front |1 |2 |3 |4 |5 |6 |7 |8 |9 |10 |11 |12 |13 |14 |15 |16 |17 |18 |19 |20 |21 |22 |23 |review |
The log odds of neonatal mortality and the
impact of twin pregnancy was formulated as quadratic functions of gestational age (GA). In
the logistic regression model, Twin was dummy coded (singleton=0, twin=1). Other
covariates were also included in the regression equation. For a more in-depth discussion of varying-coefficient models, see Hastie T, Tibshirani R. Varying-coefficient models. J Royal Stat Soc B 1993;55:757-96. For the estimation of confidence intervals in the present context, see Figueiras A et al. Regression models: Calculating the confidence interval of effects in the presence of interactions. Stat Med 1998;17:2099-105. |