This study had a prospective longitudinal cohort design, with inclusion and baseline assessment within the first 2 weeks after stroke onset and a follow-up assessment 3 months later (±2 weeks).

Participants were treated in an evidence-based comprehensive stroke unit, with a special focus on independence in daily life after a stroke (22). The stroke unit has a multidisciplinary approach focusing on early mobilization, physiological homeostasis, nutrition and fluid intake to provide best possible treatment (23, 24). While hospitalized, participants were screened for inclusion into the study by an experienced physiotherapist. Recruitment took place 3 days a week. Only patients available for inclusion on one of these days were invited to participate.

All baseline assessments were performed during hospital stay within 1 day from inclusion in the study. This included the baseline assessments, the independent variable (the Brief Resilience Scale), and all covariates included in the regression analysis. Participants were invited back to the hospital for the follow-up assessment at the outpatient clinic 3 months later. All assessments were performed by the same physiotherapist at both time points.

In line with the Norwegian guidelines for treatment and rehabilitation after stroke (22), participants were discharged from the stroke unit in accordance with their physical and cognitive level for further follow-up in the primary health care system to either a rehabilitation center, outpatient clinic or at home with or without organized home-based treatment.

Patients admitted to the stroke unit at St Olavs Hospital, Trondheim University Hospital in Norway were asked for their participation. Patients with first-ever or recurrent acute ischemic or hemorrhagic stroke were eligible for inclusion if the onset of stroke was within 14 days post-stroke, and their modified Rankin Scale (mRS) score (25) was 0–4 points. Aphasia or language problems were accepted for inclusion as long as the patients scored 4–6 points on the item “orientation” on the Scandinavian Stroke Scale (SSS) (26), and they were capable of providing informed consent. However, persons with a life expectancy of <6 months, such as advanced and progressive cancer and unstable heart conditions, or serious impairments that could have significant impact on a functional outcome or unstable medical condition after an acute stroke were excluded.

At baseline, age and sex, days from stroke to inclusion and the degree of disability prior to the stroke (measure by the mRS) were scored. mRS is ranging from 0 (no symptoms) to 5 (severe disability), while 6 denotes death. The scale has shown strong validity and reliability (27). From the medical records of the participants the type and location of stroke, severity of the stroke measured by the SSS, and possible reperfusion treatment were assessed. The SSS, ranging from 0 (worst) to 58 (best) has shown to be a valid measure of stroke severity (28, 29).

The baseline assessments also included the degree of disability, measured by the mRS and the Barthel Index (BI) and fatigue measured by the Fatigue Severity Scale (FSS-7). Fatigue is reported to be up to three times higher three months after a stroke than in the general population (30). FSS-7 has shown to be reliable and valid in the Norwegian stroke population (31, 32). Each item was scored on a 7-point Likert scale, with a total FSS score ranging from 7 (best) to 49 (worst). The total score was averaged to yield a score from 1.0 to 7.0. Higher scores indicate higher fatigue levels. FSS-7 was completed by self-report.

The Barthel Index obtained at 3-month follow-up was added as the dependent variable to assess independence in basic ADL, ranging from 0 (worst) to 100 (best) (33). Barthel Index has proven good reliability in the stroke population (34), however the documented ceiling effect is a weakness (35).

The Brief Resilience Scale (BRS) obtained at baseline was used to measure the ability to cope with a serious event. It consists of six items with five different response categories, giving a total score ranging from 6–30. The total score is further divided by the number of items answered giving an average score of 1 (worst) to 5 (best). BRS has shown to be a reliable tool to assess an individuals' ability to recover from stressful circumstances (36, 37). The BRS used in its this project was the original version in English. The BRS questionnaire was completed in a collaboration between the participant and the physiotherapist.

Covariates included in the regression analysis were identified and selected from previous literature. These were age, sex, fatigue at baseline, degree of independence prior to stroke, and stroke severity at admission (38–40).

