Abstract TP310: Inter- Rater Reliability of Telephonic Modified Rankin Scale Obtained for a Telestroke Network

Introduction: The modified Rankin Scale (mRS) has been established as a standard measurement of stroke disability and is widely used in major stroke clinical trials. Previous studies compared inter-rater reliability between face-to-face and telephone assessments; we aimed to evaluate the reliability of the mRS between two rater’s assessments both conducted by telephone. Objective: Identify the inter-rater reliability of mRS of patients interviewed twice by mRS-certified raters over telephone. Methods: Patients seen by telemedicine (TM) between 03/2017-04/2017 at UTHealth Lone Star Stroke Consortium spoke sites with discharge diagnosis of ischemic or hemorrhagic stroke and contacted via telephone. They underwent a structured telephone interview conducted by two trained raters of the UTHealth TM research team. The 2 interviews were conducted within 2 weeks of each other. Scoring agreement was measured using the κ statistic. Results: Total of 65 patients were in-window for 90 day mRS telephone interviews; excluding deaths, mRS of 30 patients were collected by both interviewers. The overall agree ment between two raters was 43.3% (13 cases). It was most common to disagree by one category (10 cases, 33.3%). The unweighted kappa value (κ) for inter-rater agreement was 0.29 (p = 0.0006), and the weighted kappa value (κ w ) was 0.50 (p = 0.0000). Conclusion: Unlike previous studies, we found reliability of mRS obtained via structured telephone interview in our TM population is uncertain. The inter-rater reliability of mRS assessed by telephone shows discrepancies between interviewers, as >50% of patients did not receive the same score. Our study is limited by small sample size and further studies are needed to confirm our findings. With clinical trials relying on telephone mRS for longitudinal measurement of patient outcome and some requiring mRS on the same patient multiple times during a study, it is important mRS be both reliable and reproducible between different raters.