In the article by Rahman et al., effect of arsenic exposure through ingestion of contaminated tube well water on infant death and fetal loss was studied. The outcome was determined in terms of early fetal loss, late fetal loss, neonatal death, postnatal death and infant death.
For this study, a large number of pregnancies were studied. Community health workers visited the area monthly and collected the record of missed menstrual periods and used these to identify pregnant women. Once pregnancy was identified (and confirmed), pregnant ladies were advised to visit nearby health facilities. Household surveillance, on monthly basis, yielded the data on outcomes of pregnancies. Using these criteria, 51,500 pregnancies were identified. Only those pregnancies were included in the study for which the history of drinking water source and the arsenic concentration in that particular source was available. Using these constraints, only 29,134 pregnancies were selected for inclusion in this study.
Hope this helps.