Skip to main content

Table 3 Model comparisons for mortality and hospitalization in the general population cohort, full and age-restricted

From: Comparing comorbidity measures for predicting mortality and hospitalization in three population-based cohorts

Model

Full cohort (n = 662,423)

Age 65+ (n = 137,700)

 

c-statistic

(95% CI)

Brier

score (SD)

Δc(%)

c-statistic

(95% CI)

Brier

score (SD)

Δ c(%)

Death

Base model

0.880

(0.877, 0.884)

0.012 (0.098)

--

0.729

(0.723, 0.735)

0.046

(0.183)

--

+ # diagnoses

0.901

(0.898, 0.904)

0.012 (0.096)

0.021 (2.36)

0.769

(0.764, 0.775)

0.046

(0.179)

0.040 (5.53)

+ Charlson

0.905

(0.902, 0.908)

0.012 (0.095)

0.025 (2.81)

0.785

(0.779, 0.790)

0.045

(0.177)

0.056 (7.64)

+ Elixhauser

0.913

(0.910, 0.916)

0.012 (0.093)

0.033 (3.73)

0.805

(0.799, 0.810)

0.044

(0.173)

0.076 (10.36)

+ # drugs

0.894

(0.890, 0.897)

0.012 (0.096)

0.013 (1.53)

0.764

(0.759, 0.770)

0.045

(0.178)

0.035 (4.83)

+ CDS

0.889

(0.886, 0.892)

0.012 (0.097)

0.009 (1.01)

0.751

(0.745, 0.757)

0.046

(0.181)

0.022 (3.00)

One or more hospitalizations

Base model

0.652

(0.651, 0.654)

0.138 (0.238)

--

0.563

(0.559, 0.566)

0.215

(0.171)

--

+ # diagnoses

0.722

(0.720, 0.724)

0.130 (0.233)

0.070 (10.68)

0.668

(0.664, 0.671)

0.202

(0.189)

0.105 (18.60)

+ Charlson

0.671

(0.669, 0.672)

0.136 (0.238)

0.018 (2.81)

0.613

(0.610, 0.616)

0.210

(0.179)

0.050 (8.92)

+ Elixhauser

0.682

(0.680, 0.683)

0.134 (0.236)

0.029 (4.47)

0.630

(0.627, 0.633)

0.206

(0.184)

0.067 (11.92)

+ # drugs

0.688

(0.686, 0.690)

0.134 (0.235)

0.036 (5.46)

0.625

(0.622, 0.628)

0.207

(0.181)

0.063 (11.10)

+ CDS

0.672

(0.671, 0.674)

0.136 (0.236)

0.020 (3.05)

0.604

(0.601, 0.607)

0.210

(0.177)

0.041 (7.34)

Two or more hospitalizations

Base model

0.706

(0.704, 0.709)

0.047 (0.187)

--

0.571

(0.567, 0.576)

0.110

(0.246)

--

+ # diagnoses

0.782

(0.779, 0.785)

0.045 (0.179)

0.075 (10.67)

0.686

(0.682, 0.690)

0.105

(0.235)

0.115 (20.07)

+ Charlson

0.731

(0.728, 0.734)

0.047 (0.184)

0.024 (3.42)

0.633

(0.629, 0.638)

0.108

(0.241)

0.062 (10.92)

+ Elixhauser

0.748

(0.745, 0.751)

0.046 (0.181)

0.042 (5.91)

0.653

(0.649, 0.658)

0.106

(0.237)

0.082 (14.40)

+ # drugs

0.744

(0.742, 0.747)

0.046 (0.182)

0.038 (5.37)

0.638

(0.633, 0.642)

0.107

(0.238)

0.067 (11.66)

+ CDS

0.729

(0.726, 0.732)

0.047 (0.184)

0.023 (3.18)

0.619

(0.614, 0.623)

0.108

(0.241)

0.048 (8.32)

  1. Base model includes age, age2, sex, income quintile, and geography
  2. CDS = Chronic Disease Score
  3. Δc = difference in the c-statistic between the base and full models; c-statistics in boldface font are significantly different from the c-statistic for the base model, according to the method of DeLong et al. [39]