A multivariate example of case-cohort analysis: neonatal mortality in Switzerland, 1979-81.

A study of all births in Switzerland during the years 1979-81 is used to illustrate the advantages of the case-cohort design for this epidemiologic analysis of neonatal mortality. The example shows that familiar associations between infant mortality and explanatory variables such as sex and weight at birth can be precisely estimated using only a sample from the full cohort.