= Solution
Because the second model is saturated, its fitted worsening probabilities are the observed proportions:
$$
\widehat p_C=\frac{101}{5728+101}=0.01733,
\qquad
\widehat p_L=\frac3{1364+3}=0.002195,
\qquad
\widehat p_S=\frac{27}{4413+27}=0.006081.
$$
Thus the <drug efficacy as a risk reduction> is
$$
\widehat E_L
=1-\frac{\widehat p_L}{\widehat p_C}
=1-0.1266
=\boxed{87.3\%},
$$
and
$$
\widehat E_S
=1-\frac{\widehat p_S}{\widehat p_C}
=1-0.3509
=\boxed{64.9\%}.
$$
These are reasonably similar to the published estimates of $90.0\%$ and $62.1\%$.
The interaction coefficients provide the same quick comparison on the odds scale:
$$
e^{-2.08156}=0.125
\quad\hbox{and}\quad
e^{-1.05847}=0.347.
$$
They are treatment-versus-control <odds ratios>[odds ratio]. Since worsening is rare in every group, each odds ratio is close to its <risk ratio>, giving the approximate efficacies $87.5\%$ and $65.3\%$.
Back to article page
