Study setting
This study was conducted in 25 HPDs of Uttar Pradesh with a population of 69.6 million accounting for 35% of the state population [19]. These HPDs were identified as poor performing districts by the Government of India (GoI) through the ranking of districts within all states based on a composite index of health impact and outcome indicators in the areas of maternal health, child health and family planning from the Annual Health Survey 2010–11 data [20]. This also includes 6 additional districts geographically contiguous to the 19 identified HPDs having poor maternal and child health indicators. The 25 HPDs in Uttar Pradesh, thus identified, are spread across different regions and are the highest contributors to maternal and infant mortality in the state. The University of Manitoba and India Health Action Trust (an India-based non-profit organisation) [21] established the Uttar Pradesh Technical Support Unit (UPTSU) to provide techno-managerial support for improving maternal, neonatal and child health outcomes in these HPDs. Within the 25 HPDs, 100 Community Development (CD) blocks equivalent to sub-districts, from 294 CD blocks were selected as focused CD blocks for intensification and as learning labs. The UPTSU provides mentoring support to Front Line Workers (FLWs) at the community level and clinical staff at the facility level as well as health systems strengthening support to ensure adequate availability, utilization and quality of antenatal, intra-partum and postnatal care services for women and children. Frontline health workers include the Accredited Social Health Activists (ASHAs) who are the closest available health functionary for the community. An ASHA covers a population of approximately 1000 people in a village and the closest available health facility for a village i.e. a sub-centre has around 10 ASHAs in its catchment area wherein a sub-centre caters to a population of almost 11,000 on an average in the state of Uttar Pradesh [21]. An ASHA is responsible for generating demand and mobilization of pregnant women for health services. This includes counselling them on birth preparedness, safe delivery and antenatal and postnatal care practices.
Study design and participants
Community Behaviour Tracking Survey (CBTS), a cross-sectional quantitative survey, was conducted between June–October 2018 in rural areas of 25 HPDs of Uttar Pradesh by the UPTSU. The CBTS was designed to provide estimates of indicators of Reproductive, Maternal, Newborn and Child Health (RMNCH) program at block and districts levels in the selected geography. Accordingly, four population subgroups were chosen for data collection to measure different indicators under RMNCH. The survey methodology of CBTS is detailed elsewhere [22]. The Primary Sampling Unit (PSU) of this survey was the working area of an ASHA that caters to approximately 1000 people. Simple random sampling was used to select 20 CD blocks each among 100 and 194 CD blocks in 25 HPDs. Using a simple random sampling approach, 2811 primary sampling units (PSUs) within the selected blocks were chosen for one of the four survey groups i.e. the survey group with the maximum required number of PSUs. The required number of PSUs for the remaining survey groups were randomly selected from the already selected 2811 PSUs. For this analysis, data from a total of 2646 PSUs selected in Group-1 has been used. This process is depicted in Additional file 1. In each selected PSU, all households were visited to identify women who had a pregnancy outcome including live birth, stillbirth and abortion within 59 days preceding the survey. This timeframe was chosen to obtain more recent information to avoid possible recall bias. Overall, 13,908 women were identified based on the above criteria and 12,041 of them were interviewed (86.6% response rate). The majority of women not interviewed were not available during interview or had not returned from the facility or mother’s place (92.3%) and 0.7% had died during or after delivery or abortion. Among the interviewed women, 9458 women who had a delivery 2 months prior to the survey excluding those who had an abortion, were included in the analysis.
Research investigators were recruited and trained to administer a structured questionnaire in the local language (Hindi) to collect data on household and respondents’ background characteristics, relevant information on antenatal care, delivery care and postnatal care. The research investigators first determined the PSU boundary. With a random start, they visited all households and gave each household a number. They used a screening questionnaire in each household to identify women who recently completed a pregnancy, followed by informed verbal consent, and then, if eligible, proceeded with the implementation of the survey questionnaire.
Variables
The respondents were asked “During this pregnancy where did you plan to deliver: at home or at any health facility?” and the responses were recorded as “at health facility”, “at home” and “not planned”. Similarly, the respondents were also asked “did you or your family identify in advance a vehicle you would use to reach health facility for delivery or in case of emergency?” and the responses were recorded as “Yes” or “No”. Based on these questions, a categorical variable for birth preparedness was computed as 1) identified both health facility and transport, 2) identified only health facility, 3) identified only transport, and 4) not planned. Women who planned for home delivery were included in the category “not planned”. Women were considered to have birth preparedness for this analysis if they identified both health facility and transport to reach the health facility for childbirth. The primary outcome variable, institutional delivery, is dichotomous (Yes, No) and defined as delivery conducted in a health facility, including both private and public facilities. The study included relevant confounders based on existing literature to understand the effect of birth preparedness on institutional delivery [12,13,14,15,16,17,18]. This includes the number of visits for antenatal check-ups during pregnancy (0, 1, 2, 3, 4 or more), number of times the ASHA contacted the women during pregnancy (no contact, 1–2 times, 3 or more times), 1st trimester ANC registration (Yes, No), identified hypertension during pregnancy (Yes, No), identified anaemia during pregnancy (Yes, No), 3rd trimester antenatal check-up (Yes, No) as well as background characteristics including maternal age (< 20, 20–24, 25–29, 30+), years of education (no education/< 5 years, 5–10 years, 10+ years), parity (1, 2, 3, 4+), religion (Hindu, non-Hindu), wealth quintile (poorest, poorer, middle, richer, richest), caste (Scheduled Caste (SC)/ Scheduled Tribe (ST), Other Backward Class (OBC), Others). Caste is a social stratification system based on heredity that has existed historically in India. A caste can be equivalent to social class in India where the ‘backward class’ in Article 16 (4) of the constitution of India refers to OBC, SC and ST based on social backwardness which leads to economic and educational backwardness [23, 24]. The caste category “Others” has been used to denote castes whose members are economically and socially advantaged as compared to the socially disadvantaged caste category members (SC/ST, OBC). SC and ST caste categories for any state or union territory may be defined by the government of India through a public notification and the OBC list is notified by the National Commission for Backward Classes. Hence, while caste may be synonymous with socio-economic status, the wealth quintile has been considered as an appropriate methodology for quantifying it. Wealth quintile was computed based on the composite score derived from a set of household indicators such as access to drinking water (tap water, hand pump/protected well, others), access to toilet facility (flush, pit, no toilet), type of dwelling (pucca, semi-pucca, kaccha) and ownership of household assets (electricity, black & white television, colour television, mobile, land telephone, refrigerator, air conditioner, bicycle, motorcycle, car, water pump, tractor) using Principal Component Analysis (PCA) [25]. Participants were then categorized into five wealth quintiles: the first 20th percentile group, representing the relatively poorest quintile of the participants and 5th quintile, representing the relatively richest participants.
Data analysis
Descriptive analysis was conducted to describe the background characteristics of the pregnant women and estimate the proportion of pregnant women with birth preparedness. Bivariate analysis and multivariate logistic regressions were used to test the association of background characteristics and antenatal care related factors with birth preparedness as well as that of birth preparedness with institutional deliveries. Multivariate results are presented in the form of adjusted odds ratios (aOR) with 95% confidence interval (95% CI). All analyses were done using STATA version 15.1 [26].