Predictors of long-term return-to-work at five-year follow-up for mild-to-moderate subacute stroke patients enrolled in an early supported discharge program

3.1 Study design

The ESD program was set up in 2007 by the National University Hospital (NUH) in Singapore. It is a community-based rehabilitation program which provides home therapy to help stroke patients maximize their functional outcomes and live as independently as possible in the community. The ESD program at NUH is led by a multidisciplinary team comprising a neurologist, an occupational therapist, and a physiotherapist. Interventions are mainly targeted at helping stroke patients to optimize their functional status and reintegrate into the community. All patients receive home-based treatment within 3 days of discharge that lasts from 2 to 6 weeks. Depending on their needs, patients usually receive home-based treatment by the occupational therapist and the physiotherapist for a total of 5 (minimum) to 20 (maximum) one-hour sessions. The occupational therapy interventions include ADL/IADL retraining, upper limb retraining, cognitive and perceptual retraining, and home modification advice. The physiotherapy interventions include strengthening exercises, indoor and outdoor ambulation, and stairs practice. There are weekly discussions on patients’ progress among the team, including the neurologist. It is not a specialized vocational rehabilitation program designed to target RTW specifically, but it addresses RTW goals depending on the needs of the ESD patients. Patients suffering from acute stroke with residual mild-to-moderate disability (≤3 months of stroke onset) and who have available caregivers are eligible for the program. All records of ESD patients, including demographics details, pre- and post-ESD outcomes, and one-year follow-up data, are recorded in an existing ESD standing database (registration number: NUH/2016-00027).

The present study on RTW utilized a prospective cohort design to examine the relationships between various factors, such as demographic, functional, personal, psychosocial, and work-related factors, and RTW at five-year follow-up for ESD patients. RTW was defined as return to paid employment on a full-time or part-time basis. Details of RTW were collected through questionnaires administered by the Principal Investigator (PI) via phone follow-up. Additional medical records were extracted from the ESD standing database.

3.2 Participants

Consecutive stroke patients enrolled into the ESD program between 1st January 2012 and 31st December 2014 (i.e. 4–6 years, with an average of 5 years post stroke) were identified from the ESD standing database. Patients were included in the study if they were of working age (18–60 years old) and were gainfully employed at the time of their hospitalization.

Patients who met the inclusion criteria received an invitation letter by post (in both English and Chinese) providing details of the nature and content of the research study. Using a prescribed script, the PI contacted the potential participants two weeks after the letter was posted to obtain verbal consent for their participation in the research study. Verbal consent was documented on a data collection form, and if consent was given, the PI administered the questionnaire over the phone, which took around 10 to 15 minutes. The study was approved by the Human Ethics Committee of the Hong Kong Polytechnic University (reference number: HSEARS20180118001) and the NHG Domain Specific Review Board (reference number: 2018/00037) in Singapore for research involving human subjects.

3.3 Data collection

The following data were extracted from the ESD standing database: age, race, gender, type of stroke, medical history, length of hospital stay, inpatient functional independence measure (FIM) score, and pre- and post-ESD outcome measure scores, including FIM, to evaluate functional status [10]; the Frenchay Activities Index (FAI) to measure participation in instrumental activities of daily living [11]; Motricity Index (MI) to evaluate motor impairment [12]; Modified Rankin Scale (mRS) to measure the extent of disability after stroke [13]; level of assistance and aids required for ambulation; and FAI scores at one-year follow-up. Sociodemographic details, such as age, gender, race, and number of people working in full-time paid employment before stroke onset, were recorded at time of discharge from NUH. Comorbidities, including diabetes mellitus (DM), hypertension (HTN), hyperlipidemia (HL), atrial fibrillation (AF), ischemic heart disease (IHD), and the presence of repeated stroke; type of stroke, and length of stay in hospital (in days) were also recorded (Table 1).

An RTW questionnaire was designed and the content of the questionnaire was established after a comprehensive literature search on issues relating to RTW. The content was further refined and simplified after seeking the opinions of three experts in occupational therapy with a minimum of two years of working experience in stroke rehabilitation. The format of the questionnaire was mainly closed-ended multiple choice questions or one-word answers, except for one question which involved getting the participants to rate, on a scale of 0 to 100, their ability to perform at work. The questionnaire contained a total of 21 questions that included years of education, whether they had returned to work, duration to RTW, time sustained in current job, intention to resign, number of job changes within the last five years, type of services received to RTW, barriers to and facilitators of RTW, and self-perceived capacity to RTW. In addition, we also asked participants about their previous and current types of jobs, with “white collar” work defined as spending a large proportion of time engaged in performing administrative duties or holding office jobs and “blue collar” work as jobs that involve mainly manual labor [14]. The number of questions answered by patients who had returned to work differed from the number answered by those who had not. To achieve as many responses as possible, the PI called the patients at different times of the day (e.g. during lunch hours or after work hours) and at weekends over a period of 3 to 4 weeks.

The data were analyzed using IBM SPSS Statistics 24. The baselines of the demographics between patients who had and had not returned to work at five-year follow-up were compared using Fisher’s exact test and independent samples T-test for the categorical and continuous variables respectively. Logistic regression (stepwise forward) was used to identify the predictors of RTW at five-year follow-up for all patients. Odd ratios (ORs) with a 95% confidence interval (CI) were calculated to estimate the association between the predictors and RTW. Linear regression (forward selection) was used to identify the predictors of perceived ability to RTW at five-year follow-up for those who had returned to work successfully. Statistical significance was set at p≤0.05.

