Modeling repeated count measures with excess zeros in an epidemiological study.