## Denominator problems

(go to Outline)

Another major problem with surveillance is the lack of denominator to calculate incidence rates. Surveillance systems provide data only on cases of disease, not on the population from which they came. The population denominator, which is often unknown in humanitarian emergencies, must come from somewhere else.

__Uncertain population denominator__

In Goma, Zaire in 1994, deaths were counted by counting the bodies which were picked up for burial by the size of the road. For the month between July 14 and August 14, 48,347 bodies were counted. To calculate the mortality rate, this number of deaths was divided by the population. But what was the population? Some early very rough estimates put the number as high as 1,000,000. Later estimates were 500,000 - 800,000. One camp had an estimated population of 350,000 until a more accurate assessment was done using aerial photography, when the estimate dropped to 180,000. Which number do you put in the denominator?

One recurrent problem in acute humanitarian emergencies is initially using a very rough estimate of total population which everyone knows is probably inaccurate for want of anything better. Then a registration, census, or more accurate assessment is done. Often, the apparent incidence rate of disease jumps suddenly because the population denominator has suddenly declined. To avoid such sudden changes in disease rates, most public health workers continue to use the old, inaccurate estimate unless there is other evidence of a sharp rise in the incidence rate. Or they will apply past disease totals to the new population estimate and recalculate past rates so that they can be accurately compared to current rates using the new population estimate.

In addition, the size of the population in emergencies often changes rapidly. Keeping track of an accurate population size is often very difficult unless there is an ongoing registration of people leaving or entering the population.