Table 3. Parameter estimates, model fit, and predicted accuracy of Poisson regression models for neonatal mortality rate among selected Pacific Islander neonates in the 2014–2018 US birth cohort.
| Model | Coef | Est | SE | 95% CI | P-value | AIC | Non-bootstrapped | Bootstrapped | |||
|---|---|---|---|---|---|---|---|---|---|---|---|
| ROC-AUC | ROC-AUC 95% CI | ROC-AUC | ROC-AUC 95% CI | ||||||||
| GA-based model | β 0 | 19.827 | 1.175 | 17.525 | 22.129 | <0.001 | 676.524 | 0.897 | 0.859–0.935 | 0.897 | 0.895–0.899 |
| β 1 | -0.274 | 0.200 | -0.665 | 0.117 | 0.170 | ||||||
| β 2 | -4.258 | 0.228 | -4.704 | -3.811 | <0.001 | ||||||
| BW-based modela | β 0 | 3.168 | 0.581 | 2.029 | 4.307 | <0.001 | 715.439 | 0.879 | 0.834–0.923 | 0.879 | 0.877–0.882 |
| β 1 | -0.413 | 0.334 | -1.067 | 0.241 | 0.216 | ||||||
| β 2 | -0.168 | 0.017 | -0.201 | -0.135 | <0.001 | ||||||
| BW-GA- based model | β 0 | 26.125 | 5.083 | 16.164 | 36.087 | <0.001 | 676.927 | 0.895 | 0.856–0.935 | 0.896 | 0.893–0.898 |
| β 1 | -0.295 | 0.201 | -0.689 | 0.098 | 0.141 | ||||||
| β 2 | -5.238 | 1.011 | -7.219 | -3.256 | <0.001 | ||||||
| β 3 | -0.307 | 0.167 | -0.634 | 0.020 | 0.066 | ||||||
| β 4 | 0.051 | 0.029 | -0.006 | 0.107 | 0.078 | ||||||
Coef, coefficient. Est, estimate. SE, standard error. 95% CI, 95% confidence interval. AIC, Akaike information criterion. ROC-AUC, area under the Receiver Operating Characteristic curve. GA, gestational age. BW, birth weight. DF, degree of freedom.
a Scaling parameter for labelled models was set to 1 due to overdispersion (Pearson Chi-square/DF>2).