Text this: Problem drug use prevalence estimation revisited: heterogeneity in capture-recapture and the role of external evidence.