Contraceptive use is the result of high fertility, not the cause of low fertility. And in a situation where family planning is widespread, it is simply the universal mechanism whereby fertility is controlled or not (Gray, 1994: personal communication).
as 'having no access to health services'. Children from the sampled panchayat where there were no health posts until the time of survey were also categorised as children having no access to health services'. On the other hand, as most of the urban centres were close to the district hospital, children bom in the urban sample area were classified as having access to
health services. However, this variable examines the infant and child mortality
differentials only by considering whether a household has any access to health services. The limitation of this variable is that it does not consider the extent of use of health care services delivered through health posts.
It is argued that educated women are likely to delay the timing of family formation and first birth (Cleland and Van Ginneken, 1989: 82; Lindenbaum et al., 1989: 121; Hobcraft, 1993: 161). On the basis of these premises it can be hypothesized that maternal education influences child survival through maternal age at childbirth and length of birth interval. Thus this chapter examines whether the two maternal factors, namely age of the mother at the time of childbirth and length of interval between births as described in the analytical framework of this study in Chapter Two, are operating as an intermediate factors between the socio-economic and outcome variables to influence child survival in Nepal. As contraception is likely to be one of the means of planning the timing of births, it is also worth examining whether educated women in comparison to those with no education are using modem contraception to avoid too closely spaced births.
4.3
Analytical approach
The preliminary analyses in this chapter which examine the direct effects of the socio-economic and health-related factors on infant and child mortality are based on cross tabulations. The statistical significance of the differentials in infant and child mortality according to the independent socio-economic and health-related factors are assessed by using the same univariate and multivariate logit model technique used in Chapter Three. This part of the analysis identified the socio-economic and demographic factors that have net significant influence on infant and child mortality. This is followed by the inclusion of proximate variables (maternal factors) into a model to examine how far the effects of socio-
economic and health-related factors are mediated through selected maternal factors in influencing infant and child mortality in Nepal. In the analysis a considerable attenuation in the estimates related to the social and health factors explaining their relationship with infant and child mortality can be expected after including the maternal factors in the model if the links among variables are working as hypothesized in the causal model shown in Chapter Two.
4.4
Socio-economic differentials in infant and child mortality
Table 4.1 shows the distribution of the number of infant deaths, the number of births exposed to the risk of dying, and infant mortality rates per 1000 live births classified according to socio-economic variables obtained from the 1976 NFS and 1986 NFFS data sets^. The higher risk of death to infants of mothers with no education, illiterate fathers, working mothers, parents who have never used any contraception and those bom in rural areas from both surveys is evident from Table 4.1. Infant mortality rates between 1966 and
1985 have declined by 41 per cent. The decline is statistically significant at the one per cent level of significance. The decline in infant mortality over the period according to categories of each variable is also statistically significant. Three variables used from the
1986 NFFS reveal the lowest exposure to death among infants of households which had more than three cows, owned half a hectare or more of land and had access to health posts. The risk of death among children in a household with 1-3 cows is marginally higher than for those which do not have any cows. The association between this variable and infant mortality in the multivariate analysis, however, appears to be more clear after controlling for the influence of other variables. The evident influence of socio-economic variables on infant mortality could be due to differences in nutrition, hygiene, health care and environment as a consequence of different household socio-economic status.
Table 4.1 suggests a higher risk of death among infants of working mothers as against not working mothers. In a study from 29 countries, using WFS data, Hobcraft et al.
5/ For infant mortality rates the 1976 NFS refers to the period 1966-75 and the 1986 NFFS refers to the period 1976-85.
(1984: 202-203) noted a higher risk of death for infants of mothers who worked outside the home in comparison to those who did not work, during the first months of life in 12 countries and beyond one month to one year of life in 14 countries in a. Furthermore, Hobcraft et al. (1984: 221) observed that the importance of the mother's work status was progressively declining in influencing child survival chances as the age of the child increased. The higher risk of death to infants of working mothers was also found in Lesotho (Banda et al., 1990: 7) and in India (Sandhya, 1986: 95). The evident differential in infant mortality according to work status of mothers in this study could be due to a different propensity for breastfeeding among working as against not working mothers. However this is largely unexplained because the information on breastfeeding was available only for selected births and is not explored further.
The influence of contraceptive use on child survival, as observed in this study, was also observed for Thailand (Frenzen and Hogan, 1982: 401). Frenzen and Hogan explained this result as the outcome of the length of birth spacing maintained by the use of contraception. Unlike the case of Thailand, the analysis for Nepal carried out in the later sections of this chapter did not produce any evidence of the influence of contraceptive use on birth spacing. Thus contraceptive use in this study may be explaining child survival prospects through utilization of health care services delivered in conjunction with family planning services. This seems to be possible on the grounds that parents who are contraceptive acceptors are also likely to use health services for their children from the same source as contraceptive services. The evident decline in infant mortality over the period could be due to the emergence of health posts in 1966/67 and the increase in health care delivery services through various institutions in many parts of the country after 1970. For example, Dasvarma (1984) for Indonesia argued that about 60 per cent of infant death and 71 per cent of death among children aged one to four could be avoided by the implementation of effective hygiene and preventive medical programs.
Table 4.2 shows the distribution of child deaths, the number of children exposed to the risk of death, and child mortality rates per 1000 (4ql) from the 1976 NFS and 1986