Indmedica Home | About Indmedica | Medical Jobs | Advertise On Indmedica
Search Indmedica Web
Indmedica - India's premier medical portal

Indian Journal of Community Medicine

Investigating Causes of Perinatal Mortality by Verbal Autopsy in Maharashtra, India

Author(s): Ragini Kulkarni, Sanjay Chauhan, Bela Shah, Geetha Menon, Chander Puri

Vol. 32, No. 4 (2007-10 - 2007-12)

ISSN No. 0970-0218

Ragini Kulkarni, Sanjay Chauhan, Bela Shah, Geetha Menon, Chander Puri


Objective: To investigate the causes and contributory factors of perinatal mortality by verbal autopsy in Maharashtra. Materials and Methods: Rural and urban areas in six districts in Maharashtra were selected by Probability proportional to size sampling. Verbal autopsies for perinatal deaths were conducted using standard tools and by visiting households; cause of death was assigned according to the International Classification of Diseases-10 using a standard algorithm. Statistical analysis was done using the SPSS-11 version software. Results: A total of 83 perinatal deaths (31 stillbirths and 52 early neonatal deaths) were investigated out of which cause of death for perinatal deaths could be assigned in 96.4% deaths. The leading causes of perinatal deaths were prematurity (19.3%) and complications of placenta, cord and membranes (12.9%) among stillbirths, while low birth weight (36%) and prematurity (26%) accounted for early neonatal deaths.

Keywords: Early neonatal deaths, perinatal deaths, stillbirths, verbal autopsy

One of the millennium development goals is to reduce under-five mortality rate by two-thirds, between 1990 and 2015. Estimates around the start of the new millennium put India’s under-five mortality rate as 95 per 1000 live births.1 Neonatal mortality that constitutes deaths between 0 and 28 days contributes to 46% of the total under-five deaths2 and 61.3% of the infant deaths. The WHO’s definition, more appropriate in nations with less well-established vital records of stillbirths is: late foetal deaths (28 weeks gestation and more) and early neonatal deaths (first week) in 1 year to the total number of live births in the same year, usually expressed as a rate per 1000 live births. Globally, some three-quarters of neonatal deaths happen in the first week after birth.3 Thus, nearly two-thirds of the infant deaths and about half of the under- five childhood deaths in our country occur in the neonatal period. In India, two-thirds (48%) of all neonatal deaths occur in the first 7 days after birth. Moreover, the early neonatal mortality rate (ENMR) for the year 2003 in India is 25/1000 live births, while for Maharashtra it is 19/1000 live births. The stillbirth rate in India and Maharashtra are almost similar, i.e., 9 and 10/1000 live births, respectively. Though these estimates are available, stillbirths are thought to be largely under-reported. The perinatal mortality rate in India is 33/1000 live births, while for Maharashtra it is 28/1000 live births (31 in rural and 23 in urban areas).1

The vital registration data on perinatal deaths, especially stillbirths, is inadequate in India. Medically certified cause of death for stillbirths and neonatal deaths is inadequate more so in the rural areas as a majority of these deaths occur in homes. In such cases, verbal autopsy can prove to be an essential tool for estimating the causes of death. It is an indirect method to ascertain the biomedical cause of death from information on symptoms, signs and circumstances preceding death, obtained from the deceased caretakers. This has found wide usage for ascertaining the causes of death without medical supervision in children.4 It helps in determining the cause and other associated factors that lead to death of an individual.

Various authors worldwide have documented the causes for perinatal mortality. The medical causes could be antenatal such as maternal diseases like hypertension, diabetes, cardiovascular diseases etc.; intranatal such as birth injuries, asphyxia, prolonged labour, obstetric complications; or postnatal such as prematurity, respiratory distress syndrome and congenital anomalies. The non-medical factors include financial constraints at individual level and education and occupation of the parents. High maternal age, primi-parity, high parity, smoking, low socio-economic status, being a single mother and early perinatal loss have been identified as risk factors for perinatal mortality. The biomedical determinants include maternal factors such as age, parity, antenatal care, bad obstetric history and pregnancy-related complications and perinatal and neonatal factors such as birth weight, gestational age, infection and congenital malformations.5

The present paper is an outcome of the multicentric task force project on Cause of Death by Verbal Autopsy coordinated by Indian Council of Medical Research (ICMR) during 2003-2005 in five states namely Maharashtra, Assam, Bihar, Rajasthan and Tamil Nadu. In this paper, an attempt is made to study the causes and contributory factors of perinatal mortality in Maharashtra.

