Indian Pediatrics 2003; 40:653-659 

The demographic and socio-economic determinants of post-neonatal deaths (n = 475) in a special project area of rural northern India (Ballabgarh) were ascertained from 1991 to 1999 using the electronic database system of the project area for data extraction, and were compared with the eligible living children of the same age using a matched population-based case-control study design. Similar determinants were also ascertained in neonatal deaths (n = 212) using the same study design. After controlling for the potential confounders using conditional logistic regression analyses, lower caste (a proxy measure for low socio-economic conditions in rural India) was found to be significantly associated with higher post-neonatal deaths (OR = 2.21). Higher maternal age (>30 years) and fathers’ lower educational levels were significantly associated with higher neonatal deaths, in addition to higher post-neonatal deaths in the same area.

Key words: Case-control study, India, Neonatal death, Post-neonatal death.

The All India Institute of Medical Sciences runs a Comprehensive Rural Health Services Project (CRHSP) in rural Ballabgarh of northern India. Child mortality rates in this project area were lower than the national rural child mortality rates during 1991-1999, but post-neonatal mortality rates are similar to the national rural average(1). The proportions of post-neonatal deaths contributing to overall infant deaths are also higher in the project area, because of a proportionate greater decline in neonatal deaths(2).

The project area has a high vaccination uptake among children aged 12-23 months (more than 90% for each of the six antigens: BCG, DTP, OPV and measles) consistently sustained from 1990 onwards(2,3). This observation probably suggests adequate utilization of the existing health care services in the project area. However, it would be interesting to observe any underlying demo-graphic and socio-economic differentials across the populations, possibly contributing to a slower decline in post-neonatal death rates in this project area. The present study set out to ascertain the demographic and socio- economic determinants of post-neonatal deaths between January 1st 1991 and December 31st 1999, and to compare these determinants with those of neonatal deaths during the same period in this project area.

Subjects and Methods

In 1999, there were two primary health centers (PHCs) in CRHSP area each catering for about 37,000 people across 28 villages. Ten multipurpose health workers (male and female) delivered health services in the community. The multipurpose health workers registered births and deaths during routine domiciliary visits. In addition to continuous collection of routine socio-demographic data, the health workers conduct a census in May and June each year. All the available socio-demographic data, including births and deaths from the two PHCs, are entered into an electronic database system from 1987 to date(2).

Health supervisors randomly select 20% of households and crosscheck the information collected. In addition, the medical officers in the two PHCs crosscheck the data for completeness and accuracy in another 5% of households. This rigorous nature of data verification indicates the reliability and accuracy of the electronic database, although no objective validation of such data was undertaken in recent past. After initial verification, the electronic database was found to have complete information from 1991 onwards.

In April and May 2000, a population-based matched case-control study was conducted. The study population comprised children aged 0-365 days old from 1991 to 1999, as recorded in the electronic database system of the project area. Cases were "all" neonatal deaths (less than 28 days old) and "all" post-neonatal deaths (28-365 days old) between January 1st 1991 and December 31st 1999. The controls were living children, who were randomly selected from the CRHSP area, using a rigorous computerized matching programme specially written for the study objective, from a cohort of 17,500 eligible children children aged 0-365 days born between January 1st 1991 and December 31st 1999, as recorded in the electronic database. The controls were matched for age (date of birth of cases with 15 days deviation), gender and village on an individual basis, which was possible using the specialized computerized matching programme. A total of 475 post-neonatal and 212 neonatal matched pairs were recurited for the final analysis. The available demographic and socio-economic variables (caste, age, literacy, occupational status of parents and the number of living children in households), as recorded in the electronic database were collected for each case and control.

Conditional logistic regression analysis was done using SAS software(4). Both adjusted and unadjusted odds ratios (OR), with 95% confidence intervals (CI), were calculated. Reference categories were based on the most favourable outcome with regard to exposure status. The Trinity College Public Health Ethics Committee, Dublin and the All India Institute of Medical Sciences, New Delhi provided ethical approval for this study.

Results

Table I shows that post-neonatal deaths after adjustment were significantly higher when these socio-economic and demographic determinants were present: lower caste (OR = 2.21), middle caste (OR = 1.61), maternal age of more than thirty years (OR = 4.32), fathers’ aged less than 25 years or between 25 and 35 years (ORs = 3.78 and 3.05, respectively). Fathers’ with schooling of less than five years (OR = 1.72) were also significantly associated with a higher likelihood of post-neonatal deaths. The maternal age of less than 20 years was protective of post-neonatal deaths (OR = 0.49), as well as fathers who were involved in trading (OE = 0.57).

Table I

Socio-economic and demographic determinants of the post-neonatal deaths.