Survival analysis was conducted on the basis of the censored data of time until first job upon RTW, for which the Kaplan-Meier method was used. Then, the survival analysis was stratified further by age (young-young <50 and old-young ≥50) and disability (without disability (FIM score >100) and mild-to-moderate disability (FIM score 54–100)) [10]. Comparison of the survival curves of the two groups within each stratification was conducted using the log-rank test. Cox regression was used to identify predictors of RTW at five-year follow-up. Hazard ratios (HRs) with a 95% CI were computed to estimate the relative risk between the predictors and RTW. To maximize adequate data for analysis, the ‘last observation carried forward’ (LOCF) and the mean substitution (MS) methods were adopted for missing data less than 10 to 15 percent [15].

4.1 Participants

Of the 69 patients who were initially identified from the database and recruited, six (8.7%) had passed away by the time of study, 11 (15.9%) remained uncontactable, and two (2.9%) declined to participate in the study (refer to Fig. 1). Fifty patients (72.5%) who completed the questionnaires were included in the analysis. However, one patient had missing pre- and post-ESD outcome measures and hence was excluded from the final analysis (n = 49). The proportion of missing data for the post-ESD outcome measures and the questionnaires was 4% and 8.2%, respectively; they were treated using LOCF and MS, respectively. There were only 30 patients who had completed the FAI scores at one-year follow-up.

OPEN IN VIEWER

Fig. 1

Flowchart of participants through the study.

The characteristics of all the patients (n = 49), including those who did and did not RTW, are presented in Table 1. The majority of the patients were male (75.5%) and Chinese (63.3%). The mean age of the participants was 50 (±7.4), the mean years of education was 10.1 (±3.2), and the average length of stay in hospital was 13.8 days (±11.9). Ischemic stroke was observed in 77.6% of the study population, hypertension in 44.9%, diabetes mellitus in 30.6%, and hyperlipidemia in 26.5%. There were no significant differences in demographics, pre- and post-ESD outcome measures, and FAI one-year follow-up between the RTW group and the non-RTW group.

Details on the patients who had returned to work (n = 33) are reported in Table 2. In the RTW group, 90.9% of the participants were the breadwinner of their family, with the majority earning an average income of between SGD $1 to $1999 (42.3%) and taking an average mean time of 5.7 (±4.6) months to RTW; 78.7% of them had returned to their “premorbid job”, with more patients having blue collar jobs (63.6%). Mean time in current job and the mean average of job changes in last five years were 46 (±20) months and 0.2 (±0.6), respectively, with only 6.1% expressing an intention to resign from their jobs in the next three months; 78.8% had received services to return to their current job, with hospital-provided services (96%) and alternative medicine (e.g. traditional chinese medicine or malay massage) (30.8%) being the two most common services received. Only 12.1% received assistance to help them sustain their current job, with adjustment to workload being the most common intervention (75%) given when they initially returned to work after stroke. At five-year follow-up, the patients gave a rating of 84.1 (±14.5) out of 100 for their perceived capacity to perform at work. The most common reason for returning to work was having support from family and friends (75.8%) (Table 3a), while the most common reason for not returning to work was physical impairments (40.8%) (Table 3b). The three factors ranked highest for a successful RTW were, in order of priority, 1) having support from family and friends; 2) financial factors, such as having to support family; and 3) having support from employers and coworkers (Fig. 2a). On the other hand, the three factors ranked highest as barriers to RTW were, in order of priority, 1) physical impairment, 2) cognitive impairment, and 3) social problems (Fig. 2b).

OPEN IN VIEWER

Fig. 2

Factors for RTW or not RTW after stroke. Note: RTW: return-to-work.

The median time from survival to RTW was 7 months, as shown by the Kaplan-Meier survival curve (Fig. 3a). The time to RTW was further divided into age group (Fig. 3b) and disability group (Fig. 3c). A statistically significance difference was found between the young-young group (median time to RTW = 6 months) and the old-young group (median time to RTW = 18 months), with p = 0.044, while only a marginally difference was found between the no disability group (median time to RTW = 5.7 months) and the mild-to-moderate disability group (median time to RTW = 12 months) at admission to hospital, with p = 0.074.

OPEN IN VIEWER

Fig. 3

Time to RTW using survival analysis. Note: RTW: return-to-work.

Using a logistic regression for RTW at five years post stroke (Table 4), social problems, such as stigmatization, lack of support from employers, family and coworkers, etc., was found to be a negative predictor of RTW (OR 0.02; 95% CI 0.002–0.22), while being the breadwinner was a positive predictor of RTW (OR 13.79; 95% CI 2.46–77.52). Similarly, with regards to the Cox regression for RTW at five years post stroke (Table 4), having social problems was negatively associated with RTW (HR 0.09; 95% CI 0.02–0.34), while being the breadwinner was positively associated with RTW (HR 4.07; 95% CI 1.74–9.51). Regarding the linear regression for perceived capability to perform at work (Table 4), the presence of cognitive impairment emerged as a significant predictor, with p = 0.001.