Materials and Methods

Sampling design and data collection tools Sampling was done by dividing the state of Maharashtra into six geographical zones and selecting one district from each zone by PPS sampling. The districts selected were Thane, Pune, Jalgaon, Akola, Yavatmal and Bhandara. Villages or wards were selected based on their proportion in the district. The ICMR had conducted a pilot study to develop and test the data collection instruments for stillbirths and neonatal deaths and the diagnostic algorithms to arrive at the cause of death. These instruments in the local language (Marathi) included both narrative histories and structured questions related to a death from the relative of the deceased. The study instruments were separate for stillbirths and neonatal deaths. Both included items related to identification, unprompted history and semi-structured questions. The coordinating team from ICMR conducted a training programme for the field investigators. On the first day, they were trained on the technical aspects of data collection and questionnaires were explained to them. On the second day, field training in an urban and a rural area was conducted.

Data collection

Field investigators collected data using the standardized tools by visiting the households. The survey was conducted twice a year at six-monthly intervals. The recall period was fixed at 6 months (i.e., deaths occurring in the last 6 months). All the selected primary sampling units (PSUs) in each of the six districts were covered. Large PSUs with more than 2000 population were represented by selecting a central point and thereafter selecting 100 consecutive households towards north, east, west and southern direction. This ensured a representative sample of 500 households from each PSU.

The choice of respondent in all the stillbirths was mother, while in neonatal deaths, in majority of the deaths, parents of the deceased were interviewed. Wherever the medical certification and medical records were available, they were used in arriving at the cause of death. International Classification of Diseases (ICD) -10 codes up to a minimum of three digits were used.

Data analysis

The elicited information on probable cause of death was analyzed using SPSS-PC version 11 in relation to sociodemographic characteristics of the household, place of death, person conducting the delivery, type of treatment received etc. A trained doctor independently assigned the cause of death by reading the collected data.

Ethical aspects

Approval was obtained from the Institutional Ethics Committee of National Institute for Research in Reproductive Health, Mumbai. Written informed consent in the local language was obtained from the respondent during household survey. Confidentiality was maintained regarding the cause of death arrived at during verbal autopsy.


Data on 83 perinatal deaths (31 stillbirths and 52 early neonatal deaths) investigated by the verbal autopsy method is discussed in the present paper.


Among a total of 31 stillbirths covered, the leading causes were prematurity (19.3%), complications of placenta, cord and membranes (12.9%), maternal complications of pregnancy and anaemia in mother (9.6% each) and antepartum haemorrhage (6%). Stillbirths due to anaemia in mother were found in rural areas, while there were no stillbirths due to this cause in urban areas. In 22.5% stillbirths, proper antenatal care was not taken by the mothers as neither the blood was tested for anaemia nor was blood pressure recorded. In 6.5% stillbirths, cause could not be assigned.

Early neonatal deaths

There were 63 neonatal deaths covered in the present study, out of which 83% were early neonatal while 17% were late neonatal deaths. Among early neonatal deaths, the majority (59.6%) occurred on the first day of life and among these, the maximum occurred during the first hour of birth [Figure 1].

Figure 1:

