Subscribe
 
|
—from Epidemiological Bulletin, Vol. 22 No. 2, June 2001— Inequalities in Infant Mortality in the American Region: Basic Elements for AnalysisHealth and living conditions in the Region of the Americas have considerably improved in recent decades. Some infectious diseases have been eradicated or controlled, populations have increased development possibilities and various public and health services are now available. During this time, life expectancy has increased and infant mortality has continuously declined. The reduction in infant mortality can be seen from different perspectives, which include looking at different levels of aggregation of the data and studying its distribution characteristics in different geographical areas. Based on the methodology and the data used for the health situation analysis presented in PAHO’s Annual Reports of the Director from 1996 to 2000 and the 1998 Edition of Health in the Americas, this article presents a broader approach of infant mortality analysis, with the purpose of detecting health inequalities. Measurement of the infant mortality rate (IMR) includes all deaths in children under 1 year occurring in the live birth population within a period of one year. Nevertheless, errors in the estimates can appear due to underreporting of infant deaths or live births (lb). Even with these limitations, infant mortality and its desagregation by major groups of causes continues to be one of the most commonly used health indicators for health situation analysis. For strategic and political reasons, infant mortality is under close scrutiny in many countries. It is not only used as an indicator of the health status of the population but is also important due to its impact on the populations’ life expectancy at birth. Since 1995, following PAHO’s Regional Initiative of Core Health Data, managed by the Special Program for Health Analysis, about 20 countries have been publishing subnational core data and indicators in a systematic way, including infant mortality. This allows us to present the behavior of infant mortality at the subnational level. Some countries went even further and produced information on the causes of mortality, the population groups most affected and the areas where highest infant mortality occurred. Central tendency and variability indicators of infant mortality from 18 reporting countries is presented in Table 1. The means of IMR range from a minimum of 6.43 per 1000 lb to a maximum of 87.3. The medians of infant mortality in some countries tend to be different at the country level, probably due to the great dispersion (variance) of the values observed. Medians are usually better central summary measures than averages because they limit the effect of extreme values (such as those seen in table 1). Rate ratios give an idea of the existing differences between geographic units in a country. The higher this ratio, the greater the health inequalities. On the other hand, the coefficient of variation is a relative dispersion measure that takes into account both the average value of the IMR and measures of its dispersion. It is calculated by dividing the standard deviation by the mean of the IMR in each country. Although not included in this analysis, the use of weighted averages can also be useful to take into account the relative weight of each geographic unit. The regional and subregional breakdown give a general idea of the infant mortality indicator in large areas of the hemisphere. Looking at the distribution pattern of the IMR allows one to point out the possible inequalities between different geopolitical units. A representation of the distribution of those rates is obtained by choosing an adequate subnational disaggregation (districts, departments etc.) and ordering the units in decreasing order of the IMRs. This gives a picture of the great heterogeneity of the distribution. Among the 18 countries of the Region of the Americas (1) that reported data on infant mortality at the subnational level, a total of 386 geographical units are available. Figure 1 shows the great variability of infant mortality in those countries. The amplitude of the inequalities also becomes evident when one observes that the maximum value (133 deaths by 1000 lb) represents 36 times the minimum value (3.71 deaths by 1000 lb). The distribution of the mortality at this level shows a great variation with approximately 5% of the geographical units showing values higher than 60 deaths per 1000 lb. In the other extreme, 20% have mortality lower than 10 per 1000 lb and more then half have values close to the regional average of 24 deaths by 1000 lb. As indicated previously, infant mortality has been declining since the 1960s, when it reached a median of about 80 per 1,000 lb in Latin America. At the end of the 1990s, the median IMR for the Region dropped to 20 per 1,000 lb (an average of 24.8 per 1000 lb). Looking at subregions of the Americas (Figure 2), the greatest IMR in the 1980-1985 period is found in the Central American Isthmus with 65 per 1,000 lb. In the same period, North America had the lowest rates (less than 15 per 1,000). By 1995-2000, North America had reduced its rate to nearly 8 per 1,000 lb and Central America had lowered its IMR to nearly half (35 per 1,000 lb) of the value observed in the previous period. Though the IMR in the Latin Caribbean also declined during this period, this region was left