The real-life reliability of the modified Rankin scale used in a stroke unit and a rehabilitation ward

Purpose:
The modified Rankin scale (mRS) is the gold standard for measuring stroke-related disability in clinical trials and everyday practice. However, inter-observer variability is a source of bias that may undermine the reliability of retrospective studies. It may also depend on clinical backgrounds of different assessing physician. Our aim was to prospectively assess real-life consistency between stroke unit physicians (SUPs) and physical and rehabilitation medicine physicians (PRMPs) using mRS in patients transferred directly from a stroke unit (SU) to a rehabilitation ward (RW).

Methods:
We enrolled 48 consented stroke patients transferred within the same hospital from SU to RW. Patients were scored in mRS by a SUP and a PRMP at the day of transfer as a standard of care. The reference mRS score (REF) was obtained by a single- blinded stroke physician using the Rankin Focused Assessment form to guide an interview.

Results:
An mRS score was reported for all patients admitted to the RW and 33 patients discharged from the SU. The overall agreement was 75.8% between the assessments of SUPs and PRMPs (κ = 0.58), 72.7% between SUPs and the REF (κ = 0.55) and 70.0% between PRMPs and the REF (κ = 0.49). A similar agreement was observed for PRMPs and the REF in the sensitivity cohort of 48 patients (66.7%; κ = 0.46). Patients with the REF mRS of 2 (n = 6) were often scored as 3 both by SUPs (4/6) and PRMPs (5/6). In patients with the REF mRS of 3 or 4 there was no clear tendency towards overrating disability.

Conclusions:
The reliability of mRS scoring in everyday practice is modest and does not seem to depend on clinical background of assessing physician.