Estimating maternal mortality: the sisterhood method.

This paper describes a new indirect technique for deriving population-based estimates of maternal mortality. The technique, called the "sisterhood method," is relevant to developing countries where the alternative data sources and approaches to estimation are often inadequate and inappropriate. The sisterhood method uses the proportions of adult sisters dying during pregnancy, childbirth, or the puerperium reported by adults during a census or survey, to derive a variety of indicators of maternal mortality. The first field trial of the method was carried out in the North Bank Division of The Gambia, West Africa, in 1987. The results indicate a lifetime risk of maternal mortality of 0.0584, or 1 in 17, approximating a maternal mortality ratio of 1,005 per 100,000 live births, which is consistent with previous estimates for this region.

Levels, trends and patterns of adult mortality in China over twenty five years.

PIP: This paper investigates the reliability of the adult mortality estimates arrived at previously for China through the use of further data and methods. The most straightforward and direct approach is through the examination of survivorship from the 1964 to 1982 censuses, which has already been done effectively. The long time span limits the extent to which the trends can be verified as opposed to the average level of mortality in the interval. The technique applied in this paper is the original version of the growth-balance method but with adjustments developed particularly for China. The growth-balance method, with adjustments when necessary, has been applied to various age specific mortality schedules, calculated from registered deaths in China. No separation by sex is made. The growth-balance was also applied to the age-specific mortality schedule and risk exposure which were the basis of the 1978 life table, which shows distinct features. In view of the tentative nature of the model and the lack of precision of the adjustment factors, the agreement between the pairs of life tables is remarkable close. The model mortality is higher than the intercensal measures at 15 to 45 years but lower at the ages above 45. There is sufficient evidence to construct a China standard life table, which could be done in several ways, such as by reference to the UN model patterns or by specifying fixed values in the 4 parameter generalizations of the logit system.^ieng

[Field performance of the sisterhood method for measuring maternal mortality]. / Evaluacion del trabajo de campo del metodo de la sobrevivencia de hermanas para medir la mortalitdad materna.

PIP: Reducing maternal mortality has now been declared a health priority in many developing countries. Translating rhetoric into action requires inputs of scarce resources. Information plays an essential role in the planning, management and evaluation of action. In many developing countries, reliable and timely information is also a scarce resource and the short term prospects for significant improvement of the conventional sources of data on maternal mortality are not encouraging. Alternative and complementary methods for generating information are required which recognize the needs and constraints of developing countries. Indirect techniques for demographic estimation have made a considerable contribution to our knowledge of child and adult mortality in these circumstances. This paper describes a new indirect technique, the sisterhood method, for estimating maternal mortality and the results of the first 2 field trials in Gambia and Peru. (Author's).^ieng

Population trends in less developed countries.

PIP: Current demographic trends in developing countries are reviewed. The author notes the growing awareness of the complexity and diversity of the population situation in different parts of the world, as both data and methods of estimation based on inadequate data improve. The consequences of these different trends are assessed.^ieng

The Population Commission and IUSSP.

PIP: The United Nations (UN) and the International Union for the Scientific Study of Population (IUSSP) have cooperated since the 1940s. In 1927 an International Population Conference in Geveva established a permanent Population Union to cooperate with the population activities of the League of Nations. The 2 institutions' successors, IUSSP and the United Nations (UN), developed close and productive linkages, collaborating to create a Multilingual Demographic Dictionary, published in English, French, Russian, and Spanish and in many other languages. Meanwhile the Union, at the request of UNESCO, prepared a pioneering study attempting to define the cultural factors affecting developing country fertility in the context of the demographic transition, In 1966 the Union and the UN collaborated to develop criteria for internationally comparable studies in fertility and family planning (FP). The resulting monograph served as a reference for many fertility studies, including the World Fertility Survey. Another study on the impact of FP programs on fertility, resulted in the organization of expert meetings and the production of a manual and monographs on FP program evaluation. There was futher cooperation in a study on mortality, internal migration and international migration, resulting in manuals on methods of analysing internal migration and indirect measures of emigration, among other things. The 1954 Wold Population Conference (WPC) and the 1965 UN WPC were organized by the UN collaborating with the Union, and the Union administered the funds used to bring developing country delegates to the Conference. Subsequent WPCs at Bucharest and Mexico City were political in nature, bu the Union contributed to both a report outlining demographic research needs. The Union also assisted the UN in organizing a series of regional population conferences, and its Committee on Demographic Instruction prepared a report for UNESCO on teaching demography, and cooperated with the Secretariat in funding the UN Regional Demographic Training Centers at Bombay and Santiago.^ieng

Childhood mortality estimated from reports on previous births given by mothers at the time of a maternity: II. Adapted multiplying factor technique.

PIP: A second method is presented for estimating childhood mortality from information on previous births and their survivorship collected from women when registering a current birth. This method, called the adapted multiplying factor technique, requires data on age of mother, total number of live births prior to the current one, and number of previous births still alive and number dead. The method is tested using 1968-1975 birth notification data and 1970 and 1976 census data from the Solomon Islands, and the results are compared with estimates derived from the preceding births technique and other methods of estimation.^ieng

Childhood mortality estimated from reports on previous births given by mothers at the time of a maternity: preceding-births technique.

