The purpose of the study is to develop a predictive model for return to work after stroke rehabilitation.

Background Return to work post stroke is a key milestone for many survivors of stroke; however, many cannot achieve this goal. Based on the previous work by Tay et al, independent predictive factors were identified for return to work post inpatient stroke rehabilitation. In further works by Koh and Tay with the same dataset put through different models such as LASSO-Full, Lasso-Routine, ElasticNet-Full, etc suggests that the good discrimination performance of the return-to-work prediction models during internal validation supports a multisite, external validation study. Performing a temporal validation study is the interim step. A smaller part of the dataset could be used to further train the development model.

Significant economic costs are incurred because of stroke, which do not include further economic costs from the downstream loss of earnings and caregiver burden. Existing prediction models of return-to-work have area under the receiver operating characteristic curves (AUROCs) between 0.65 and 0.80. However, these models have not been assessed for calibration or clinical utility. No externally validated prediction model exists for return-to-work after stroke. There are advantages to having a prediction model. One of the concerns of patients and their families involve the loss of income as a result of stroke. The prediction model would help to prognosticate, as well as assist to set appropriate rehabilitation goals for the patient. Suitable patients can be directed to return to work services, if necessary.

Health is related to one's employment and financial position. Being able to return to gainful employment can result in better general and mental health5. People with disabilities employed in the past year reported better general and mental health than their peers with the same disabilities who were unemployed.

This study could be completed in 1 to 2 years and a prospective study involving external validation can be simultaneously performed with NUHS collaborators over 2 to 2.5 years. Both projects could be published in the next 2 to 3 years upon obtaining the grant for the temporal validation study. Thereafter, a Return to work calculator could be designed and launched in the next 4 years.

Hypothesis This study seeks to collect 1375 patient data who had completed inpatient stroke rehabilitation between the time periods of 2018-2025. We seek to further train our development model and to perform a temporal validation on this model.

Inclusion Criteria:

1. Completed inpatient stroke rehabilitation in CGH
2. Diagnosis of a first ever stroke
3. Patients who were working prior to stroke
4. Consent given

Exclusion Criteria:

1. Not first ever stroke
2. Did not require inpatient stroke rehabilitation
3. Not working prior to stroke
4. No consent obtained

The criteria for recruitment is first-ever stroke and premorbid work.

Data is protected under the SingHealth Data Protection Policy which is compliant to the Singapore Personal Data Protection Act 2012