The aim of this observational study is to perform a temporal and geographical external validation of the EPOS (Early Prediction of Outcome after Stroke) model for the prediction of independent gait after stroke. The EPOS model measures the early presence of leg strength on the affected side and sitting ability to predict recovery of independent walking six months after stroke. Compared to the EPOS model development study, the prediction time point of independent gait in this study will be three months rather than six months post-stroke and the patients will be more heterogeneous. Due to the differences in the new cohort, it is expected that the performance of the models will be lower than in the development cohort, but still be adequate.

Worldwide, stroke is the second leading cause of death and is also the third leading cause of death and disability combined. According to the Global Burden of Disease, Injuries, and Risk Factors Study (GBD) by the Collaborator Network on the topic of stroke, the absolute number of incidence strokes and their deaths have steadily increased in recent years. The most important risk factors include high blood pressure, a high body-mass index, high plasma glucose, fine dust pollution and smoking. In Switzerland, an average of 20'423 people suffered a stroke per year between 2016 and 2020. Due to the growth and ageing of the population, the number of hospitalizations due to cardiovascular diseases is increasing, but deaths have decreased over time. According to a recent study, 75% of stroke survivors are able to walk independently after three months. On average, independent walking is achieved six days post-stroke whereby stroke severity is most strongly associated with recovery of gait.

In recent years, a large number of studies have been published showing that prognosis is of growing interest in stroke rehabilitation. Medical prognostic models are used to inform patients and their families about the course of their disease and to plan possible treatments. Early prognosis of recovery after stroke enables early discharge planning and efficient use of health care resources. In addition, therapy goals can be realistically formulated and thus recommendations for care or adaptations of the home environment can be made. Currently, prognostic models are rarely used in clinical practice. This is partly because most of the models have not been externally validated, which is a prerequisite for implementation in practice. External validation involves using new data from a similar population to investigate the generalizability of the model. In summary, a good and useful prognostic model should be accurate, generalizable and ideally clinically effective.

A narrative review described six prognostic models for independent gait after stroke. Only three of these models measure predictor and outcome variables at a specifically defined time point: the Australian model, the TWIST model and the EPOS model. The Australian model and the TWIST model have been updated but not externally validated and the EPOS model has been externally validated once. According to the review above, the EPOS model seems to be a promising model for predicting independent gait after a stroke. This is due to the early and fixed time point of simple and fast clinical tests as well as the prognosis time point of six months. The EPOS model shows that early presence of lower limb strength and sitting ability can predict recovery of independent walking six months after stroke. Predictive variables were measured using the Trunk Control Test with the item "sitting balance" (TCT-s) and the Motricity Index leg score (MI leg). The dependent outcome variable was the Functional Ambulation Categories (FAC). In the external validation, the dependent outcome variable FAC was measured after three months. The results show good performance of the model from the third day onwards for predicting independent gait three months after stroke.

In summary, these findings raise the question of whether the use of the EPOS model in post-stroke patients within the first 72 hours is accurate and generalizable for predicting independent walking three months after stroke. Although this first external validation shows promising results, further external validation studies need to be undertaken before the model can be applied in clinical practice. The aim of this study is to perform a temporal and geographical external validation of the EPOS prediction model for independent gait after stroke. Compared to the development study, this study will be conducted in Switzerland and the prediction of independent gait will be based on the time point of three months rather than six months post-stroke. In addition, all post-stroke patients will be included, regardless of the type of stroke, the affected circulation or positive stroke history. Furthermore, patients with pre-existing comorbidities are also being considered, as long as they were ambulatory before the stroke.

Inclusion Criteria:

1. aged 18 years or older
2. having suffered an ischemic or hemorrhagic stroke, confirmed by computerized tomography and/ or magnetic resonance imaging
3. being unable to walk independently (FAC < 4) within 72 hours
4. giving written informed consent to participate

Exclusion Criteria:

1. not able to walk independently before hospital admission (FAC < 4)
2. clinically significant comorbidities that prevent walking ability
3. deficits in cognition or communication, so consent cannot be given for study participation

admitted to hospital with an acute stroke