Factors Associated with Low Birthweight in Rural Bangladesh

Abstract

This study examines factors associated with low birthweight (LBW) in rural Bangladesh. Enrolled in early first trimester, 350 women were followed for duration of pregnancy and data gathered on maternal factors such as social, demographic, anthropometric, biochemical measures and newborn's birth weight within 48 hours of birth. Almost a quarter of babies (24%) were born with LBW and mean birth weight was 2961 g. Bivariate analysis found associations between LBW and mother's age, parity, weight and hemoglobin level at booking, weight gain and health problems during pregnancy, tobacco consumption, and gestational age. But no such association was seen for birth spacing, mother's height, economic status, educational level, body mass index, mid upper arm circumference and number of ANC visits. Multivariable analysis revealed gestational age, hemoglobin levels at first visit and weight gain during pregnancy as significant predictors of LBW in this rural setting. Although antenatal care provision is absolutely necessary, intervention approaches that go beyond clinical or primary care settings are also warranted for better nutrition of women. Concerted efforts in health and non-health sectors are necessary for improvement in health and social status of women in order to reduce low birthweight in Bangladesh.

Introduction

Low birthweight (LBW) is strongly associated with neonatal mortality and developmental problems in childhood and adulthood.¹ World Health Organization estimates that 90 per cent of LBW infants are born in less developed countries (LDCs), while developed countries have a rate of around 7 per cent.^2,3 LBW may be responsible for higher mortality figures in LDCs in two ways: one by increasing infant mortality rates, and two, as some studies have shown, by contributing to chronic disease in adult life.⁴

Although LBW seems to be a universal problem across LDCs, striking regional variations exist. For reasons that are not clearly understood, LBW rates are very high in South Asia with as many as 50 per cent of all newborn being LBW.⁵ Limited studies in Bangladesh have shown LBW rates ranging between 31 per cent and 47 per cent.^6,7 Factors associated with LBW in South Asia have been found to include pre-term births, maternal anthropometric measures, anemia, smoking, birth spacing, socioeconomic status, and antenatal care among others.^6–9 However, most of those studies conducted bivariate analysis only and therefore did not control for inter-relationships between multiple factors. Furthermore, these urban or hospital-based studies may not be comparable to rural settings where 85 per cent of Bangladeshi population lives. Determinants of LBW may vary across different settings and there is a need for analysis within different environments. The general paucity of community-based studies on LBW in rural Bangladesh prompted this research to examine the extent of the problem and variables influencing LBW so that appropriate intervention measures can be initiated for modifiable risk factors.

Methods

Data were collected in the years 2000–2001 in approximately 30 villages in rural Nabinagar/Nayerhat area of Dhaka district. The villages, situated in the service area of Gonoshasthaya Kendra (GK), a non-governmental organization (NGO) providing primary health care services, are representative of rural Bangladesh. Residents practice subsistence agriculture, more than two-thirds have no formal education, of Muslim religion (>80 per cent), and live in extended families.

Trained female health workers conducted house-to-house visits to identify women who missed their last menstrual cycle confirming pregnancy status through the pregnancy home test kit. Gestational age was estimated by asking the first day of last menstruation period. This short recall period reduced errors in estimation. The first 500 pregnant subjects were recruited at this first visit. At booking (first visit or enrolment), interviewers obtained anthropometric measurements, collected blood and urine samples, and administered a questionnaire to determine presence of other risk factors such as tobacco smoking/chewing and socioeconomic status. Health workers visited subjects once every month until the end of second trimester and every two weeks in the third trimester to record frequency of antenatal care visits, and various health problems during pregnancy such as ‘vaginal bleeding’, ‘urinary tract infection’, ‘diabetes’, ‘hypertension’, and ‘high fever’. Newborn's birth weight was measured for both hospital and home births. Birth weights of babies delivered in local hospitals or clinics were recorded from official documents. For home births, a village-level information network system consisting of already enrolled pregnant women, traditional birth attendants (TBAs), health workers, elderly women and members of women's groups (samity) ensured that newborns were weighed within 48 h of delivery. LBW was defined as less than 2500 g.

Of the 500 recruited, 150 women dropped out or were excluded from analysis for various reasons: abortion (n = 15), miscarriage (n = 12), stillbirth (n = 16), withdrawal from study (n = 5), moving out of study area for work (n = 10), incomplete information (n = 6), failure to weigh newborn within 48 hours (11), death of infant before weight measurement (n = 13), birth of twins (n = 7), and congenitally malformed babies (n = 5), woman moved to maternal home (outside study site) for delivery (n = 37), and floods preventing data collection (n = 13).

Independent variables were recoded into dichotomous categories based on commonly used cut-off points.^6–9 Association of LBW to different independent variables was investigated by bivariate analysis using Pearson χ-square test and odds ratio. Variables with bivariate associations at significance levels of p-value <0.2 were included in a multivariable logistic regression procedure. STATA software package was used for all statistical analysis and a p-value of <0.05 was considered statistically significant. Bangladesh Medical Research Council (BMRC) and Gono University, Savar, Bangladesh provided ethical clearances.

