for multiple testing were performed in case of significant differences. To evaluate the course of PAM levels for individual patients, descriptive statistics were used.
For patients who filled in the PAM at baseline and at 6 months, PAM levels at these time points were graphically shown in a Sankey diagram.

Results
Between March 2016 and December 2019, 506 stroke patients were included in the SCORE study ( Figure 1). Of them, 28 (5.5%) were excluded from the current analyses because 26 did not complete a PAM at any time point, and 2 had a maximal PAM score of 100 at one time point and a minimal PAM score of 0 at another time point. The frequency of an ischemic stroke was lower in these excluded patients than in the included 478 patients (64.3% versus 82.1%, p<0.001). Other characteristics were not significantly different between these groups (results not shown). Table 1 Table 2).

PAM scores over time
At 3 month follow-up, 367 patients completed the PAM with a mean score of 60.7 (SD 14.8) and at 6 months 335 patients had a mean score of 61.9 (SD 18.0) ( Table 3). In the univariate analysis, the PAM score did not significantly improve over time (β=0.80 (95% Confidence Interval (CI) -0.14-1.73), p0.094) ( Table 4). Further analysis of the significantly related covariates showed that old age and worse communication, memory and thinking and mood and emotions had a negative effect on the PAM score as a function of time. In the multivariate analysis, including the significant related covariates, the PAM score did improve over time  [13][14][15] . Moreover, the mean PAM score in the present study was much lower than the mean score (75.3) of stroke patients in the study of Kidd et al. 12 The authors stated that patient activation was probably lower based on interviews with these patients. 12 In contrast, the mean PAM score in the present study was higher than in a cross-sectional study with stroke patients from a tertiary hospital (60.2 versus 51.56). 16 The authors hypothesized that this low patient activation might be due to underdeveloped health literacy and healthcare awareness. 16 A strength of this study is that it gives insight in the course of patient activation in stroke patients during the first 3 months after stroke when most recovery takes place. 29 Also up until 6 months when it is thought that a plateau effect is reached. 29 Another strength of our study is that PAM levels are described. This information at the individual patients' level of knowledge and skill to self-manage, allows physicians and therapists to target self-care education and provide support for each patient's needs while presumably being more effective in supporting patient's self-management. 9 Previous studies found that a low level of patient activation was associated with low income, using less preventive screening measures (e.g. health screening), unhealthy behaviours (e.g. smoking), worse clinical indicators (e.g. systolic blood pressure), more visits to the emergency department, more admissions to the hospital 30 , more unmet medical needs and inappropriate use of the healthcare system 31 . In contrast to these previous studies, patients in PAM level 1 did not report significantly more unhealthy behaviours pre-stroke. However, they did have more comorbidities than patients in PAM level 4. Moreover, patients in level 1 had lower HRQoL, lower self-rated communication, memory and thinking and mood and emotions compared to patients in level 4. In other words, patients who are more severely affected by their stroke, have a lower level of patient activation. Furthermore, the number of patients with a low level of activation (level 1 and 2) was >35% at all time points. In addition, the PAM score decreased markedly in a number of patients over time. This subgroup of patients may specifically need attention and support. For a patient with a low level of activation, it could be of value to introduce a tailored intervention on those aspects of patient activation that he/she has difficulty with. In case the level of activation of patients in level 1 does not improve, the care they receive might be more directed to compensation strategies. Patients in level 2 and 3 might benefit from interventions targeted at patient activation as a part of rehabilitation. Interventions were proven to be effective in increasing patient activation in patients with diabetes and other chronic conditions and the highest increase was seen in patients with the lowest activation levels. 10,32 In stroke patients, 3 different interventions were studied, which aimed at improving patient activation. [13][14][15] Of the three only 1 was found significantly effective. 15 This intervention was a home-based social worker-led case management program combined with a website providing stroke-related information. However, the exact mechanisms remain uncertain. 15 These interventions have not yet been tested in more affected stroke patients who receive rehabilitation. This should be addressed in future research.

Study limitations
Since the PAM has not yet been validated specifically in stroke patients, this can be considered a limitation of this study. Based on our data and two previous studies 12, 28 , there is some doubt regarding the content validity of the PAM, i.e. the degree to which the content of an instrument is an adequate reflection of the construct to be measured, looking at relevance of the items, as well as comprehensiveness and comprehensibility. 33 In our study, two patients (0.4%) had a maximum score of 100 and a minimum score of 0 at another time point and 11 patients (3.8%) increased or decreased 3 levels between baseline and after 6 months. It is unclear whether these patients were truly differently activated or whether there was a problem with comprehensibility due to cognitive or communicative limitations. These doubts are further substantiated by the study of Kidd et al. 12 where there seemed differences in patient activation described by PAM scores and interviews and a study done in a population with neurological conditions which showed that individual activation levels were underestimated due to differences in item difficulties 28 . This advocates for validation of the PAM in a population of stroke patients who receive rehabilitation. Since the Minimal Important Change (MIC) of the PAM in stroke patients is unknown, it was not possible to interpret whether the slight improvement observed in the present study is perceived as an important change by stroke patients. This advocates for determining the MIC of the PAM in stroke patients.
A larger percentage of patients with haemorrhagic stroke were excluded from our analysis.
Although the percentage of excluded patients was low (5.5%), we cannot preclude that this could have influenced the generalisability of our results. Furthermore, the 293 patients with paired measurements on the PAM at baseline and at 6 months differed significantly from the 185 who did not have paired measurements on age, living situation, smoking and mobility.
Therefore, the course of PAM levels might not be generalisable to the whole population.

Conclusion
The mean PAM score in stroke patients increased over time, but only slightly. Moreover, about a third of patients remained at low levels of patient activation and patients decreased in their level of patient activation. This indicates that there is room for improvement since no specific interventions for increasing patient activation are part of current rehabilitation treatment. Further research is needed to determine the effectiveness of interventions to improve patient activation for this specific population.