2.1 Study design and settings

This was a retrospective mixed methods study, which used facility- and community-based approaches to examine the deaths that occurred during pregnancy, childbirth, and puerperium. The study was conducted between November 2013 and October 2015 at Malda Medical College (MMC&H) and Hospital and Medical College and Hospital, Kolkata (KMC&H), which are situated in the northern and southern part of West Bengal, India, respectively. These two hospitals are the major tertiary-level healthcare units in eastern India, and they conduct more than 27,000 deliveries annually (https://www.wbhealth.gov.in/). The majority of pregnant women with complicated and irreversible problems from neighboring rural and urban areas, and also from other parts of India, visit these hospitals to use basic and comprehensive emergency health facilities. Therefore, these hospitals are a representative sample to help understand the contribution of eclampsia to maternal mortality in the community as a whole.

2.2 Data sources

2.2.2 Primary sources

Forty community-based maternal death reviews (20 women who died from eclampsia and 20 women who did not die from eclampsia) were performed in the community. The 40 cases were selected from the 317 maternal deaths that occurred in the studied hospitals. We used the residential address maintained in the selected hospitals to reach the respondents. We interviewed the family members and others who had witnessed any stage during the process leading to death. These people were mainly family members, neighbors, and relatives. In some cases, both sets of relatives from the parent's side and the in-law's side were also interviewed together. In primary data collection, this study adopted the maternal mortality questionnaire, which incorporated the National Rural Health Mission-validated community-based maternal death review questionnaire.

Information pertaining to the sociodemographic characteristics, health-seeking behavior, the referral history, premorbid condition, types of delays leading to maternal death, and strategy required to prevent maternal death were also obtained by using semistructured open-ended questions. This information was obtained from 12 health workers of various hierarchies who were associated with the maternal deaths or who participated in the provision of care to the deceased. In-depth interviews were performed in these 12 caregivers who comprised 2 accredited social health activists or auxiliary nurse midwives, 6 healthcare providers from first, second, and third levels healthcare facilities, and 4 were from the selected hospitals. Figure 1 shows the process of selecting interviewees. Most of the interviews were conducted in Bengali (regional language), but a few interviews were also performed in Hindi (official language of India). Written informed consent was obtained from all family members of the deceased women and also from hospital authorities.

2.3 Definitions of outcome and exposure variables

2.4 Data analysis

Bivariate analysis with the χ ² test was applied for the comparison between deceased cases of eclampsia and those without eclampsia in relation to various exposures. The binary logistic regression model was applied to estimate adjusted odds ratios (AORs) with their respective 95% confidence intervals (CIs) to identify the main confounders associated with death due to eclampsia. Differences were considered significant if p was <0.05. STATA version 13.0/SE was used for all statistical analyses. Personal interviews were conducted with the family members and medical staff. Most of the interviews were digitally recorded and later transcribed and translated into English. The English transcript materials were imported into QRS Nvivo, and categorized and analyzed by adopting the thematic approach.

The credibility, transferability, dependability, and confirmability of the data were taken into consideration to ensure the reliability and trustworthiness of the qualitative aspect of the study. The data were double-checked, the length of time during the interview was checked and a rapport was created with the respondents to produce valid data. The findings of this qualitative study were compared with quantitative findings. The qualitative findings were also corroborated with other similar national and global level studies. There was no controlled environment to conduct the research.

3.1 Underlying causes of death

The total number of maternal deaths in the selected health facilities was 317 over a 2-year period (Table 1). More than 70% of the maternal deaths were the result of direct obstetric causes, and 16.7% of these deaths were due to indirect obstetric causes. Eclampsia contributed to one-third of deaths, and it accounted for the leading cause of maternal death in West Bengal.

