Skip to main content
. 2019 Nov 22;87(1):63–74. doi: 10.1002/ana.25637

Table 3.

Best Independent Early Predictors of 30‐Year EDSS >3.5 and ≤3.5. All models include EDSS 10 at or before 30 years.

Predictor Odds Ratio (95% CI) p Predictor Combinations
30‐year EDSS >3.5, best independent predictors up to 1 year a
Baseline infratentorial lesion count, ≥1 vs 0 16.8 (2.0–139.7) 0.009 0 0 ≥1 ≥1
1‐year deep white matter lesion count, ≥1 vs 0 6.7 (1.7–26.0) 0.006 0 ≥1 0 ≥1
Model‐predicted probabilities for 30‐year EDSS >3.5 (95% CI) 13% (0–26) 49% (33–64) b 94% (83–100)
Model‐predicted probabilities for 30‐year EDSS ≤3.5c 87% 51% 6%
30‐year EDSS >3.5, best independent predictors up to 5 years d
Baseline infratentorial lesion count, ≥1 vs 0 8.0 (1.5–41.4) 0.013 0 0 ≥1 ≥1
5‐year deep white matter lesion count, >5 vs ≤5 5.1 (1.7–15.6) 0.004 ≤5 >5 ≤5 >5
Model‐predicted probabilities for 30‐year EDSS >3.5 (95% CI) 18% (5–30) 52% (36–71) 63% (22–100) 90% (76–100)
Model‐predicted probabilities for 30‐year EDSS ≤3.5 82% 48% 37% 10%

All models include EDSS = 10 at or before 30 years.

a

Model n = 80. Overall model accuracy using 0.5 probability cutoff (95% CI) = 71% (60–81).

b

There were no subjects with this lesion combination.

c

These probabilities and their CIs are 100% minus the >3.5 probabilities.

d

Model n = 79. Overall model accuracy using 0.5 probability cutoff (95% CI) = 75% (64–84).

CI = confidence interval; EDSS = Expanded Disability Status Scale.