Analysis of Birth Weights of a Rural Hospital

Background: Low birth weight remains a major reason behind childhood malnutrition. The NFHS findings show no dent in this problem. Objective: This study was undertaken to explore change in birth weights in a period from 1989 to 2007 and any associations thereof. Materials and Methods: All birth records of a private rural hospital spanning two decades (1989-2007) were analyzed for birth weight, age of mother, gender, birth order of the baby, proportion of pre-term babies and low birth weight babies. Results: No change was observed in the average birth weights (average 2.71 kg) over the period. Although the birth weight shows some expected variance with the age of mother, it was found to have no relation with the baby’s birth order and gender. The low birth weight proportion is about 24% and shows little difference before and after the series midpoint of year 1998. Conclusion: The birth weights have hardly changed in this population in the two decades.


Introduction
Birth weight is a major determinant of child's health and nutrition.In India, birth weight has remained low, with the NFHS reported proportion of low birth weight (LBW) babies about 23% for rural and 19% for urban population. (1)he proportion of LBW has improved only marginally from NFHS 1 (2) and 2 rounds. (3)However, NFHS data does not offer quantitative estimates of birth weights.The NFHS 3 reports association of low birth weight to place of residence (urban or rural), age of mother, religion and caste, birth order of the baby, education, wealth and use of tobacco.NFHS 3 reports LBW for Maharashtra state to be 22%. (4)However, internet search for birth weight studies from India yields very little literature, some of which is cited below.
A study from urban population of Bhilai reports LBW proportion of 23% which is rather high for urban population. (5).A study from South India using primary data from Ambedkar district in Tamil Nadu reports that the mean birth weights (2.77 kg) have shown a very marginal improvement (70 g) in the period from 1969 to 1993. (6).It has been observed from the same study that low birth weight (<2.5 kg) proportion reduced significantly from 27.2% to 15.9% in rural and 19.1% to 10.8% in urban areas over the decades.In a hospitalbased study in Kolkota (2005), the proportion of LBW was 34% with a mean BW of 2.64 kg. (7)The proportion of low birth weight babies in a Kerala study (1997) was found to be 18%. (8)stematic studies with standardized methods are rare in Indian literature.In this situation, secondary data from hospitals and gram panchayat birth records can be highly rewarding, though the issue of standardization of weighing scales and methodology in recording of birth weights are bound to remain a concern.Also the large proportion of home births create huge gaps in birth weight information, mainly as they are likely to be from disadvantaged communities and hence from the lower end of the birth weight distribution.We took up this study with an impression that the proportion of low birth weight babies must be declining over the two decades at least in hospital data.The secondary data available to us from this rural hospital should provide some insights.

Materials and Methods
This retrospective study is an analysis of the childbirth records of a private nursing home in Dindori block of district Nashik (Maharashtra) from 1989 to 2007.This block of Nashik district has a mixed population of tribal and nontribal communities.The block has both prospering grape growing villages as well as tribal communities migrating in search of seasonal labour after one paddy crop in monsoon.The block has two rural hospitals, 10 primary health Ashtekar, et al.: Birth weights of a rural hospital centers (PHC) and 66 health sub-centers.The block has 95 doctors (private and public sector) distributed in clusters. (9)he block has 393 Anganwadis and good transport infrastructure.Jeeps, cars and autos have now nearly replaced bullock carts for patient transport.Road traffic has apparently grown many times in the two decades.Mobiles are ubiquitous in the last five years even in tribal areas.The economy has visibly progressed.
The hospital in this study was started in 1988 with a view to offer rational ethical care to rural people.The Obstetric-Gynecology work dominated clinical work in all these years.This study includes all the 2586 births conducted in the hospital in this period.The families reported from several villages (14 to 133) in the two decades and had both tribal and non-tribal families.The case records were available in year wise bound books.Incomplete case records were deleted.Factors such as weight, mother's education, and hemoglobin were not mentioned on the birth record because they were recorded on the ANC (Ante Natal Care) record sheet, which was not available for this retrospective study.The baby weighing machines reportedly changed three times in the study period, but the same type and brand was used (weighing pan with spring balance).There was no conscious attempt for standardization or validation of the scale; hence instrument error cannot be eliminated.The observer remained nearly the same for most births except in the last five years when one more doctor joined the hospital.The spot location of the scale has also remained unchanged for the two decades-in the delivery room, ensuring the same eye level from the delivery table.The practice of weighing the baby after wrapping in old cotton sari-cloth adds about 25 g to the bare baby weight, (we found that two such clothes together weighed 50 g).

