RESUMO
PIP: This work describes the methodology and demonstrates the utility of usual limits of variation as an instrument of diagnosis and analysis of the status of infant mortality at a given moment. The method is illustrated using Cuban data for 1980-84 and the 1st 6 months of 1985 at the national and provincial levels. The infant mortality rate was always under 20/1000 live births during the 5 years. Monthly level provincial data on births and infant deaths were used to calculate infant mortality rates and to identify the limits of usual variation. The processing was done manually and the results were expressed as rates represented graphically, showing the monthly maximum, minimum, and median. 4 separate areas of the graphs were identified: the success zone under the minimum curve, the security zone between the minimum and average curve, the alert zone between the average and maximum curve, and the alarm zone above the maximum curve. Cuba's highest infant mortality rates have occurred in the months of June and July for the 1980-84 series. A rate above the maximum suggests a deterioration of the infant mortality situation. The same series can be calculated for early or late neonatal or postneonatal mortality, for mortality by specific causes, or in smaller geographic areas. The usefulness of the technique is diminished if the area or the period of analysis becomes too small. Established limits of usual variation should be updated at least every 2 years.^ieng