PIP: This paper demostrates that, by using the Preceding-Births technique (Brass 1969), data can be collected to determine indices of the most recent childhood mortality using a totally different approach independent of the age of the mother reporting on the survivorship of previous children. This method is particularly simple since it requires only 1 question on the registration or notification form: is your last child still alive? No complex analysis and no dependence on other data sources or model life tables are necessary. An advantage of this method is the low % of selection bias. These techniques are concerned with the limited use of only that part of the maternity history data concerning mortality of recent births. Their use is recommended particularly for monitoring trends and evaluating intervention. In the interim before full registration exists in less developed countries, this is a valuable and viable means of estimating childhood mortality. Model tables describe distributions of intervals between live births reported from Colombia, Italy, USA, England and Wales; mortality schedules taken from African standards; and an estimation of proportion dead by the Preceding-Births technique in the Solomon Islands.^ieng

Censored cohort parity progression ratios from birth histories.

PIP: Despite recent progress in birth data collection and analysis techniques, it remains difficult to establish convincingly that a quantum fall in fertility due to control practices within unions is beginning to be appreciable in a population. A precise measurement of the size of fall associated with such alterations in behavior is difficult. 1 reason for the difficulty of defining measures of significance is the lack of accurate and detailed data. Another reason is the concommitant changes in other demographic characteristics that often occur at the same time as the adoption of control practices. Some of the most common are increases in age at marriage, reduction in length of postpartum amenorrhea, and shorter delays between marriage or cohabitation. Despite evidence of fertility falls over some 15 years, it has been hard to establish the size and nature of this change. The technique for doing so is the life table analysis of birth intervals. The strength of the life table approach is that it adjusts for populations at risk, such as the number of women in an age group of incomplete fertility who have experienced differing numbers of births at different intervals before the survey. The life table analysis deals with the problem of censoring. However, the estimates are seriously biased by selection for speed of reproduction. For some populations the control methods do not seem satisfactory. The most straightforward and effective approach is to subdivide the women by age at earlier birth for any particular interval by order, and to calculate separate life tables for the subdivisions. Individual countries are discussed to illustrate particular features of the analysis (Sri Lanka, Nepal, Korea, Pakistan, Latin America). The approach has been very successful in delineating trends in the parity progression ratios of higher order births. More research is needed on cutoff points of cohort and birth order for the indices, and the best ways of combining the measures to show trends.^ieng

Policies for the reduction of mortality differentials.

PIP: Effective policies for the reduction of mortality differentials can only be formulated from a knowledge of what these differentials are and some understanding of what determines them. This review draws attention to the present limitations of the information. Before turning to a discussion of policies to reduce mortality differentials, attention is directed to differentials by socioeconomic characteristics in developed countries and to mortality differentials in adult and child mortality in developing countries. Britain has the longest series of differential mortality according to individual characteristics. The classification used was occupation, with later grouping into "social classes." Infant mortality is given in table form by the social class of the father, and male adult mortality is presented in a table for the 1921-1971 period. Differentials were consistently larger for the acute and "environmental" diseases than for congenital anomalies and conditions arising from pregnancy and birth. The standardized indexes of adult male mortality showed a smaller range of variation. A good case can be made for the argument that poor health resulted in changes in occupation and hence a downward move in social class. Studies of characteristics other than occupation in the developed countries are uncommon, but a survey in the United States linked birth and death registration records with a family questionnaire from 1964-1966. In families with a household income of under $3000, the infant mortality was 60% higher than in families with a household income over $10,000. In the developed countries, adult female mortality is lower than adult male mortality at all ages. The cumulative evidence supports the old suggestion that in some developing countries female mortality is, in contast, higher than that of males. The abundance of estimates of childhood differentials in mortality in developing countries makes it necessary to be selective. A particularly systematic comparative study of child mortality was made in Latin America by Behm who examined its implications for the socioeconomic determinants of fertility. The ratio of mortality up to age 2 for children born to women with no education was 3-4 times that for children born with 10 or more years of education. There were differentials by income, occupational class, and urban-rural residence, but these were smaller than that by education and did not explain it. A longitudinal survey in England and Wales revealed significant mortality differentials by family structure. A little evidence exists that this may be even more important in developing countries. Policies to reduce morality differentials must change the environment or operate on the capacities of the families to control it. The evidence is that broad measures to improve the environment or social and health activities will reduce mortality but not socioeconomic differentials so long as the spread of control capacities over families remains the same. Available evidence supports the view that even a modest effort in family welfare policies for the more disadvantaged among the population could result in a good return in reduced mortality differentials.^ieng

Measurement of death registration completeness using the growth balance procedure applied to data from India.

"In a previous note...a new method was presented for the simultaneous evaluation of the completeness of death registration and census coverage." In the present article, the method is applied to mortality data for males in Kerala, India, over the period 1961-1971.

A procedure for comparing mortality measures calculated from intercensal survival with the corresponding estimates from registered deaths.

PIP: Presentation of a new technique introducing intercensal death registration data into the census survival calculation. The technique is applied to 1960-70 data from Thailand.^ieng