Results
The data covers period from 1989 to 2007 -19 calendar years.The number of births recorded in the first four years in the hospital has been clubbed since the numbers were small as the hospital was new in the locality and home births were the norm.1].Total number of twin births was 16 in the entire series.
Each of the twins has been entered separately for birth weight.The average birth weight, in our data, is 2.71 kg with an SD of birth weight 0.48 kg [Table 2, and Figure 1], which remains unchanged through the period with some exceptions.The data on birth weight of babies are available year-wise.This can be looked upon as a time series or one can make use of regression analysis where average birth weight (Y) is regressed on year (X).Since the number of children in each year is different based on which averages are worked out, the weighted regression analysis is appropriate.(Here 'weight' being number of children in that year.)In this analysis, year is on X scale and average BW on Y scale.The number of children in each age year varies from 1989 to 2007 [Table 3].We have fitted linear, quadratic and cubic polynomial to this data.This takes care of non-linear nature of data as seen in scatter plot [Figure 1], and the regression is shown in Figure 2. The analysis is shown below:  curve is also not a good fit.(We performed analysis in the presence and absence of two suspected outliers -data for year 2 and 14).Since this fit is not 'good' and all the regression coefficients are not significant, it supports the claim that birth weight has all most remained the same although there are some unexplained outliers.We conclude that the average birth weight has remained nearly stagnant over the study period.There is no gender difference in the average birth weight.The t-test for testing equality of means shows that the difference in mean birth weight for male (N=1335, mean BW 2.72 kg) and female babies (N=1114, mean BW 2.70 kg) is not significant (P value >0.05).
The overall proportion of low birth weight (<2.5 kg) babies in the period (2498 cases including pre-term babies) was 24%, with some unexplained spikes in the year 1995 and 1996.
The year 1998 is midpoint of the series.Incidentally an ultrasound machine was introduced in service in 1998 to improve ANC diagnosis.When tested for BW difference in the series before and after 1998; there was slight difference in proportion of low birth weight (LBW) babies before and after 1998 (25% and 24%) which was statistically not significant (P value >0.05).
The data on age of mother were available for 1947 cases.Only 2% of the mothers were below 18 years of age.The mention of age on the case paper is approximate and may not be very factual, as often happens in rural areas and more so when a woman arrives for childbirth in labor pains.

Discussion
Birth weight is a major determinant of childhood This study shows no influence of gender of the baby on BW.This, against the new understanding on gender differences on birth weight (mean for boys 3.46 kg and mean for girls 3.23 kg) in WHO Growth standards, (10) may be because of the overall low positioning of this sample in global distribution of birth weights.However, this difference may manifest only at optimal conditions.The order of birth also makes no difference to birth weight in this study.
Did the ANC make a difference to the BW?This is a difficult question to answer from our data.This is because we have no access to records of ANC visits.We also do not know how many of these women were registered for ANC with PHCs.Many unregistered cases need to be admitted for delivery in this hospital.We assume that the data is homogenous for hemoglobin (Hb) since many rural women have a starting level Hb under 8 g and rural women's weights do not vary highly.The proportion of low body mass index (BMI) for rural women in Maharashtra in NFHS3 is as high as 43%. (4)

Conclusion
Birth weights have not changed much in the study period in this rural block despite the changing economy of the region.We need to act on the proximate social-familial and gender factors for improving birth weights.

Figure 1 :Figure 2 :
Figure 1: Average BW as time series Figure 2: Regression of average BW on time series The work picked up from 1993 onwards, with a maximum of 202 births in 2003.The expected annual number of deliveries in the block (2.64 lakh general population in 2001) is 6600+ at a birth rate 25.Other deliveries took place at other hospitals including the PHCs and Rural Hospitals and at homes.The data (2586 births) represents about 2% of the childbirth cases in the block over the period of 19 years.The average BW is generally influenced by, among other things, proportion of pre-term babies.The data shows a variable proportion of pre-term deliveries (3% in 2002 to 15% in 1995) with an average level of 6% [Table

Table 4 : Distribution of BW according to age of mother Age Group(in Yrs) No of cases Mean BW (kg) SD of BW (kg)
The mean BW for this data is 2.71 kg.The average BW does not change across different age groups of mother [Table4].BW is also not significantly different between age groups, mothers below (BW 2.71 kg) and above 21 years (BW 2.72 kg).There is no difference in mean BW of babies of mothers below or above 18 years of age (P value 0.86).The correlation coefficient between age of a mother and birth weight of baby was weak (0.159), though statistically significant.