"We can observe the values for `Age:Smoker[T.Yes]`, the coefficient and standard error are very close, the confidence interval span over positive and negative range and the p.value is above 0.05. Considering all of the above, the provided data and the test we performed are **not sufficient information to say there exist a toxic effect with smoking for women**. **Considering current scientific knowledges, I doubt those results**, but I have no hypothesis on the reason for such difference in between the results and actual scientific data. It might be interesting to look again in the data and original article to find any explaination. "