Association between clinical and environmental variables with the occurrence of hospitalization and ACU in patients with SCD from multivariate analyses
Aged up to 3 y . | Aged up to 6 y . | ||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|
Model description | Comparison model | LRT P | RN2 | AIC | AUC | Model description | Comparison model | LRT P | RN2 | AIC | AUC |
ACU = ED visits + hospitalizations | ACU = ED visits + hospitalizations | ||||||||||
Base model | NA | NA | 0.10 | 547 | 0.70 | Base model | NA | NA | 0.11 | 350 | 0.70 |
Base model + E∗ | vs base model | 8 × 10–4 | 0.12 | 538 | 0.72 | Base model + E† | vs base model | .004 | 0.14 | 343 | 0.74 |
Hospitalizations | Hospitalizations | ||||||||||
Base model | NA | NA | 0.09 | 526 | 0.68 | Base model | NA | NA | 0.13 | 373 | 0.72 |
Base model + E∗ | vs base model | 2 × 10–4 | 0.12 | 514 | 0.72 | Base model + E‡ | vs base model | 4 × 10–4 | 0.17 | 357 | 0.76 |
Aged up to 3 y . | Aged up to 6 y . | ||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|
Model description | Comparison model | LRT P | RN2 | AIC | AUC | Model description | Comparison model | LRT P | RN2 | AIC | AUC |
ACU = ED visits + hospitalizations | ACU = ED visits + hospitalizations | ||||||||||
Base model | NA | NA | 0.10 | 547 | 0.70 | Base model | NA | NA | 0.11 | 350 | 0.70 |
Base model + E∗ | vs base model | 8 × 10–4 | 0.12 | 538 | 0.72 | Base model + E† | vs base model | .004 | 0.14 | 343 | 0.74 |
Hospitalizations | Hospitalizations | ||||||||||
Base model | NA | NA | 0.09 | 526 | 0.68 | Base model | NA | NA | 0.13 | 373 | 0.72 |
Base model + E∗ | vs base model | 2 × 10–4 | 0.12 | 514 | 0.72 | Base model + E‡ | vs base model | 4 × 10–4 | 0.17 | 357 | 0.76 |
AIC, Akaike information criterion value; AUC, areas under the receiver operator characteristics curves; base model, hydroxyurea treatment + chronic transfusion + genotype; E, environmental factors; LRT P, P value calculated using likelihood ratio test; NA, not applicable; RN2, pseudo-R2 (RN2) values calculated based on the logistic regression model.
E = % of households with children living >1 mile from a supermarket.
E = % of households with children living >1 mile from a supermarket + % of individuals with a bachelor’s degree.
E = % of households with children living >1 mile from a supermarket + % of individuals with a bachelor’s degree.