Study setting
This study was conducted in Chandigarh, a Union Territory (UT), located in northern part of India, which was having a population of 1,055,450 in 2011 [
16]. About 97% the population reside in urban area and 3% live in rural area [
17].
Study design
This study used following three approaches. Firstly, to estimate stillbirth rate, ‘catch and re-catch’ method was used [
9]. Secondly, to assess probable causes of stillbirth verbal autopsy technique was utilized followed by coding of the causes according to the International Classification of Diseases, Tenth Revision (ICD-10) and classification of causes according to the schema proposed by Lawn et al. (2011) [
4,
18,
19]. Thirdly, to determine the risk factors of stillbirths, a case-control design was adopted.
Data were collected using three questionnaires. First questionnaire was on verbal autopsy which was adapted from World Health Organization (WHO) verbal autopsy standard manual [
18]. The second questionnaire was prepared on the basis of review of literature to capture additional information from hospital records. And the third questionnaire was on Social Capital which had been developed by Narayan et al. (2001) and it was earlier used by World Bank [
18,
20‐
23]. These questionnaires were pretested and translated in local language before use.
Data collection
Four interviewers having graduate level qualifications were recruited and trained. The data collection was done in two steps. First step was to identify all stillbirths among residents of Chandigarh Union Territory (UT) from the records, and the second step involved locating the women in the community who had delivered a stillborn baby (case) and selection of women who had a live-birth (control) in the same neighbourhood as that of the case (Fig.
1).
To identify all stillbirths in Chandigarh UT, the data were collected from three sources: (a) five major hospitals (b) office of the registrar of birth and death known as Civil Registration System, and (c) community-based auxiliary nurse midwives (ANMs). After collecting data from above three sources, duplications were removed, and final list of stillbirths was prepared.
One control (live-birth) for each case (stillbirth) was selected systematically from the ANM’s Birth Register from the same neighbourhood area to which the case belonged, matching for the month of birth also. The controls were selected according to the serial number of the register to avoid selection bias i.e., first live-birth in the same month when stillbirth had occurred. After the selection of cases (stillbirths) and controls (live-births), the interviewers visited the households to interview the respondents.
A total of 301 stillbirths were line-listed. Out of these, 181 mothers could be contacted; rest of the mothers were found to be either shifted out of the city after the birth of the baby (n = 78) or could not be found at the address which was provided to the hospital (n = 42). A total of 205 live births (controls) were enrolled but 181 (88.3%) mothers could be interviewed. Twenty four mothers could not be interviewed due to their non-availability despite repeated visit (n = 19) and due to refusal to give consent (n = 5).
Statistical analysis
Following variable definitions were used in the study:
Stillbirth: As per World Health Organization (WHO) definition, operationally, stillbirth was defined as ‘a late foetal death with ≥ 28 completed weeks of gestation’. Recent estimates of stillbirths published in The Lancet were based on the same operational definition [
4,
5,
24]. The weight and body length were not taken into consideration to define stillbirth as it was not feasible to do so.
Antepartum and Intrapartum stillbirth: Classification of antepartum stillbirth and intrapartum stillbirth was done mainly based on history of ‘baby stopped moving’ and or ‘baby looked macerated’. In addition, responses to two open-ended questions were also taken into consideration, which were asked at the end of interview to ascertain the cause of stillbirth: (a) what did doctor tell were the causes of the death of this baby, and (b) in your opinion when did the baby die- before labour or during labour?
Socio-economic status: It was assessed using modified Kuppuswamy scale 2014 [
25].
Social capital: It captures the existence of community networks; civic engagement; local identity and a sense of solidarity and equity with other community members; and trust and reciprocal help and support [
26]. World Bank report in 2004 added few more dimensions, i.e., groups and networks; trust and solidarity; collective action and cooperation; information and communication; social cohesion and inclusion; empowerment and political action [
23].
Maternal hypertension: Mothers was diagnosed with hypertension before or during the pregnancy, or having pre-eclampsia or eclampsia during the pregnancy were included in this category [
27].
Foetal growth: Small-for-gestational-age (SGA), large-for-gestational-age (LGA) and appropriate-for-gestational-age (AGA) were classified using Fenton’s chart wherever sex and weight were available from the health record [
28].
Gestational age: It was estimated on the basis of the reported date of last menstrual period.
Preterm: A baby born less than 37 weeks of gestation was considered to be pre-term.
Congenital malformation: Following six verbal autopsy questions were used to identify congenital malformations; whether the head was not formed or smaller head, whether the head of the child was larger, any swelling or opening in the back of the baby, whether there was any defect in the palate, any deformity in the hand and feet or any other malformation.
Estimation of stillbirth rate
The number of live-births in the population of Chandigarh in 2014 were estimated by using the crude birth rate reported by the Sample Registration System (SRS) in 2014 and by projecting the population from 2011 census to year 2014 using the population growth rate reported by Census of India for Chandigarh. The number of stillbirths were estimated using capture-recapture method.
In the capture and recapture, Lincoln-Petersen unbiased formula (N) = [(A + 1) × (B + 1)/(AB + 1)] − 1 was used, wherein ‘A’ (
n = 276) denotes the number of stillbirths captured from the hospital records and Civil Registration System (CRS), ‘B’ (
n = 264) denotes the number of stillbirths captured from the community-based records of Auxiliary Nurse Midwife (ANM), and ‘AB’ (
n = 239) denotes the number of stillbirths captured by both the system [
9]. We have clubbed Civil Registration System (CRS) and Hospital data as one catch, because Hospitals report to Civil Registration System (CRS), hence, these cannot be considered as independent sources. Assumption of capture and recapture method is that sources should be independent and the chance of being captured by each source should be equal [
9]. The clubbing of data helped us to fill some missing data from one of the tertiary care hospital which had refused to share data with us. Stillbirth rate was defined as number of stillbirth per 1000 births in one calendar year.
Causes of stillbirths
Stillbirths were classified according to the method described by Lawn et al. (2011), where they classified the stillbirth into intrapartum and antepartum, and most likely maternal cause and foetal cause [
4]. The verbal autopsies and available health records of 181 stillbirths were reviewed by two community physicians separately. They assigned one foetal and one maternal cause code according to the International Classification of Diseases (ICD) 10th revision [
4,
19]. If there was a consensus code between two community physicians, then that code was taken as a final cause. If there was no consensus between the two community physicians, a third community physician of a higher rank reviewed the case and assigned a final cause.
Risk factors of stillbirths
Statistical differences in the characteristics of cases and controls were compared by t-test for quantitative variables and by Pearson’s chi square test or fisher’s exact test for categorical variables. Logistic regression models were fitted to adjust for the effects of confounding variables. All variables were considered while arriving at the final model to evaluate the risk of stillbirth by estimation of adjusted odds ratio (aOR) with 95% confidence intervals (CI). Data were analysed using SPSS 22.0.0 (Statistical Package for the Social Sciences).