| Previous: Estimating Cause-specific Adult Mortality | Up: Examples | Next: R Function Reference |
Sometimes information about the rate of a specific cause of death is available from sources other than the verbal autopsy survey. For example, the death rate due to homicide and accidents can usually be determined by direct questioning with little error. The maternal death rate is typically lower than a certain known percentage due to sex and age group restrictions. In some other situations, a rough range of the death proportions for specific diseases may be available from other surveys on health outcomes. Researchers may also be willing sometimes to use analyses of neighboring countries or previous analyses in the same region to put bounds on some of the true proportions.
Below we give an example of how to incorporate such information into the va procedure. In this example, the cause of death ``a13'' and ``a18'' are fixed at 21.6% and 8%, and the fraction of cause ``a14'' is expected to be estimated between 0.5% and 1%.
res2 <- va(formula=cbind(S1+...+S49)~cod,
fix=c("a13=0.216", "a18=0.08"),
bound=c("0.005<a14<0.01"),
data=list(FBA,DSS), nsymp=16,
n.subset=10,prob.wt=1,printit=TRUE,boot.se=FALSE)