Fig. 3.

Stroke outcome in relation to ΔIGF-I. a. Distribution of crude ΔIGF-I values and crude 3-month mRS scores (N = 349). The box shows the overall Odds Ratios (OR) and 95% confidence intervals of an ordinal regression with mRS score as the dependent variable. For convenience, the line represents a crude correlation (r = − 0.114, p = 0.033). b. Unadjusted ΔIGF-I-quintile distribution (%) of stroke outcomes as indicated by mRS scores of 0–2 (good or favorable) or 3–6 (poor or unfavorable) 3 months and 2 years after IS. The p-values from the Chi-square analysis comparing distributions of good and poor outcome are shown. c. Functional outcome 3 months post-IS, shown as OR and 95% CI for associations (binary logistic regression) of favorable mRS score with unfavorable functional outcome for each of the ΔIGF-I quintiles relative to ΔIGF-I q1 (q1 is a reference with OR = 1, shown as a hatched line). Models 1–4 are shown with successively added adjustments for sex (S), age (A), traditional cardiovascular covariates (C), initial stroke severity (I), and day of the first blood sample (D), together with their respective numbers (N) with complete datasets. The boxes show the p-values for the overall associations using ΔIGF-I quintiles as a continuous variable with the same respective adjustments (p-trends). d. Functional outcomes 2 years after IS shown as OR and 95% CI for associations (binary logistic regression) between favorable mRS score and unfavorable functional outcome for each of the ΔIGF-I quintiles, as in B