with the highest figure (nearly 45 per 1,000). The reduction of the IMR between 1980-1985 and 1995-2000 in the Andean Region, Brazil, the Central American Isthmus, and the Latin Caribbean fluctuated between 30 and 45%, while the reduction in the Non-Latin Caribbean, the Southern Cone and North America was between 20 and 25%. However, it is important to take into account that the latter three subregions started at a much lower IMR. It is easier to have an impact on higher IMRs and as a result the observed changes are difficult to compare. The distribution of IMRs within countries can be visualized in an even more precise way using box plots. When representing several countries in a single graph, one can observe that the median of the IMR in the subnational sphere shows variations among countries. In figure 3, considerable variations are observed in the median of the IMR in the same group of countries of the Americas as mentioned previously. This median goes from a minimum of 5.7 deaths per 1000 lb in Canada to 83 in Bolivia. This means that the probability of dying in children under 1 year in the country with the worst situation is 15 times greater than in the country with the best situation. Even so, 14 of 18 countries reached the proposed goal of Health for All in the Year 2000 to reduce the levels of infant mortality below 30 per 1000 lb. However, the averages presented here tend to give an image of the situation that does not account for the important differences that are occurring inside the countries. Indeed, the box plots show that there exists an important variation in the IMR within the countries and that these differences vary by country. An additional means of quantifying these differences is to present Z values (2) of the IMR of different subnational units with respect to the national average. The greater the value of the Z score (on the positive or negative side), the further the IMR with respect to the national average. This type of analysis reflects the existing inequalities assuming that the expected value is that of the national level. For example in the United States, infant mortality is greater in the District of Columbia (figure 4). A more precise analysis of the rates in different populations shows that minorities, particularly the black minority, present higher infant mortality in every state in this country (Figure 5). Similarly, the calculation and the graphic representation of Z values in departments of Guatemala (figure 6) identifies patterns with large variations in the IMR in the different geographical areas of these countries. This information can also be presented in maps (figure 7). Finally, in the study of the inequalities that exist in the IMR, it is important to take into account their relation to the different biological, socioeconomic and cultural determinants of society. In particular, the demographic and socioeconomic features of a population are factors that determine their living conditions. In this case, it is of interest to explore regression analyses to observe the relation between the IMR and other indicators, for example the population with access to water. Thus, in subnational units of Peru (figure 8) an inverse relationship between the IMR and the proportion of the population with access to drinking water is observed, with correlations of -0.65 and -0.66, respectively. The negative correlation indicates that the IMR decreases as the access to drinking water increases. The methods presented so far give a general idea of the situation, however a deeper inquiry can include other levels of analysis. For example, the analysis of the structure by cause of infant mortality in a country and at different periods makes it possible to show occasional changes in the mortality profile. Figure 9 shows specific mortality rates for several causes in children under one year of age in countries corresponding to different levels of Gross National Product (GNP), a measure of the wealth of a country (3). One can observe that mortality declined in all the presented countries. The conditions originated in the perinatal period represent a major proportion of mortality in all the countries. However in some countries such as Nicaragua, communicable diseases dominate the mortality profile. Some countries such as Canada and Chile present a substantial decrease in both groups of causes while the reduction was greater in communicable diseases in Nicaragua. External causes usually do not weigh as much in infant mortality, but in some countries such as Chile they represent a proportion that should not be disregarded. Situation analyses makes it possible to determine priority problems, and therefore represents the first step towards the establishment of priorities in health intervention. Detailed analyses of important indicators such as the IMR allows one to define priorities in specific areas, taking into account inequalities that exist within and among countries. Such analyses also allow for consideration of special population groups and the performance and quality of the services directed to them as well as general and cause-specific mortality rates. Note: Figure 9. The scales are different. (Estimated rates per 100,000 children under 1 year old.) Notes: References: Source: Prepared by Dr. Carlos Castillo-Salgado, Dr. Enrique Loyola, and Ms. Anne Roca of PAHO’s Special Program for Health Analysis (SHA). Return to Index |
|
