The first reason that mortality estimates are difficult to obtain in many forced migrant situations is because there is usually no existing vital registration system. If a registration system is in place, it is likely to break down in an emergency. Yet even if complete registration data were available, they would be likely to be flawed. Deaths are often underreported because people may fear that their family food rations will be reduced if their family size decreases. Cultural taboos about death may also lead to underreporting. Because the total population, which is the denominator when calculating mortality rates, is commonly overestimated, mortality rates are much more likely to be understated than overstated (Toole and Waldman, 1997). In addition, in order to estimate age-sex-specific mortality rates, one needs to know the composition of the population, and this information, if available, is often faulty.

Estimates are thought to be most reliable in situations in which refugees are in an organized camp setting and least reliable where internally displaced persons are spread over a large area. Generalizing findings beyond a specific sample is difficult, however, because of differing survey methods and lack of data on groups that are hard to locate or count.

Brent Burkholder discussed methods for estimating mortality rates in refugee camps. Because the total number of persons is used as the denominator when calculating a crude mortality rate, estimating the size of the entire population is important. Burkholder used the example of Goma, Zaire, in the summer of 1994 to illustrate how different methods of counting the total population led to very different estimates. Médecins Sans Frontières marked grids on crude maps to show population densities of certain areas. Individuals within selected areas were enumerated and their numbers extrapolated to provide an estimate of the total number of persons in the camps, which was 750,000. The French military, on the other hand, counted the number of shelters in aerial photographs and extrapolated from small ground surveys of population density to estimate a total number of 500,000; this count was likely to be an underestimate because many people did not have shelters at that point.

The second important component of an estimated crude mortality rate is the total number of deaths over a period of time. Burkholder discussed several collection techniques for mortality data in emergency situations. Techniques range from burial site observation to collection of hospital or death records to surveys of community leaders or the population as a whole. In Goma, it was difficult to hide deaths (which refugees might have been inclined to do to avoid losing food rations) because graves could not be dug in the volcanic soil. Body collectors were employed to count the number of dead. At first, this led to overestimates of mortality because many collectors believed that they would be paid per body. As Burkholder explained, after this misunderstanding was cleared up, the estimates were thought to be fairly reliable.