Study design
The cross-sectional study design was used to describe the magnitude and associated factors of LBW.
Study area and period
The study was carried out during September 2016 to June 2018, in different districts of West Bengal state, India. West Bengal is located between 85° 50′ and 89° 50′ east longitudes and 21° 38′ and 27° 10′ north latitudes. The state capital is the metropolitan city of Kolkata (formerly Calcutta). As per the report of the 2011 national census, West Bengal is the fourth most crowded state in India with a population more than 91 million [24]. This contributes to 7.55% of the country’s total population. The population density in this state is 1030 persons per square kilometer. The total population growth in this decade was 13.84%.
Sample size and sampling technique
The required sample size was determined using single population proportion with the following assumption: 22% prevalence of LBW in West Bengal according to National Family Health Survey-III (NFHS-III) [25], 99% confidence interval, and 3% marginal error. As a multistage, sampling technique was employed to select respondent, a design effect of 2 was used. Also, 20% was added for exclusions and another 20% was added for non-responses. Thus, the final sample size was 3554 (rounded off to 3600). From an original sample of 3600 pregnant women who were registered in National Rural Health Mission, 2611 were participated in the present study. The authors disqualified 989 respondents due to multiple reasons such as migration to another geographical area (342), home delivery (251), miscarriage (49), gestational age higher than 12 weeks at the first interview (212), unknown last menstrual period (119), and twin delivery (16). Women suffering from surgical condition were not included in the study.
In the present study a multi-stage stratified cluster sampling, followed by a systematic random sampling method were employed for selecting participants from three districts (Howrah, Purba Medinipur, and Paschim Medinipur) of West Bengal. Three blocks were selected randomly from each district followed by selection of eight clusters (community health center) (six from village and two from town/municipality approximately wherever possible) from each selected block. Then systematic random sampling method was employed to select the participants from the clusters.
Before start of the present study, the study team made a short visit to the study communities to meet local community leaders, head of the family, and to the pregnant women and brief them the purpose of the study, procedures of data collection, importance of this study, and risks associated with the study. The participants agreeing to take part in the study, written consent was obtained from them during household visits. Before commencement of this study, ethical clearance and prior permission was obtained from the Institutional Ethics Review Committee.
Study variables
The main outcome variable of the study was birth weight of infants. The independent variables include socio-demographic variables (residence, religion, cast, level of education, occupation, economic condition, and socioeconomic status of household, etc.), variables related to the maternal condition (mother age, mother weight and height, BMI, hemoglobin level, blood pressure, gestational period, birth interval, number of children, chronic illness of the mother, dietary counseling during pregnancy, addition habit, maternal antenatal care (ANC) follow-up, iron folic acid tablet consumption, etc.), and environmental health condition (availability of latrine, type of family, and number of individuals in the house).
Data collection procedure
Pregnant women were followed monthly. A pre-designed and pre-tested semi-structured questionnaire which was prepared by reviewing similar literatures was administered by house-to-house visit to the participants by the interviewers. The questionnaire included three sections. The first section of the questionnaire was related to socio-demographic background of the study participants. Second part of questionnaire contained questions related to the maternal condition, and third part of questionnaire was which help to assess household environmental conditions. The questionnaires were initially prepared in English and then translated in to Bengali. The Bengali version was again translated back to English. Prior to the actual data collection, the questionnaire was pretested on 50 pregnant women and that were not included in the main study.
The weight of the pregnant women was recorded by a portable weighing machine to the nearest 0.1 kg and height was recorded by anthropometer to the nearest 0.1 cm at the first interview at 8-12 weeks of gestation (considering negligible weight gain till 12 weeks of gestation). Hemoglobin of the respondent was measured with a finger prick sample of capillary blood and analyzed immediately using a hemoglobinometer (STAT-Site M Hemoglobin Analyzer, USA). Blood pressure was measured by auscultatory method, with the help of a sphygmomanometer (mercury type) and a stethoscope. Socioeconomic status was assessed by modified Kuppuswamy’s scale [26]. All babies were weighed using a digital non-hanging type salter scale and rounded to the nearest 10 g within 1 day of after childbirths.
Quality control of the study
To assure the quality of the data, properly designed data collection instrument and training of data collectors were done. The data were collected by five groups of examiners (15 male and 15 female field examiners; B.Sc. in Human Physiology, fluent in the local language (Bengali)) which consist of six individuals in each group. The principle investigators and the supervisor purposively selected the members of each group. The principle investigators and the supervisor were given 2 days training to the examiners on procedures, techniques, and collecting the data including measuring weight and height, calculating gestational age, etc. While collecting data, the instruments were regularly checked for their accuracy. The principal investigators go to field to check the interview and anthropometric measurement of the field examiners and supervised and reviewed every questionnaire for completeness and logical consistency and made corrections on the spot.
Statistical analysis
Frequencies and percentages were used for categorical variables to summarize data. Difference was assessed by employing Chi-square test for categorical variables. In order to investigate the association of predictor variables with outcome variable (LBW), both bivariate and multivariate analysis were used. To determine the effect of each variable on LBW, odd ratio was calculated based on 95% confidence interval in the adjusted and unadjusted model. All analyses were performed using the Statistical Package for Social Sciences (Version 20). A p value < 0.05 was considered statistically significant.