Abstract 2466: Inter-observer Reliability and Validity of Pre-morbid Modified Rankin Scale
Background and Purpose: The modified Rankin Scale (mRS) is the most prevalent outcome scale in stroke trials. Utility of standard mRS is limited by inter-observer variability. Pre-morbid function, described using mRS, is often used to determine eligibility for clinical trials or interventions. The clinimetric properties of pre-morbid mRS have not been described previously. We assessed the hypothesis that pre-morbid mRS would have acceptable reliability and validity for clinical use.
Methods: Over a four week period (June 2011), four researchers assessed consenting, sequential stroke unit admissions across two University Hospitals. Paired interviewers (trained in mRS) were randomised to perform independent and blinded mRS assessment of patients. Clinical and demographic details were collated independent of interview. Inter-observer variability was calculated for pre-morbid and standard mRS using kappa (k) and weighted kappa (kw) with 95% confidence interval (95%CI) and percentage agreement. Validity was assessed by comparing pre-morbid mRS with other markers of function. Number of medications and comorbidity index were compared to pre-morbid mRS (average of paired scores) using rank correlation; proportions living alone with no external care were compared for those with pre-morbid mRS≤1 versus pre-morbid mRS>1 (chi-square test).
Results: Seventy-four stroke survivors were assessed (four proxy assessments); including a variety of stroke types (TACS:13; PACS:27; POCS:9; LACS:15). Median age 72 years (IQR:62-79); median time since event 5 days (IQR:3-9). Median standard mRS was 4 (IQR:2-4). Median pre-morbid mRS was 1 (IQR:0-3; range:0-4); 27 (38%) patients had premorbid disability (mRS>1). Reliability for standard mRS interview was: 56% agreement; k=0.40 (95%CI:0.27-0.52); kw=0.55 (95%CI:0.39-0.71). Reliability of pre-morbid mRS was: 70% agreement; k=0.58 (95%CI:0.46-0.70); kw=0.70.(95%CI:0.53-0.87) Spearman’s Rho for pre-morbid mRS and comorbidity was 0.31 (95%CI:0.08-0.50); for pre-morbid mRS and number of medications was 0.33 (95%CI:0.11-0.52). There was no association between need for carers and pre-morbid mRS (p=0.10).
Conclusions: Inter-observer reliability of pre-morbid mRS is limited but comparable to standard mRS. Poor correlation between markers of previous function and pre-morbid mRS suggest suboptimal validity. Based on these data premorbid mRS may not be a suitable trial entry criterion. There is scope for improvement, improved training and guidance specific to premorbid mRS is a potential solution.
- © 2012 by American Heart Association, Inc.