All data were analyzed using IBM SPSS Statistics version 25. Unless otherwise stated, demographic data were reported as mean values and standard deviation (SD) for all participants, retrieved from the participants‘ medical record. Residuals were visually inspected for normal distribution by Q-Q plots with both dependent and independent variables showing a normal distribution. Homoscedasticity was controlled through a scatter plot of the residuals. Paired sample t-tests were applied to study changes in ADL and resilience from baseline to 3 months later. Linear regression was used to analyze associations with Barthel Index at 3 months as the dependent variable and resilience at baseline as independent variables in both bivariate and multivariate regression analysis. Age, sex, fatigue at baseline, stroke severity at admission and the modified Rankin Scale prior to inclusion were added as covariates. The covariates were selected based on clinical judgement and the literature. We checked for collinearity with VIF <10 and correlation values below 0.9. Significance level was set at p < 0.05 for all associations.

This is the first study to assess the association between resilience in the early phase and basic ADL 3 months later in a stroke population. Assessing resilience within 2 weeks after a stroke is assumed to give highly valid response to the ability to bounce back from a serious event, which should be considered a major strength of this study. Another strength was that all included participants underwent treatment in an evidence-based stroke unit with a focus on early rehabilitation to regain independence in basic ADL. The multidisciplinary approach with focus on physiological homeostasis, nutrition, and fluid intake and also on early mobilization and goal achievement (23, 24), might have facilitated positive coping strategies in a stressful situation (45).

Ninety-eight subjects accepting participation represented about 15% of patients admitted to the stroke unit during the recruitment period. The low recruitment rate was caused by patients not meeting our inclusion criteria, failing to be screened inside the inclusion window, or refused participation. Another important reason for the low inclusion rate was because the physiotherapist responsible for inclusion was working part-time, which might have increased the risk of failing to include patients with a short hospital stay. Unfortunately, we do not have a detailed screening log reporting the specific reason for being excluded from the study. The low recruitment rate might represent a selection bias. However, comparison with data from the Norwegian stroke registry suggests that our participants were comparable to most stroke survivors in Norway regarding stroke severity and degree of disability (46). Of the 98 participants first included in the study, 34 (34%) were lost to follow-up. However, secondary analysis of baseline characteristics of age, sex, modified Rankin Scale prior to the stroke, stroke severity, functional impairment, and independence in basic ADL did not show any clinical or statistical difference between those included and excluded from the study. Therefore, we would argue that the participants included in the final analysis were representative of those originally included in the study.

We applied rather broad inclusion criteria to ensure inclusion of a population representing the heterogeneity of stroke survivors. However, in line with several other stroke studies (47, 48) the baseline characteristics (Table 1) suggest that participants had only mild-to-moderate disability after stroke, and with the ceiling effect of the Barthel Index after 3 months further underlines this (Figure 3). The ceiling effect is likely to influence on our results with a low variance making associations less likely. This should be considered when interpreting our results. Including stroke survivors with more severe strokes and without the ability to move independently could have given different results. Even though, the Barthel Index has been widely used as an assessment tool for ADL function for stroke survivors, adding a more extensive and responsive measure like the Functional Independent Measure (49) or the Stroke Impact Scale (50) would be more appropriate to applied in a future study.

The inclusion criteria allowed both first time- and recurrent strokes to be included in the study. Participants with a recurrent stroke may have suffered from impairments such as movement disorders or fatigue prior to testing, which could influence on the results. This variation has been accounted for by adjusting for pre-stroke function as measured by the modified Rankin Scale in the multivariate regression analysis. However, in future studies with a larger sample size, a sub-group analysis differentiating between first ever and recurrent strokes could add additional information about sub-groups of stroke patients.

The relatively wide inclusion window ranging from 1 to 12 days after stroke onset may influence on the results, as spontaneous recovery is at its highest the first days and weeks after a stroke. This inclusion window was chosen to recruit as many participants as possible. However, the average day from stroke onset to inclusion was 4.3 (SD 2.8) days, indicating that most participants were included within the first week after the stroke.

The linear regression was chosen instead of dichotomizing Barthel Index in order to maintain statistical power given the relatively small sample size. We chose to add 5 different covariates to the analysis. These were chosen based on the literature and clinical judgment. Sex has previously been questioned as a predictor of ADL outcome after a stroke (51). However, the bivariate analysis showed a borderline significance level (p = 0.049), which was an argument to include sex as a covariate in the multivariate analysis.

In conclusion, our results showed that early measurement of resilience was not associated with independence in basic activities of daily living 3 months after stroke, indicating that resilience is a personal trait which do not have an impact on the outcome of physical adversity. However, future research should investigate whether resilience is related to the outcomes of psychosocial adversity after stroke.