Variables	Cases No (%)	Controls No (%)	OR (95% CI) (Unadjusted)	OR (95% CI) (Adjusted)
Mothers’ age
<20	19 (4)	44 (9)	0.57 (0.32, 1.01)	0.49 (0.26, 0.94)
20-30	289 (61)	346 (73)	Reference	Reference
>30	167 (35)	85 (18)	2.36 (1,70, 3.28)	4.32 (2.49, 7.49)
Fathers’ age
<25	106 (22)	136 (29)	0.63 (0.41, 0.98)	3.78 (1.79, 7.99)
25-35	293 (62)	276 (58)	0.85 (0.58, 1.24)	3.05 (1.64, 5.65)
>35	76 (16)	63 (13)	Reference	Reference
Caste (a proxy for socio-economic status)
Lower	200 (42)	132 (28)	2.12 (1.53, 2.93)	2.21 (1.53, 3.18)
Middle	104 (22)	121 (25)	1.22 (0.84, 1.78)	1.61 (1.05, 2.45)
Upper	171 (36)	222 (47)	Reference	Reference
Mothers’ literacy skills
No schooling	378 (79)	349 (73)	1.51 (1.10, 2.06)	0.99 (0.68, 1.46)
Just literate	97 (21)	126 (27)	Reference	Reference
Fathers’ educational levels
No school	113 (24)	79 (17)	2.21 (1.51, 3.23)	1.53 (0.97, 2.43)
Schooling <5 yrs	209 (44)	179 (38)	1.78 (1.31, 2.41)	1.72 (1.20, 2.46)
Schooling >5 yrs	153 (32)	127 (46)	Reference	Reference
Mothers’ occupation
Non working	436 (92)	429 (90)	1.20 (0.77, 1.88)	1.33 (0.76, 2.32)
Working	39 (8)	48 (10)	Reference	Reference
Fathers’ occupation
Agriculture	184 (39)	149 (31)	Reference	Reference
Service	137 (29)	138 (29)	0.79 (0.57, 1.10)	0.81 (0.55, 1.19)
Traders	53 (11)	66 (14)	0.64 (0.42, 0.97)	0.57 (0.36, 0.91)
Unskilled	57 (12)	46 (10)	0.96 (0.61, 1.51)	1.03 (0.62, 1.72)
Others	44 (9)	74 (16)	0.43 (0.27, 0.69)	0.57 (0.34, 0.97)
Number of living children
One	95 (20)	140 (29)	Reference	Reference
Two	100 (21)	113 (24)	1.36 (0.92, 2.02)	1.11 (0.71, 1.74)
Three	88 (18)	82 (17)	1.56 (1.05, 2.31)	1.23 (0.76, 1.99)
Four	68 (14)	58 (12)	1.74 (1.12, 2.70)	1.08 (0.61, 1.90)
>/=5	124 (27)	82 (18)	2.28 (1.53, 3.39)	1.10 (0.61, 1.99)

 

Table II shows that neonatal deaths after adjustment were significantly higher in mothers’ older than 30 years (OR = 3.83), and in fathers’ with schooling of less than five years (OR = 2.28). Young mothers’ less than 20 years were protective of neonatal deaths (OR = 0.39), as well as in households having two living siblings (OR = 0.46).

Table II

Socio-economic and demographic determinants of the neonatal deaths.

Variables	Cases No (%)	Controls No (%)	OR (95% CI) (Unadjusted)	OR (95% CI) (Adjusted)
Mothers’ age
<20	22 (10)	32 (15)	0.74 (0.40, 1.37)	0.39 (0.17, 0.88)
20-30	126 (59)	156 (33)	Reference	Reference
>30	64 (31)	24 (12)	3.34 (1.91, 5.85)	3.83 (1.63, 9.02)
Fathers’ age
<25	70 (33)	69 (32)	0.34 (0.16, 0.72)	1.87 (0.56, 6.19)
25-35	107 (50)	132 (62)	0.26 (0.13, 0.56)	0.92 (0.33, 2.55)
>35	35 (17)	11 (6)	Reference	Reference
Caste (a proxy for socio-economic status)
Lower	59 (28)	68 (32)	0.71 (0.43, 1.19)	0.86 (0.47, 1.57)
Middle	45 (21)	47 (22)	0.77 (0.42, 1.40)	1.33 (0.64, 2.76)
Upper	108 (51)	97 (46)	Reference	Reference
Mothers’ literacy skills
No schooling	151 (71)	143 (67)	1.41 (0.91, 2.19)	1.06 (0.59, 1.88)
Just literate	61 (29)	69 (33)	Reference	Reference
Fathers’ educational levels
No school	47 (22)	21 (10)	3.64 (1.90, 6.94)	1.63 (0.94, 2.81)
Schooling <5 yrs	97 (46)	91 (43)	1.67 (1.08, 2.59)	2.28 (1.02, 5.10)
Schooling >5 yrs	68 (32)	100 (47)	Reference	Reference
Mothers’ occupation
Non working	196 (92)	200 (94)	0.75 (0.35, 1.58)	0.83 (0.34, 2.02)
Working	16 (8)	12 (6)	Reference	Reference
Fathers’ occupation
Agriculture	84 (40)	73 (34)	Reference	Reference
Service	62 (30)	70 (33)	0.75 (0.47, 1.20)	1.00 (0.58, 1.75)
Traders	26 (12)	27 (13)	0.80 (0.41, 1.56)	0.91 (0.42, 1.94)
Unskilled	20 (9)	9 (4)	1.90 (0.79, 4.54)	2.09 (0.78, 5.59)
Others	20 (9)	33 (16)	0.53 (0.28, 1.02)	0.76 (0.34, 1.73)
Number of living children
One	60 (28)	59 (27)	Reference	Reference
Two	34 (16)	59 (27)	0.59 (0.34, 1.03)	0.46 (0.24, 0.88)
Three	37 (17)	41 (19)	0.93 (0.53, 1.65)	0.75 (0.37, 1.53)
Four	31 (15)	24 (11)	1.29 (0.67, 2.51)	0.80 (0.35, 1.82)
>/=5	50 (24)	29 (14)	1.93 (1.02, 3.64)	0.86 (0.37, 1.98)

