The English statistician, William Farr (1807–1883) once remarked, "Death is a fact. All else is inference." He meant that everyone dies, but though useful deductions can be made about life and death, the information is imperfect and there are traps in interpretation for the unwary. Nevertheless, facts about deaths are a reliable and consistent source of information about the health of various populations. The information can be arranged in a hierarchy of reliability, ranging from certainty (the fact of death), through near certainty (age and sex), to the place and circumstances of death. The exact cause of death is probably the least reliable piece of information for most deaths. Even among persons dying in a hospital after being diagnosed with a specific disease, autopsies sometimes shows the ante-mortem diagnosis to be wrong. When people die at home after being attended by a family doctor, the cause of death stated on the death certificate is correct only an estimated 50 to 60 percent of the time. The accuracy of death certifications is highest when death is due to a violent cause such as a traffic crash or gunshot wound, and lowest when death is sudden and there is no autopsy. The certified cause of death is often whatever happens to be most fashionable—heart attack or stroke being most in vogue in the late twentieth and early twenty-first centuries.
The aggregated data in death certificates are used to calculate mortality rates that can be manipulated in various ways to show general and specific health indicators and trends, and to make predictions about the likely future course of events. The commonly used varieties of mortality rates are: crude; age-standardized; cause-specific; infant
Crude Mortality Rates. These rates reflect the number of deaths in a defined population during a specified period—usually a year—divided by the midyear population. Because of variations in age composition and other factors, crude rates are seldom useful for comparisons.
Age-Standardized Mortality Rates. Irregularities in crude rates can be reduced by making adjustments. The simplest and best way to do this is usually to calculate what the rate would be if the population concerned had the age composition of a standard population—one in which the composition is precisely known, such as that of a census year.
Cause-Specific Mortality Rates. The greatest value of mortality rates for studies of common conditions like cancer and coronary heart disease is in comparisons of cause-specific mortality rates in different populations (different regions, occupations, time periods). Such comparisons have illuminated understanding of many causal and associated factors and have prompted much detailed study of these diseases.
Infant and Perinatal Mortality Rates. The indicator of health most commonly used for comparing nations and trends over time is the infant mortality rate. This is the number of deaths in one year of infants under one year of age, divided by the number of liveborn infants. It is strongly correlated with social and economic conditions. The perinatal mortality rate is a more sensitive indicator of the standards of care for women before, during, and immediately after childbirth. The perinatal mortality rate is the number of fetal deaths between twenty-eight weeks gestation and one-week post-partum divided by the number of live births in a year multiplied by one thousand. It is considered to be a good indicator of the quality of care received by pregnant women.
Maternal Mortality Rates. This is a measure of the risk of dying from puerperal causes—causes associated with pregnancy, childbirth and the postpartum period. The World Health Organization defines this as any time up to forty-two days after termination of pregnancy, irrespective of the duration of pregnancy or its outcome in a live birth, stillbirth, miscarriage, or abortion. Maternal mortality rates are very low in the industrial nations, reflecting high standards of care during pregnancy and childbirth. In countries where women have no other recourse than induced abortion to control their fertility, almost a million women die annually of puerperal causes—a terrible loss of life that could be prevented by easier access to family planning. Cultural, political, and religious opposition inhibits these societies from addressing this appalling problem.
Standardized Mortality Ratio (SMR). This is the ratio of the number of deaths observed in a specified population to the number that would be expected if that population had the same mortality rate as a standard population. The standard population is arbitrarily chosen; it may be a recent census year or an artificial, computer-generated one. The SMR is a very useful statistic, often used to compare outcomes in two or more groups under study.
Throughout the twentieth century, death rates fell steadily everywhere in the industrial nations, reflecting improved living conditions, greater longevity, and improved control over causes of premature death. A sad exception was the experience of the nations of Eastern Europe and the former Soviet Union after the collapse of communism. In these nations, social chaos and the decay of what was once a relatively efficient pubic-health system led to lethal epidemics of diphtheria and other communicable diseases that were previously controlled by immunization programs. This experience demonstrated the necessity of maintaining effective and efficient public health services.
JOHN M. LAST