The usefulness of Stata in the analysis of complex veterinary surveys
Infectious bovine rhinotracheitis is caused by the bovine herpesvirus type 1. It is an enzootic disease on the B List of the Office International des Epizooties (O.I.E.). Programs to eradicate bovine herpesvirus type I have been implemented in several European countries to facilitate the free trade of cattle, semen and embryos within the European Community. Therefore, Belgium has an incentive to control and eradicate this viral infection. In the initial stage of the eradication campaign it is essential to survey the infection prevalence. Also, it is important to investigate the survey results for possible risk factors that might be associated with bovine herpesvirus-1 positivity among cattle. The national bovine herpesvirus 1 seroprevalence (apparent prevalence) in the Belgian cattle population was determined by a serological survey that was conducted from December 1997 to March 1998. In a random sample of unvaccinated herds (N=309), all cattle (N=11,248) were tested for the presence of antibodies to glycoprotein B of bovine herpesvirus 1. The age and sex of the animals and the type (dairy, mixed or beef) and size of the herds were registered. The survey is an example of a stratified one-stage cluster sampling design. STATA has some very useful commands for analysing surveys. The data set was analysed using the Svylogit and Gllamm STATA commands, which provided similar results. The strengths of Svylogit and Gllamm will be highlighted. We will also compare the command Gllamm with the SAS procedure NLMIXED, which produced similar results on the analysed data set. The binary response is the apparent prevalence, which is the serological discrete test result (positive/negative). The true infection status was mimicked via an ado-file, using expert opinion on the uncertainty regarding the test misclassification probabilities. The results based on the analysis using this new response were compared with those based on the analysis of the original response.