 

Discussion

The findings indicate that demographic variable (parental age), and socio-economic variables (caste and fathers’ level of education), are significant determinants of a higher likelihood of post-neonatal deaths in the CRHSP area from 1991 to 1999. The maternal age and fathers’ educational levels are also significant determinants of higher neonatal deaths in the project area during the same study period. The major strength of this study design is minimisation of selection bias in the study sample, because of a rigorous computerized matching programme. The greatest limitation of the study is that only those variables recorded in the electronic database were available for analysis. Hence, the possibility of confounding in the study findings is more likely, because of the unmeasured factors(5).

The study included ‘all’ neonatal and ‘all’ post-neonatal deaths from 1991 to 1999, as recorded in the electronic database, and hence no formal sample size calculation was done. However, the power of this study is more likely to be greater than 80%, based on an expected OR of two with an alpha level of 0.05 and an allowance of 25% for con-founding(6,7). Hence, the findings are less likely to have occurred by chance.

The variable lower caste can be used as a proxy measure for low socio-economic conditions in rural India(6,8). Post-neonatal deaths are twice more likely in families belonging to lower caste, indicating the role of underlying poverty prevalent in this society. This observation is also reflective of the underlying slower decline in post-neonatal deaths in the project from 1991 to 1999(1). Interestingly, post-neonatal death rate was used as an index for general poverty in study in the United Kingdom(9). Nonetheless, further explorations are necessary to test this hypothesis of using post-neonatal mortality as an adequate summary measure for population health.

Unlike other studies(10,11) mothers’ literacy status did not influence both the neonatal and post-neontal mortality. In contrast, the fathers’ educational levels influenced both the outcomes.

A higher likelihood of neonatal deaths in older mothers (more than 30 years), suggest that obstetric factors are more likely to influence this observation. Perinatal risk factors(12), such as pre-term birth, birth asphyxia, low birth weight and congenital anomalies are relatively common in older mothers, and these are the common causes for neonatal deaths in the project area, as reported recently(2). The probable causes for observed higher post-neonatal deaths in older mothers’ and younger fathers’ are, however, difficult to explain.

Younger mothers in the project area have a protective effect on both neonatal and post-neonatal deaths that may partly be explained by strong family ties and extended parental support to young pregnant women, consistent with findings elsewhere(13). Parental occupational status and large family size are not significant determinants of both neonatal and post-neonatal deaths in this population.

In conclusion, the study findings demonstrate that lower caste (a proxy measure for low socio-economic conditions) is an important socio-economic determinant of post-neonatal deaths in the project area that needs to be addressed for a possible greater decline in post-neonatal death rates in the future. In addition, maternal age and fathers’ educational levels influencing signifi-cantly on both neonatal and post-neonatal deaths merit attention for future child survival interventions in these groups of the society.

Acknowledgements

Technical support of Mr. Guresh Kumar for the computerized matching program; and assistance from Professors V.P. Reddaiah, S.K. Kapoor and Dr. K. Anand for providing me with data of Comprehensive Rural Health Services Project, All India Institute of Medical Sciences, New Delhi, is acknowledged. I also thank Professor John Kevany and Dr. Jean Long at the Department of Community Health & General Practice of Trinity College (Dublin) for assistance in my M.Sc. project design, and Dr. Kathleen Bennett at St. James’s Hospital (Dublin) for the conditional logistic regression analyses.

Funding: Ireland Aid Program, Department of Foreign Affairs (Ireland).

Competing interests: The author (ZK) was a Medical Officer in the Ballabgarh Project Area prior to his joining the M.Sc. program. This study is a re-analysis of data collected during my M.Sc. program at the University of Dublin (1999).

 References