CHILD MORTALITY AND ITS COVARIATES
5.2. The covariates
5.2.3. Demographic variables
The demographic variables of interest in this study included sex of the child, children ever born to each mother, birth order of the children, birth interval, m other’s age at marriage, and m other’s current age at the time of the survey (Appendix table A5.3).
The categorical distribution of the observations for this study shows that 52 per cent of the children included in the analysis were male, 48 per cent female. The sex ratio of mortality of the child population is in the expected direction as found in other countries, with girls having a relative advantage over boys. As observed in almost all the low-mortality countries, there is a pattern of higher male than female mortality for children aged less than one and this pattern invariably persists for children aged 1-4 years (Dyson, 1977). In many countries with high fertility and mortality, however, the pattern reverses, leading to a higher female than male mortality in the post-neonatal period. Such findings have been reported in countries of South Asia, parts of Western Asia and in some populations of northern Africa (Dyson, 1977: Behm and Vallin, 1982: 28). Similar findings have been reported by Rutstein (1983: 25) using World Fertility Survey data from 29 countries including Pakistan. His findings suggest that for all countries together, mortality was one per cent lower for male children between the ages of one and two and was 4 per cent below that of females for boys between the ages of two and five years. Higher female than male child mortality has been reported in almost all studies conducted in Pakistan (Irfan, 1986; Rukanuddin and Farooqui, 1988; Ahmed et al., 1992).
Excess female child mortality has been attributed to various socio-cultural and economic factors which emanate from a strong son-preference attitude. Studies conducted in India (Basu, 1992) and Bangladesh (D ’Souza and Chen, 1980; D ’Souza, Bhuiya and Rahman, 1980; Chen, Huq and D ’Souza, 1981) provide evidence that intra-family allocation of food in favour of males results in a higher level of malnutrition among girls than boys. Different medical care was also found to be one of the major factors leading to higher female mortality.
As in almost all surveys conducted in Pakistan (Arnold and Sultan, 1992), the data collected for this study also suggest that the average number of children ever born to women towards the end of their childbearing years was more than six. The variable for this analysis was categorized into four distinct groups to determine as to what extent the total number of children born to a woman affected the survival probabilities of these children. In ascending order the categorizations of the children ever born in various subgroups are: less than three children; three to five children; six to eight children and nine or more children. The first category comprised 14 per cent of the total children; the second 46 per cent; the third 28 per cent and the fourth category 12 per cent.
The influence of birth order and birth interval on child survival have been issues of great demographic concern. In Pakistan, the analysis available on these variables suggests that first born children and those of higher birth orders are at a greater risk of mortality than the
others (Rukanuddin and Farooqui, 1988). Studies also suggest that a birth interval of less than two years can greatly increase the mortality risk of the previous or the next child (Cleland and Sathar, 1984; Rukanuddin and Farooqui, 1988). With an obvious interest in these variables, the two factors were included in the analysis to see what kind of impact these variables had on the child population. Four levels of birth order were entered into the analysis: first birth order; second to third; fourth to fifth; and sixth or higher.
Of the total children 26 per cent were of the first birth order; 40 per cent of birth order 2- 3; 21 per cent 4-5, and 13 per cent sixth or higher. Four categories were also created for birth-interval variables. The first category included 20 per cent of the total observations who were born at an interval of about one year; the second contained 32 per cent of children, with a birth interval between one to two years; the third category provided information about 12 per cent of children who were born between an interval of 2 and 3 years; and the fourth category included a total of 36 per cent of the children who either were born between three and four years after the previous child or were the first born.
