Thinking outside the curve, part II: modeling fetal-infant mortality

Table 1 Mortality for White Singleton Infants with Heavily Smoking Mothers

Quantity	@ 1000 g	@ 2000 g	@ 3000 g	@ 4000 g
Risk in component 1: logit^-1{ $\overset{\land}{θ}$ } [point estimate] Confidence interval	110.1 (23.2, 392.2)	---	---	---
Risk in component 2: logit^-1{ $\overset{\land}{θ}$ } [point estimate] Confidence interval	460.3 (138.3, 819.2)	35.9 (16.9, 74.7)	16.2 (8.2, 31.5)	7.6 (2.3, 25.0)
Risk in component 3: logit^-1{ $\overset{\land}{θ}$ } [point estimate] Confidence interval	---	41.3 (6.1, 232.1)	4.0 (0.7, 20.8)	2.4 (0.2, 29.6)
Risk in component 4: logit^-1{ $\overset{\land}{θ}$ } [point estimate] Confidence interval	---	---	4.7 (3.0, 7.2)	2.8 (0.3, 28.3)
Odds ratio, component 1 vs. component 2: exp{ $\overset{\land}{θ}$ } [point estimate] Confidence interval	0.15 (0.01, 2.46)	---	---	---
Odds ratio, component 2 vs. component 3: exp{ $\overset{\land}{θ}$ } [point estimate] Confidence interval	---	0.87 (0.06, 12.7)	4.13 (0.41, 42.0)	3.19 (0.09, 115)
Odds ratio, component 2 vs. component 4: exp{ $\overset{\land}{θ}$ } [point estimate] Confidence interval	---	---	3.51 (1.44, 8.56)	2.74 (0.14, 53.1)
Odds ratio, component 3 vs. component 4: exp{ $\overset{\land}{θ}$ } [point estimate] Confidence interval	---	---	0.85 (0.19, 3.79)	0.86 (0.04, 21.1)

Mortality risks and odds ratios are estimated at selected birthweights, based on 25 samples of size 50,000 from the population of white singletons born to heavily smoking mothers. Confidence intervals are constructed using Equations (6) and (7) with C ₀ = 4.0 and φ = .2465.

ISSN: 1471-2393