In the absence of lists from which a random sample could be drawn, stage-wise area sampling was combined with systematic sampling so that every hospital user in Dhaka city had an equal chance of being selected. Firstly, 17 residential areas were randomly selected from a generated list of all major residential areas in Dhaka. The selected areas included Motijheel, Purana Paltan, Kamlapur, Lalbagh, Gopibagh, Imamganj, Dhanmandi, Mohammadpur, Lalmatia, Mirpur, Gulshan, Banani, D.O.H.S, Uttora, Mohakhali, Rampura and Malibagh. There was general consensus that these areas included different socioeconomic groups of hospital users. From each area, streets were randomly selected. Residential homes were chosen next using systematic sampling. Interviewers were also given a letter of introduction from a wellrecognized private university so residents could see that the study was authentic. Finally, a telephone number was provided for respondents with questions orĀ