Table 3.

Multivariable logistic regression analysis evaluating the association between bioavailable estradiol and hospital mortality

Dependent variable	Independent variables	Odds ratioa (95 % CI)	p Value
Hospital mortality	Bioavailable estradiol ≥ median	4.27 (1.65–11.06)	0.003
	Age in years	1.04 (1.01–1.07)	0.009
	Charlson index ≥ median	4.15 (1.53–11.27)	0.005

The final parsimonious multivariable model included age in years and Charlson comorbidity index dichotomized by the median value (3). This model had good discrimination (c-statistic = 0.73) and acceptable calibration (Hosmer-Lemeshow χ² = 13.4, p = 0.10)

^aOdds ratios refer to the change in likelihood of hospital mortality for patients with bioavailable estradiol concentration ≥211 pM, a 1-year increase in age, and a Charlson comorbidity index ≥3, all after adjustment for the other two variables