Table 3 Summary of predictive variables... | American Society of Hematology

Table 3

Summary of predictive variables in the multivariate models

Early predictive variables	Late rates dichotomous at median no. of events/y				Late rates dichotomous at 1 event/y
Early predictive variables	P	OR	95% CI	R²	P	OR	95% CI	R²
For outcome: late pain
Early pain, nondactylitis	.002	1.93	1.26-2.95	0.060	—	—	—
Early ACS	—	—	—		.009	1.57	1.12-2.21	0.053
For outcome: late ACS
Early ACS	<.001	2.23	1.52-3.29	0.149	<.001	2.51	1.67-3.86	0.236
For outcome: late pain and ACS
Early ACS	.001	1.92	1.36-2.70	0.152	<.001	1.93	1.42-2.62	0.123

Early predictive variables	Late rates dichotomous at median no. of events/y				Late rates dichotomous at 1 event/y
Early predictive variables	P	OR	95% CI	R²	P	OR	95% CI	R²
For outcome: late pain
Early pain, nondactylitis	.002	1.93	1.26-2.95	0.060	—	—	—
Early ACS	—	—	—		.009	1.57	1.12-2.21	0.053
For outcome: late ACS
Early ACS	<.001	2.23	1.52-3.29	0.149	<.001	2.51	1.67-3.86	0.236
For outcome: late pain and ACS
Early ACS	.001	1.92	1.36-2.70	0.152	<.001	1.93	1.42-2.62	0.123

ORs are calculated by binary logistic regression analysis, which adjusts for sex, genotype, early nondactylitis pain, early dactylitis, and early ACS episodes as appropriate. ORs correspond to a single unit increase in the predictive variable (an episode of ACS or pain, where appropriate). Nagelkerke R² values are reported as coefficients of determination and indicate the proportion of variability in the outcomes explained by the models.

— indicates not predictive.

View Large