Results

Average birth weight was 2961g (sd ± 483.7) with 24 per cent (n = 84) of babies being of low birthweight (LBW). More than two thirds of mothers were between 20–29 years old with a mean gestational age of 38.1 weeks, average parity of 2.4 and mean weight of 50.4 kg at booking. Weight gain of mothers averaged 0.81 kg per month and 35 per cent of subjects gained less than 7 kg during the pregnancy. Mean hemoglobin level of mothers was 8.94 g/dl at enrolment, and 9.78 g/dl in the third trimester. More than half (53.4 per cent) of the women used smokeless (chewing) tobacco in the form of zarda (sweetened tobacco), ala-pata (tobacco leaf) and gul (dried and powdered) with pan. Detailed profile of pregnant mothers is in Table 1.

Table 1

Profile of mothers

Variables	Frequency	Per cent
Age in years:
16–19	53	15.1
20–24	142	40.6
25–29	92	26.3
≥30	63	18.0
Schooling:
No education	151	43.1
Primary	64	18.3
High School and above	135	38.6
Weight in kg:
<40	40	11.4
40–49	128	36.6
50 and above	182	52.0
Height in cm:
<145	51	14.6
146–150	137	39.1
>150	162	46.3
BMI:
<20	116	33.1
≥20	234	66.9
MUAC in cm:
<22	92	26.3
≥22	258	73.7
Antenatal check up:
<6	161	46.0
≥6	189	54.0
Health problems:
Yes	115	32.9
No	235	69.1
Weight gain:
<7 kg	123	35.1
≥7 kg	227	64.9
Hb per cent:
<8 g/dl	167	47.7
≥8 g/dl	183	52.3
Parity:
0	123	35.1
1–3	185	52.9
4 and above	42	12.0
Tobacco chewing:
Yes	187	53.4
No	163	46.6
Birth interval:*
<24 months	43	18.9
≥24 months	184	81.1
Socio-economic class:
Poor	147	42.0
Middle class	184	52.6
Rich	19	5.4
Gestational age:
<37 wks	55	15.7
≥37 wks	295	84.3
Iron tablet:
Poor compliance	93	26.6
Good compliance	257	73.4
Occupation:
Housewife	261	74.6
Others	89	25.4

*Primipara excluded.

Table 2 shows the results of testing the association of 15 factors found to influence LBW in previous studies. LBW has bivariate associations with mother's age, weight and hemoglobin level at booking, parity, weight gain and health problems during pregnancy, tobacco consumption, and gestational age. No association was found between LBW and birth spacing, mother's height, body mass index (BMI) or mid upper arm circumference, economic status, educational level, and number of ANC visits.

Table 2

Bivariate associations of birth weight with different independent variables

	LBW			Normal weight			95 per cent CI of OR
	n	Per cent	n	Per cent
Age group in years:
<20	19	35.8	34	64.2	2.2 (1.11–2.45)**
20–29	47	20.1	187	79.9	–
≥30	18	28.6	45	71.4	1.6 (0.8–3.13)
Weight at booking:
<48 kg	46	40.0	98	60.0	2.1 (1.23–3.52)**
≥48 kg	38	18.3	168	81.7	–
Height at booking:
<150 cm	51	28.7	127	71.3	1.7 (0.99–2.88)
≥150 cm	33	19.2	139	81.8	–
BMI at booking:
<20	33	30.6	75	69.4	1.65 (0.95–2.83)
≥20	51	21.1	191	78.9	–
MUAC at booking:
<22	25	27.2	67	72.8	1.2 (0.7–2.24)
≥22	59	22.8	199	77.2	–
Parity:
0	33	26.8	90	73.2	1.6 (0.88–2.80)
1–3	35	18.9	150	81.1	–
≥4	16	38.1	26	61.9	2.6 (1.2–5.77)**
Weight gain:
<7 kg	40	32.5	83	67.5	2.0 (1.18–3.41)**
≥7 kg	44	19.4	183	80.6	–
Birth interval:1
<24 months	12	27.9	31	72.1	1.32 (0.57–3.03)
≥24 months	39	21.2	145	78.8	–
Socio-economic class:
Poor	39	26.5	108	73.5	1.3 (0.76–2.23)
Not poor	45	22.2	158	77.8	–
Mothers education:
0–5 years	59	27.4	156	72.6	1.67 (0.95–2.95)
>5 years	25	18.5	110	81.5	–
Hemoglobin at booking:
<8 gm/dl	51	30.5	116	69.5	2.0 (1.18–3.40)**
≥8 gm/dl	33	18.0	150	82.0	–
Tobacco consumption:
Yes	57	30.5	130	69.5	2.2 (1.28–3.86)**
No	27	16.6	136	83.4	–
Health problems:
Yes	38	30.9	85	69.1	1.75 (1.03–2.99)**
No	46	20.3	181	79.7	–
Gestational age:
<37 wks	23	41.8	32	58.2	2.8 (1.42–5.25)**
≥37 wks	61	20.7	234	79.3	–
ANC visit:
<6	43	25.9	123	74.1	1.22 (0.72–2.05)
≥6	41	22.3	143	77.7	–

¹Primipara excluded.

