The modified Rankin scale (mRS) is commonly used as a clinical outcome measure in aneurysm trials, but inter–observer reliability in treated patients has not been tested.

We reviewed the literature on inter–observer reliability studies of the mRS. Sixty patients with ruptured (n=47) or unruptured (n=13) aneurysms treated with endovascular methods (n=34) or surgical clipping (n=26) were independently evaluated by a neurosurgeon, a stroke neurologist, and a novice research assistant, and a simplified mRS score assigned. Results were analyzed using Gwet's AC_1/2 reliability coefficients (K_G).

No previous reports validating the reliability of the mRS in treated aneurysm patients were identified. Using the mRS 0–5, inter–rater agreement was almost perfect (K_G=0.89 [0.86–0.93]). Agreement between raters remained almost perfect regardless of the rater's expertise. Agreement was almost perfect (K_G=0.87 [0.77–0.96] when the mRS was dichotomized 0–2 vs 3–5, but fell to moderate when dichotomized 0–1 vs 2–5 (K_G=0.59 (0.42–0.75). Agreement using the 0–2 vs 3–5 dichotomized mRS remained almost perfect for coiled (K_G=0.90), clipped (K_G=0.82), ruptured (K_G=0.84), and unruptured (K_G=0.95) aneurysms. Dichotomization of results at 0–1 vs 2–5 would have resulted in an (undesirable) significant difference in good outcomes between raters (P=.003), but not at 0–2 vs 3–5 (P=.52).

The simplified mRS appears to be a reliable clinical outcome measure for treated cerebral aneurysm patients. When needed, dichotomization is more reliable at mRS 0–2 vs 3–5 than at 0–1 vs 2–5. The simplified mRS is a promising tool in the functional assessment of aneurysm patients recruited in pragmatic care trials.