3.1. Errors in the estimates
The data was recorded in MINITAB (version 15) and analysed by descriptive statisticalmethods. The data used to calculateW^ hb,W^ cb and W^ dt were found to be normally distributed and average values are given ±95% confidence intervals. Other estimates of error are explained where they are given.
Thb: This was estimated from official records (BBS, 2007). In each case, the number of beds in the sampled hospitals was checked during data collection, and was in line with the official values.
Tcb: The total number of beds in clinics (6136) was based on the estimate previously reported in ‘‘The Study on the Solid Waste Management in Dhaka City” (JICA, 2005) and derived from government records (BBS, 2007). There is no indication of the error in this estimate but it is likely that it may be a slight underestimate.
Tdt: This was estimated by assuming that the 44 sampled diagnostic centres were representative of the population of
551 centres in terms of the number of tests per centre, allowing Tdt to be calculated as 14,000 ± 1100. This approach is considered reasonable, as the 44 sampled centres were found to be quite homogeneous in terms of the number of tests per centre per day (25 ± 2). However, it is likely that 551 is a slight underestimate of the number of centres.