In the study by Rahman et al. (2007), the effect of arsenic exposure (through drinking of contaminated tube well water) on fetal loss and infant death was studied for Matlab, Bangladesh.
The target population was pregnant women. In all, 29,134 pregnancies were included in this study. The health workers visited the area regularly and collected the health and demographics data. The pregnancies were identified by missed menstrual period and after 8 weeks of gestation, these pregnant women were advised to visit health care facilities. Monthly household surveys provided information of pregnancy outcomes. Only those pregnancies were included for which the drinking water history and tube-well arsenic concentration data (during pregnancy period) was available. These criteria reduced the eligible sample size to 29,134 from a total data set of 51,500 pregnancies.
Hope this helps.