CUADRO 1: RESUMEN DE LAS PRINCIPALES CARACTERÍSTICAS CLIMÁTICAS DEL PAÍS
CUENCAS HIDROGRÁFICAS
In stable populations, it is better to col- lect information of death prospectively where, as each death occurs, it is report- ed to public health or government authorities. Such systems allow the cal- culation of recent death rates as fre- quently as is required, and this data can be collected easily. In vital statistics systems, death reporting is often mandatory for persons charged with burying bodies. If deaths occur pre- dominantly in clinics or hospitals, med- ical personnel may be responsible for death reporting. In many societies, reli-
gious leaders record deaths. In humani- tarian emergencies, someone may be hired to count deaths and monitor the area designated as the graveyard or cre- mation site.
All of these systems of counting deaths require separate information on the size of the population from which the deaths occurred in order to calculate mortality rates. Such information for the popula- tion denominator may come from census counts; census projections; population registration; or the monitoring of births, deaths, immigration and emigration.. In humanitarian emergencies and other sit- uations in which data on population size is poor, techniques have been developed to estimate population size.
But which population size is used as the denominator if deaths are counted for a time period during which the popula- tion size fluctuates? One approach would be to calculate the average popu- lation during the time period (add the population at the beginning of the time period to the population at the end of the time period and divide by 2, produc- ing the arithmetic mean population).
Example 2.1 Expression of mortality rates
Even though they may appear to be very different, mortality rates expressed using different population constants or time periods indicate the same rate of death in a population. For example:
• Take a mortality rate of 9.6/1,000 population/year. • This could be expressed as 0.8/1,000 population/month. • This could also be expressed as 0.26/10,000 population/day.
The equivalency of these rates can be seen if we calculate from one rate to the other. For example, if the rate of 0.26 deaths per 10,000 per day remained constant for an enti- re year, we would expect that about 95 people would die (0.26 x 365 days in a year) out of each 10,000 people in the population.
A second approach would be to deter- mine the population at the mid-point of the time period and use that (this is called the “mid-interval population” and is the method used most commonly in vital statistics systems).
However, in many situations, the popu- lation denominator for a mortality rate itself is only a rough population esti- mate and no accounting can be made for changes in the population during the time period of interest. This is often true in humanitarian emergencies which have the additional complication of high rates of in-migration and out-migration.
TYPES OF MORTALITY RATES
Many different rates are used to measure mortality:
• Crude mortality rate/crude death rate (CMR)
• Age-specific mortality rate (ASMR) • Age-specific mortality rate for children
under 5 years (ASMR-U5) • Under-5 mortality rate (U5MR) • Cause-specific mortality rate • Infant mortality rate (IMR) • Maternal mortality rate (MMR) Crude Mortality Rate (CMR)
The crude mortality rate (CMR), also called the crude death rate or CDR, is
defined as the number of people of all ages and both sexes who die in a given time interval divided by the total pop- ulation at the mid-point of that time interval. The CMR always includes the length of the time interval and a standard population size, called the population constant. For example, a CMR may be 8.5 deaths per 1,000 per- sons per year.
It is calculated by the following formula where:
• the numerator of the fraction in parentheses is the number of deaths which occurred in a specific popula- tion during a certain time period. Only deaths which occurred during this time period should be included in the numerator of the mortality rate. • the denominator of the fraction is the
number of people in the population in which these deaths occurred. This population should be well defined, and only persons fitting this definition should be included in the denomina- tor. For example, if you are calculat- ing the mortality rate for a certain province, only people who lived in that province during the time period of interest should be included in the population denominator.
DEFINING AND MEASURING MORTALITY CHAPTER
2
This means that each person has an average likelihood of 0.006393 (or 0.6393% chance) of dying during the 8-month period.