3.2 Sociodemographic and pregnancy-related characteristics in women with and without eclampsia-related death

Variation in the mortality rate started to increase steadily after 18 years of age and continued up to 25 years in women with and without eclampsia who died (Figure 2). The risk of dying in women during gestation and childbirth due to eclampsia-related causes was higher in those aged 19−25 years. The majority (85%) of deceased women who had eclampsia lived in rural areas (Table 2). Primigravidas (40%) appeared to have the highest risk of developing eclampsia. The number of eclampsia-related maternal deaths decreased as the gravidity increased, while the gravidity of women was inversely related to non-eclampsia-related maternal deaths. Similar findings were also observed for parity. The number of eclampsia-related maternal deaths was higher in women residing more than 49 km from the final hospital. There was no significant difference in the number of maternal deaths due to eclampsia and non-eclampsia causes in women who lived at a relatively shorter distance (within 0−33 km) from the final hospital (Figure 3). After a distance of 33 km, the variation in maternal deaths due to eclampsia and non-eclampsia causes increased with increasing distance up to 100 km, and then eclampsia cases surpassed non-eclampsia cases.

3.3 Heath-seeking behavior and obstetric characteristics in women with and without eclampsia-related death

A total of 81% of all deceased women who had eclampsia were referred from various other lower-level healthcare units (Table 3). The majority (approximately three quarters) of eclampsia-related maternal deaths were found in those without ANC visits, whereas 50% of non-eclampsia-related maternal deaths occurred in those without ANC visits. Additionally, only 6 deceased women who had eclampsia had full antenatal check-ups (three to four times), and 35 of 209 deceased women who did not have eclampsia used ANC three to four times. Of the 105 deceased women who had eclampsia, 13% sought medical care within 5 h of developing complications. However, 36% of these women sought medical care between 6 and 12 h, and more than half of them took more than 12 h to seek admission for management of complications. Eighty-three (79.0%) and 144 (68.9%) women were unstable and 22 (2%) and 65 (31.1%) were stable on admission in women with eclampsia-related death and those with non-eclampsia-related death, respectively. The most commonly observed mode of delivery was cesarean section in women with eclampsia-related death (54.3%) and in those with non-eclampsia-related death (46.5%).

3.4 Distribution and association of women with and without eclampsia-related death in relation to the type of delay

A total of 80% of women with eclampsia-related death and 56.9% of those with non-eclampsia-related death had any type of delay (Table 4). More than two thirds of women with eclampsia had delay in seeking care (type 1 delay), 46% had a delay in reaching a first level health facility (type 2 delay) and approximately one quarter had a delay in receiving adequate care in a facility (type 3 delay). Type 1 delay contributed to two thirds of death due to eclampsia, but contributed to only 39% of death due to non-eclampsia causes (Table 4).

3.5 Factors associated with maternal death related to eclampsia

The binary logistic regression showed a robust association between gravidity (AOR: 0.59; 95% CI: 0.53–0.96), ANC (AOR: 2.13; 95% CI: 1.19–3.12), mode of delivery (AOR: 2.41; 95% CI: 1.27–3.23), and any delay (AOR: 0.60; 95% CI: 0.27–0.89) for eclampsia-related maternal death (Table 5). The odds ratio of eclampsia-related maternal death was two times higher in women without ANC than in women who received ANC.

4.1 Antenatal history, complications during pregnancy and awareness

The community-based study showed a mixed reaction towards ANC check-ups. Some respondents replied favorably regarding ANC check-ups and some replied against it. All of the patients had a history of access to ANC. Many of them even had access to three or more ANC visits. However, many of the deceased women's relatives never took the women to more than one ANC session. Some of them visited two thirds of 1 km from a health subcenter for antenatal check-ups. Some women also visited private hospitals and some even consulted homeopathic doctors. The antenatal and complication history is discussed in the context of patients with eclampsia and those without eclampsia for better understanding the circumstances leading to death.

An important finding from the interviews is that women who had suffered from eclampsia were deprived of ANC at the second trimester. One of the respondents reported “Our ‘Boudi’ was admitted to Ramharipur PHC for a few days in her first trimester, but after discharge, she did not have any problems. Therefore, we did not take her there any more. We took her to the hospital once more when she developed a severe headache 1 to 1½ months before her expected day of delivery” (deceased 22-year woman's sister-in-law).