Wolfers and Scrimshaw (1975) found that babies born less than 15 months after the previous birth were more likely to die as a result of prematurity. They concluded that the probability of survival can be longer and maximized if the pregnancy interval was between 15 and 38 months. Amongst many other studies, similar findings were reported by Winikoff (1983) and Puffer and Serrano (1975). Rutstein (1983: 35), using the World Fertility Survey found that children born less than two years after the previous birth were more likely to die even after surviving the first year of life. These findings suggest that a shorter birth interval leads to sibling competition and maternal depletion, especially in conditions of malnutrition. A shorter birth interval not only results in insufficient breastfeeding, depriving the preceding child of nutrients considered essential for an infant, but also prevents the mother from devoting her full attention to caring for either of the children. The examination of birth interval and child mortality also reveals that the birth interval tends to be shorter with the death of the child at the beginning of the birth interval. This is usually explained in terms of parents’ wish to replace the dead child as soon as possible. This effect, on the other hand, can be an unconscious consequence of abrupt cessation of breastfeeding and early resumption of ovulation (Knodel, 1968; Jain, 1969). On the effect of birth spacing on childhood mortality in Pakistan, Cleland and Sathar (1984) concluded that mothers who conceived the next child within twelve months experienced twice as much early childhood mortality as mothers whose children were born after a longer interval. They believe that wider spacing of births, between 24 and 35 months, in Pakistan can bring about a 15 per cent reduction in child mortality.
The most common finding on the association between birth order and child mortality is that the risk of mortality is high for first and higher-order births. The ‘U ’ pattern of relationship of birth order to child mortality was reported for many Asian countries (Trussed and Hammerslough, 1983: 16; Martin et al., 1983:422-424; Meegama, 1980). Similar results were found in some projects of the Pan American Health Organization (Puffer and Serrano, 1973:249-250).
Using World Fertility Survey data, Rutstein (1983:29-30) found a clear U-shaped curve for infant mortality with some variation amongst 29 countries. For example, it was found that only 16 of these countries showed higher first-order mortality than that of the second and third born and 19 of these countries showed higher mortality for the fourth and the sixth children. For four of the countries, Bangladesh, Haiti, Pakistan and Sudan, lower infant mortality was found for the seventh and higher birth orders. However, for all countries
together, a similar pattern of rising mortality with increasing birth order was found to be consistent for children between the ages of one and five years. More recently, Pant’s (1995), analysis on covariates of infant mortality in Nepal found a direct relation between birth order and infant mortality.
The pattern of mortality in relation to birth order is considered to be associated with the age of the mother. With marriage at an early age in many developing countries like India, Bangladesh, Nepal or Pakistan, there is higher mortality for the first birth to teen-aged mothers who are biologically, mentally and socially unprepared to bear and care for a baby. Babies of higher birth orders, born to older women, are at a greater risk of dying due to the reduced biological capacity of the mother along with greater physical depletion as a consequence of high parity, malnutrition, and other birth complications (Flegg, 1982:444; Rutstein, 1983:29) and higher birth-order babies are said to be affected by competition for food and other maternal and household resources.
M other’s age at marriage and m other’s current age were included in the analysis to see if these variables had any impact on the survival probabilities of the children. M other’s age at marriage was categorized into three groups: less than 20 years; 20-24 years and 25-31 years. Children born to women in the first category constituted 67 per cent, 27 per cent were born to women in the second category and 6 per cent were born to women married at 25 to 31.
M o th er’s current age is an indicator of the number of children born and is not as straightforward as the other variables; for studying the effects of the other variables its effect may need to be controlled. M other’s age was used as a categorical variable, being split into five categories: 15-19 years; 20-24; 25-29; 30-34; and 35-39. Children born to these women by current age were 2 per cent; 14 per cent; 28 per cent; 26 per cent and 30 per cent, respectively.
The overall results of various studies conducted in a number of countries including Pakistan show that early or late age at childbearing leads to an enhanced risk to the survival of the mother and the child. The age range of 20-29 years has been determined to be the safest, when complications of pregnancy and childbirth are at a minimum. Using the World Fertility Survey data from 29 countries, Rutstein (1983) found this U-shaped relationship for most of the 29 countries including Pakistan. The results show that infant mortality was much higher among mothers aged less than 20 years, was lowest for mothers aged 20-29 years and increased sharply for mothers who were more than 30 years of age. The U-shaped relationship persisted but was stronger for post-neonatal and child mortality (Rutstein, 1983: 15,26-27). A similar relationship was also confirmed in Sri Lanka (Meegama, 1980), Malaysia (DaVanzo, Butz and Habicht, 1983) and in the Inter-American study of mortality in childhood (Puffer and Serrano, 1975).