The population census of 1981 and 1991 provide indirect estimates of infant mortality at the district level (RGI, 1997). Figure 5 shows one way scatter plot of district level IMR estimates from 1981 and 1991 census respectively. The scale of both scatter plots (1981 and 1991 census) have been fixed between 22 and 137. Estimates used to generate Figure 5 are given in table 1.

Source: RGI, Occasional Paper No.1 of 1997, Table-3.1, p-114. See table-1 in this paper.

These estimates bring out three important characteristics of infant mortality risk prevalent in the state. Firstly, the IMR has reduced in all districts. The band of one way scatter plot of district level estimates from 1981 census are towards 137 end of the plot. The estimates from 1991 census are scattered towards the lower side of the range. Secondly, there is wide inter district variation and disparity in levels of child health status. The district level IMR ranged from 70 to 137 in 1981 i.e, a difference of 67 infant deaths/1000 live births and 22 to 99 in 1991 which amounts to a difference of 77 infant deaths per 1000 live births. Thirdly, the inter district disparity appears to be increasing instead of narrowing. The difference between lowest and highest mortality districts increased from 62 infant deaths/1000 live births around 1981 census to 77 infant deaths per 1000 live births around 1991 census.

Estimates of IMR below the district level are not easily available. The SRS sample size is not large enough for disaggregated estimates below the state level. Vital registration data would have been an useful source for small area estimates, suffers from gross, under reporting. Recently a District Family Health Survey was piloted (Mahapatra, Rao & Kumar, 2000) in three districts of AP to estimate IMR of sub district level areas. This study shows substantial area wise variations in IMR. The district and division level IMR estimates from this study shown in Table 2 provide useful insights about differences in health status by geographical regions. Clearly the IMR is significantly higher in Mahboobnagar district at 115 / 1000 live births as compared to 65 and 79 in Chittoor and Nellore respectively. Infant mortality level in Nellore district (71-87/1000 live births) is close to the state average of 75 according to SRS 1999, and 72 according to NFHS, 1998. Chittoor has a slightly better situation with comparatively lower infant mortality.

Mahboobnagar is clearly much worse compared to the state level IMR. Obviously there are important socioeconomic and geographic differences in mortality experience of people in different parts of the state. Going down to the division level, the DFHS study found that four of the five divisions in Mahboobnagar district have IMR that is higher than the state average, and in only Wanaparthy division, the IMR is comparatively lower (DFHS, 2001). The IMR estimate for Nagarkurnool division is as high as 140/1000 live births corresponding to the state average IMR in the 1960s. Thus there appears to be a wide regional variation in infant mortality with in the state. Some areas of the state are clearly three to four decades behind in terms of their mortality experience. This shows the need for districts and divisional level estimates of IMR and its importance to know the exact determinants of IMR and to develop area specific interventions to reduce IMR.

Disaggregation of IMR estimates by socioeconomic status of the household is feasible only if both mortality and socioeconomic status data are available at the household level. The NFHS collected data on socioeconomic status of households and mortality experience. Table 3 shows the infant mortality rates according to mothers background obtained from the NFHS. Infant mortality declines substantially with increase in the standard of living. In households with a high standard of living the infant mortality rate was 43 deaths per 1000 live births and in households with a low standard of living the IMR was 97 deaths per 1000 live births (NFHS-2). The scheduled castes and scheduled tribes have higher rates of infant mortality compared to other backward classes and others.

The infant mortality rate declines sharply with increasing education of mothers, ranging from a high of 82 deaths per 1000 live births for illiterate mothers to a low of 49 deaths per 1000 live births for mothers who have at least completed high school.

The NFHS estimates are based on stratification of sample households by literacy status. This is ideal. But the problem with NFHS is its small sample size. The sample size reduces further as we stratify the sample by socioeconomic status. Another way to study of these relationship is to correlate socioeconomic indicators with mortality levels by small areas. Many other factors, apart from household level exposure, will affect both the socieconomic variables and mortality experience of a small area. Hence analysts generally attach lesser importance to correlational analysis compared to household level relationships between socioeconomic variables and mortality experience. However, these area wise estimates are generally based on larger sample size and hence are more reliable. The decennial census provide us with district level indirect estimates of IMR and direct estimates of female literacy level. In Figure 6 we have plotted female literacy ratio and IMR. The pattern of low IMR associated with high levels of female literacy is clearly visible.

Source: RGI, Census 1991. IMR = Indirect estimate from 1991 census. Female literacy = direct estimate from 1991 census

To examine relationship of IMR with socioeconomic development the CMIE infrastructure development index was plotted against IMR from 1991 census (Figure 7). The pattern is similar to the previous plot of female literacy rates and IMR. Districts like Hyderabad, Guntur that have high infrastructure development index show low IMR. Districts with low infrastructure development index like Mahboobnagar, Vizianagaram have high IMR.

Age of the mother is an important risk factor for infant and child mortality. Children born to mothers under 20 yrs of age are approximately 1.5 times more likely to die before their 1st birthday than children born to mothers in their 20s. Children born to young mothers are more likely to be premature, to have low birth weights, and to have delivery complications (Devitt et.al. 1996). Children born to mothers over the age of 40 are also at higher risk of death for a number of reasons, including an increased likelihood of congenital abnormalities and an increased likelihood of closely spaced births.

In AP, Infant mortality is 40 percent higher among children born to mothers under the age of 20 than among the children whose mothers are age 20-29 (84 deaths compared with 60 per 1000 live births). The age at which a women bears the first child affects these rates. IMR and MMR are high in the women who gave birth when they where between 15-19 years of age. The main contributing factor for this is their physiological growth which does not cater to the growing needs of the pregnancy. The low nutritional status also plays its part. As the age at marriage increases the child bearing age also increases and hence will aid to lower the IMR and MMR.

Source: NFHS-2 (AP), p-121, table-6.4

Figure 9 shows the infant mortality rates according to previous birth interval. Clearly births spaced less than 24 months after the previous child birth have a higher risk of infant mortality. The timing of successive births has a powerful effect on the survival chances of children in Andhra Pradesh. Infant and child mortality rates decrease as the length of the previous birth interval increases. When the intervals between births was 48 months and above the IMR is 33 and when the interval between births is less than 24 months the IMR increases more than three fold to 106 (NFHS-2).

Source: NFHS-2 (Andhra Pradesh) p-121, table-6.4

<< Back 

 Updated by Samatha Reddy Dated: 17/08/2003