However, the majority of deceased women without eclampsia had a history of using ANC in all three trimesters. When describing ANC events of a deceased woman without eclampsia, her husband noted that “We wanted our daughter-in-law to have a healthy and safe delivery. Therefore, we took her every time for a check-up” (narrated by the woman's mother-in-law).

In this study, most relatives of deceased women with eclampsia were not acquainted with symptoms of high blood pressure, headache or swelling. One respondent reported that the pregnant woman complained of a headache, and swelling of the face and limbs as follows: “Our daughter-in-law was sitting in the courtyard and suddenly complained of uneasiness and blurring of vision. We took her to a private doctor. The doctor diagnosed her with high blood pressure. We did not know before this diagnosis that the complication was due to a high blood pressure” (father-in-law of the deceased woman).

4.2 Analysis of the referral history

Community-based in-depth interviews showed that majority of the deceased women had been referred from level 1, 2, or 3 healthcare units and reached the studied hospitals in a critical condition. This referral chain was categorized into two schematic overviews of eclampsia and non-eclampsia for better understating.

Eclampsia schematic overview: Most of the women who died due to eclampsia had been taken to more than three health facilities to reach Kolkata or Malda hospital (Figure 4a). A mother-in-law of a 20-year-old deceased woman reported the following: “Our daughter-in-law complained of blurring of vision on Sunday at 4.00 p.m. In the evening, we took her to a private doctor's clinic, and he gave her medicine. He said that she would be alright, but she developed complications again late in the night. In the morning, we took her to the primary health center, which was 2 km away, and she was admitted there. She was referred to the secondary level health setup again at 5 p.m. when her condition deteriorated. We reached the local hospital at approximately 8 p.m. the second day. She was kept there for 2 days. When she was in a critical condition, she was referred to a higher-level hospital, which was a subdivisional hospital. A male neonate was delivered by cesarean section at the subdivisional hospital. Five hours after the delivery, the doctors told us that the patient was having convulsions and that she needed to be taken to Malda. Three hours after admission to Malda Medical College and Hospital, the patient died.”

Non-eclampsia schematic overview: Most of the deceased women who had eclampsia were referred to three or more hospitals before their deaths. However, non-eclampsia cases were referred to one or two hospitals (Figure 4b).

4.3 Reasons for the referrals

The main reasons for referring the patient were hemorrhage, eclampsia, and preeclampsia. Severe anemia, jaundice, and heart problems in pregnancy were also recorded in the admission documents. A poor socioeconomic status, early age of marriage, and lack of proper food intake aggravated the situation.

The irreversible condition of a patient is due to poor management in peripheral health institutions. In this study, the patients were transferred to the final hospitals without examining their blood pressure or other medical conditions. Sometimes, patients with eclampsia were only administered 5−6 g of magnesium sulfate instead of a 14 g dose. Such a situation was clearly stated by a medical practitioner in one of the final hospitals: “We do not doubt patients requiring cesarean section are not examined properly, and for managing f eclampsia, no proper doses are provided. These are the reasons that accelerate mortality.”

5.1 Strengths and limitations

This mixed method study has a number of limitations. The hospital records used in this study may not have been good quality, including incomplete entries. The number of ANC visits was analyzed in the study, but information such as clinical examinations, use of tetanus toxoid immunization and iron and folic acid tablets, could not be ascertained from the case records. This lack of information could have had a major effect on differentiating between deceased women with and those without eclampsia. A community-based maternal death review also has limitations because it solely depends upon the narration provided by family members. Therefore, this review is subject to recall bias. This study relied on community reports of signs and symptoms of illness to classify the medical causes of death. In spite of these limitations, this study contributes to understanding the severity of eclampsia in relation to maternal health in our society. Furthermore, our findings could help in identifying confounding factors and deficiencies in the healthcare delivery system, and might contribute to planning strategies and interventions for curtailing pregnancy-related death due to eclampsia.