Reliability of the modified Rankin Scale in clinical practice of stroke units and rehabilitation wards

INTRODUCTION: The Modified Rankin Scale (mRS) is the most common tool to quantify post-stroke disability in everyday practice and by certified raters in clinical trials. However, interobserver variability may affect reliability of retrospective observational studies, including clinical registries. Our aim was to assess real-life consistency between neurologists and physical and rehabilitation medicine physicians using mRS to rate post-stroke disability of patients transferred directly from stroke unit (SU) to rehabilitation ward (RW). METHODS: This is a retrospective analysis of 132 consecutive acute stroke patients transferred from single tertiary SU to RW located in the same hospital in Poland. Patients were assessed by one rater from each department at the day of transfer. We distinguished between physicians previously certified in using mRS for clinical trials and not-certified physicians using mRS in everyday practice. RESULTS: mRS at discharge from SU and on admission to RW was recorded for 105 of 132 patients. The overall agreement was 70.5% (kappa 0.55). Similar agreement was observed in the subset of 30 patients rated by certified physicians in both departments (70.0%, kappa 0.57) and in the subset of 61 patients rated by a pair of certified neurologist and not-certified rehabilitation physician (73.8%, kappa 0.58). CONCLUSIONS: Everyday consistency between raters from SU and RW in using mRS is modest as in previous validation studies. However, it may be considered sufficient for the purpose of observational studies or stroke registries. It emphasizes the need for easily accessible training in conventional mRS or implementation of specialized tools with predefined questions.