Selection method and variable | Odds ratio^{b} | z^{c} | P > |z|^{d} | 95% CI^{e} |
---|---|---|---|---|

Forward variable selection | ||||

EGA (wk) | 1.12 | 2.71 | 0.007 | 1.03-1.22 |

Backward variable selection | ||||

EGA (wk) | 1.17 | 2.63 | 0.008 | 1.04-1.31 |

Body mass index (kg/m^{2}) | 1.64 | 2.39 | 0.017 | 1.09-2.45 |

Novel infection at admission | 0.16 | −2.45 | 0.014 | 0.038-0.70 |

Smoking | 0.13 | −2.47 | 0.014 | 0.026-0.66 |

Predicted day 7 lumefantrine concn (ng/ml) | 0.995 | −2.36 | 0.018 | 0.99-1.0 |

↵

*a*Data are for 98 patients. Information on the type of infection at admission was missing for five patients.↵

*b*The odds ratio gives the expected relative change in odds when there is a 1-unit change in the value of the predictor variable when the values of all of the other variables in the model are held constant.↵

*c*The*z*statistic was used to test the hypothesis that the logistic regression coefficient is equal to 0.↵

*d*Two-tailed*P*value for the*z*test.↵

*e*The 95% CIs for the logistic regression coefficients.