**Significant at 0.05 level.

Only three variables reached the statistical significance level (p<0.05) in the logistic regression multivariable model predicting low birth weight. Gestational age (OR = 2.76, 95 per cent CI = 1.12–6.77), hemoglobin level at booking (OR = 2.48, 95 per cent CI = 1.31–4.68), and weight gain during pregnancy (OR = 2.39, 95 per cent CI = 1.14–4.98) together explained 32 per cent of the total variation in the phenomenon.

Discussion

This study examined and analysed determinants of LBW in a community-based setting in rural Bangladesh. LBW, found in 24 per cent of births in this study, is within the range of 15 per cent to 30 per cent reported from South Asian countries such as India, Nepal and Sri Lanka.^9,10 However, mean birth weight (2961 g) is less when compared to some other studies.^6,11 There are limitations in that this study was conducted in a rural area with availability of antenatal services from a non-governmental organization (GK) and therefore findings may not be able to be generalized to all rural areas. Underestimation of pre-term cannot be denied as children who died in the early hours of their birth (before measuring the weight) were more likely to be preterm and LBW.

As expected, the factors that operate through maternal influences and pregnancy related pathways were more predictive given that they are most causally proximate to maternal health. Gestational age, maternal anemia at booking (enrolment), and weight gain during pregnancy were predictive of LBW. It is generally acknowledged that etiology of LBW is multifactorial, but the potential importance of any factor taken in isolation should be interpreted with caution as they are often interrelated. Recognition of the relative contribution of these predictive factors is required for appropriate allocation of limited resources and focused interventions in Bangladesh.

Gestational age, especially pre-term birth, is most consistently associated with LBW.¹² In developed countries, intra uterine growth retardation (IUGR) comprises one third of all LBW cases and pre-term accounts for the remainder two thirds, but reverse is true for less developed countries.¹³ The focus in less developed countries remains almost exclusively on LBW as it is considered to be one of the leading causes of stillbirths and perinatal mortality.^14,15 Accordingly, in rural Bangladesh, we found almost three quarters (73 per cent; n = 61) of all LBW babies to be at-term deliveries, and 16 per cent of deliveries were pre-term (<37 weeks). However, extensive research is being conducted to determine factors associated with gestational age, especially pre-term births.

Constitutional maternal characteristics such as maternal weight at booking (<45 kg) and weight gain (<7 kg) during pregnancy have been found to be consistently associated with LBW in various studies and settings, even after controlling other factors.^16,17 One study has discussed the high degree of sensitivity and specificity of maternal weight and weight gain in pregnancy as indicators for LBW in Bangladesh.⁶ Average weight gain during pregnancy is expected to be 10–12 kg during pregnancy,¹⁸ whereas in this study more than a third (35 per cent) of the sample had gained less than 7 kg. These risk factors can be improved if nutritional intervention starts well before conception and early in pregnancy.

As expected, anemia is high in rural Bangladesh, especially in multipara, women with short stature, and teenage mothers (data not shown). Association of anemia with LBW is also a consistent observation in several studies across developing countries.^9,19 Although, all pregnant women in this study received iron supplementation, yet the prevalence of anemia remained high. Informal discussion with some subjects and community workers revealed poor compliance with iron supplementation. This was related to fears and perceptions that intake of iron tablets would increase the size of the baby, which in turn would obstruct normal delivery of the baby at home where 80 per cent births take place.¹⁴ This was associated with fears of death of the baby or mother, and financial expenses associated with hospital deliveries and after-care. These perceptions may be an important barrier to reduction of LBW and require further qualitative investigation.

Antenatal care provision was not significantly associated with LBW. This could be due to effectiveness of ANC offered by the NGO and home visits by health workers that may have evened out the effect of ANC on LBW. However, based on our findings it is clear that although provision of antenatal care and primary health care clinics is necessary, it may not be sufficient for reducing the burden of LBW. The modifiable risk factors that predict low birth weight pertain to the general health status of the mother including weight, weight gain and anemia. Thus, prevention of LBW in Bangladesh requires intensive nutritional programs to improve the general health, weight status, and reduce anemia in women of reproductive age groups. Programs directed at girls and women much before pregnancy are needed. Traditionally, health care providers have overemphasized screening for biomedical problems, while health education and health promotion were less narrowly focused.²⁰ Prevention of LBW requires interventions that go beyond efforts during pregnancy such as ANC and home visits. Long-term strategies of primary prevention involving improvement in cultural and social status of women within the family may be more effective than interventions at time of pregnancy alone.

We wish to express our gratitude to Bangladesh Medical Research Council for providing financial support to enable us to conduct this study. We would like to extend our special thanks to all the mothers and children who took part in the study.

Author notes

aGono University, Savar, Dhaka 1344, Bangladesh bTexas A&M University, College Station, Texas, USA cLouisiana State University Health Science Center, Pineville, Louisiana, USA dGazipur Sadar Hospital, Gazipur, Bangladesh