Differences in health estimates using telephone and door-to-door survey methods-a hypothetical exercise