Stating that the CMR is 377 deaths per 58,975 per 8 months has little meaning; it can- not be directly and meaningfully compared to other CMRs from previous periods or from other populations. However, if this rate is converted to deaths per 1,000 per year, as follows, it can be compared to other rates expressed in the same way:
1. To convert this likelihood to a likelihood for a standard population size, we multiply this rate by 1,000 to obtain 6.4. This means that during the period of 8 months, 6.4 peo- ple died for each 1,000 people in the population.
2. To convert this likelihood to a likelihood for 1 year: 6.4 is divided by 8/12ths (8 months divided by the 12 months in 1 year) = 9.6 deaths per 1,000 population per year. This rate means that, if the death rate for the 8-month period continued for an entire year, for every 1,000 people in the population, there would be 9.6 deaths.
Using the formula above, the rate of deaths per 1,000 per year would be calculated as follows:
Example 2.2 Calculation of CMR with standard time unit
In a specific population, 377 deaths occurred during a period of 8 months in a popula- tion with a mid-interval size of 58,975.
Using the information in the formula you get: 377
58,975 = 0.006393 or 0.6393%
Vital statistics systems which calculate mor- tality on an annual basis use the size of the population on July 1 to indicate the average population between January 1 and December 31. Such systems count the number of deaths during a year's time, divide it by the mid- interval population, then multiply by 1,000 (the population constant) to get the number of deaths per 1,000 population per year. Since the CMR reflects the overall risk of death in the population among all ages and both sexes, it is the least specific indicator of mortality. Mortality reflected in the CMR may result from causes as varied as those from violent deaths from massacres to those from neonatal tetanus. If only one indicator
of mortality can be calculated, CMR is usu- ally the one chosen. Ideally, a newly calcu- lated CMR should be compared with a pre- vious CMR from the same population to determine whether the mortality rate is ris- ing or falling. Such trend information can be used as an overall evaluation of health, nutrition and other interventions. However, when prior mortality data are unavailable, a rough rule-of-thumb can be used:
• a CMR of less than 1 death per 10,000/day indicates a reasonable health situation;
• a CMR of more than 1 death per 10,000/day reflects elevated mortality; and • a CMR of more than 2 deaths per
Calculating CMR using person-time units
The denominator of a mortality rate is the number of people in the population; howev- er, the denominator can also be seen as being based on person-time instead of the number of persons. That is, a rate of 10 deaths per 10,000 per day may be seen as the risk of death in 10,000 people during a period of one day, or the risk in 5,000 peo- ple during 2 days, or the risk in 1,000 peo- ple during 10 days. In all three examples, the denominator is 10,000 person-days (the number of people in the population multi- plied by the number of days in the time peri- od). This will be important later when dis- cussing the measurement of mortality in cross-sectional surveys. The sample size required to achieve a certain precision around the estimate of the CMR is the num- ber of person-time units required in the denominator of the rate.
In general, if births and deaths are distribu- ted evenly throughout the time period, then each person who was born or who died during the time period contributes, on ave- rage, _ a person-time period to the denomi- nator. For example, vital statistics systems
assume that births and deaths occur evenly throughout the year. Therefore, each person who was born or who died during the year contributed about _ a year to the denomina- tor. Use of the mid-interval population, as described above, captures half of deaths and half of births and is one way of adjusting the population denominator for the incomplete contribution of births and deaths. Age-Specific Mortality Rates (ASMR) Age-specific mortality rates (ASMR) restrict both the numerator and denominator to per- sons of a certain age. For example, a morta- lity rate for persons 15-49 years of age is the number of deaths of persons 15-49 years of age divided by the mid-interval population of persons 15-49 years of age (adjusted for the length of the time period).
ASMR is often used to determine if the rate of death is substantially different from that expected in any specific age group. For example, in a survey in Kosovo, the death rate among young men was much higher than expected, indicating that some factor disproportionately increased the risk of death in this age group.
Figure 2.1 Age-specific mortality rates, Badghis Province, March 2001 - April 2002.
3.5 3 2.5 2 1.5 1 0.5 0 <5 5 - 14 15 - 49 50+ Mor
tality rate (deaths/10,000/day)
DEFINING AND MEASURING MORTALITY CHAPTER