Table 2 Result of the ordered logistic regression models used to estimate the proportional odds ratios for modified Rankin score (dependent variable) at discharge or 7–10 days after the stroke
VariableOdds ratio (95% CI*)p Value*
Intention-to-treat analysis
    Post-programme vs pre-programme† (univariate) (n = 1570)0.74 (0.56 to 0.98)0.038
    Post-programme vs pre-programme (n = 1256)0.73 (0.57 to 0.94)0.013
        Arm deficit‡2.40 (1.78 to 3.23)<0.001
        Impaired speech‡1.94 (1.55 to 2.42)<0.001
        Walk on admission§0.39 (0.28 to 0.55)<0.001
        Incontinent <72 h of stroke‡6.08 (4.62 to 7.99)<0.001
        Age at stroke†1.03 (1.03 to 1.04)<0.001
        Independent prior to stroke ‡ (mRS 0–1)0.56 (0.39 to 0.81)0.002
Analysis for patients treated only in an SCU compared with pre-programme
    SCU-only post-programme vs pre-programme† (univariate) (n = 1407)0.68 (0.49 to 0.94)0.020
    SCU-only post-programme vs pre-programme (n = 1126)0.69 (0.53 to 0.91)0.007
        Arm deficit‡2.33 (1.72 to 3.15)<0.001
        Impaired speech‡1.99 (1.58 to 2.51)<0.001
        Walk on admission§0.39 (0.28 to 0.54)<0.001
        Incontinent <72 h of stroke‡5.95 (4.85 to 7.29)<0.001
        Age at stroke†1.03 (1.027 to 1.039)<0.001
        Independent prior to stroke‡ (mRS 0–1)0.59 (0.40 to 0.85)0.005

  • *Adjusted for clustering of cases within hospitals; †<1% cases missing in either group; ‡<5% cases missing in either group; §<8% cases missing in either group.

  • SCU, stroke care unit.