Timing of early neonatal deaths

The leading causes of the early neonatal deaths were low birth weight (LBW) – 19 (36%), prematurity - 14 (26%), asphyxia at birth and diseases of the respiratory system (6.3% each). In 23 cases (44.2%), the respondents informed that the deceased were not shown in any health facility and among these, the majority (78.2%) were conducted by Dais and others (relatives). In three cases, physical abnormalities were reported in the babies. Certain undesirable practices were reported, such as application of oil/turmeric over the umbilical cord (10 cases), cutting the cord with grass-cutting blade/thread (two cases), feeding of milk with cotton, piece of cloth (two cases) and giving prelacteals such as honey with water, jaggery water etc. (five cases). Information about antenatal care during pregnancy indicated that in 15.3% cases, neither blood was tested nor B.P was recorded. In 5.7% cases, twins were delivered and they were conducted by Auxiliary Nurse Midwife (ANM) or trained dais. In these cases, the mother was also anaemic and had not taken iron tablets regularly. More perinatal deaths occurred in rural areas as compared to urban areas and perinatal deaths were more among males as compared to females though the difference was not statistically significant [Table 1]. Among stillbirths, 12.9% deliveries were conducted by untrained dais; all of them were in rural areas. Among early neonatal deaths in 17.3% cases, deliveries were conducted by untrained dais [Table 2]. There were more perinatal deaths in 21-25 years and more stillbirths in mothers less than 21 years of age [Table 3].

Table 1: Distribution of perinatal deaths by area and sex

  Type of death
Stillbirths Early
Rural 23 47 70
Urban 8 5 13
Total 31 52 83
χ2 = 3.85, P = 0.05, df = 1
Female 12 21 33
Male 19 31 50
Total 31 52 83
χ2 = 0.02, P = 0.8, df = 1

Table 2: Area-wise distribution of perinatal deaths in relation to delivery personnel

  Stillbirths Early neonatal deaths
Total Urban Rural Total Urban Rural
Doctor 19 (61.2) 7 12 21 (40.4) 5 16
Nurse 3 (9.6) - 3 8 (15.4) - 8
Trained dai 3 (9.6) - 2 13 (25) - 1
Untrained dai 4 (12.9) - 4 9 (17.3) - 9
Others 2 (6.4) - 2 1 (1.9) - 1
Total 31 (100) 7 23 52 (100)   35


In the present study, we experienced that trained nonmedical field investigators could collect data on causes of perinatal deaths by verbal autopsy method. If the field staff is adequately trained, verbal autopsy method is feasible and also helpful to study the causes of death, especially for deaths occurring at homes in our country. Several studies suggest that well-trained lay people can obtain accurate information when using culturally and linguistically appropriate questionnaires.6-8

A WHO consultation report9 mentions that sections/ core questionnaires for three different age groups (neonates, children and adults) are needed. Based on field experience with the different models, there was a general trend to include both verbatim and closedended questions in a core standard tool. In most of the reviewed questionnaires in the WHO report, the narrative section precedes the closed-ended part in order to reduce recall bias and the report of falsepositive symptoms in the verbatim part. It also mentions that in the close-ended section filter questions must be included, which are general questions that are followed by more detailed and specific questions if positively answered. In the present study, there were separate questionnaires for investigating stillbirths and neonatal deaths. Both verbatim and closed-ended questions were included and the narrative part was preceded by the closed-ended section which included filter questions. The inclusion of filter questions helped to avoid respondent fatigue.

The choice of respondent in all the stillbirths was mother, while among neonatal deaths in majority of the cases, the parents of the deceased were interviewed. Garene and Fountain10 have noted that the best respondent for investigation of a childhood death is the child’s mother if she is alive and present. In her absence or unavailability, father or foster parents would be appropriate.

In the present study, the recall period was fixed at 6 months. Garene and Fountain10 found in rural Senegal the adequate period of history recall to be 3-9 months after the death has occurred. Participants in the London School of Hygiene and Tropical Medicine (LSHTM) workshop11 believed that recall of circumstances of death would be adequate up to 12 months after the event.

Table 3: Association between maternal age and perinatal deaths

Stillbirths Early
<21 16 12 28
21 to <25 9 27 36
25 and above 6 13 19
Total 31 52 83
χ2 = 7.3, P = 0.02, df = 2

Among perinatal deaths, though there was no difference between the deaths in males and females statistically, the number of deaths was more among males as compared to females. The probable reasons for this could be a welldescribed biological survival advantage in the neonatal period among girls and under-reporting of the female perinatal deaths.

Age of the mother at pregnancy was an important factor contributing to perinatal mortality. Babies born with mothers at age less than 25 years were found less likely to survive as compared to mothers at higher age. Sixty-four (77.1%) of the perinatal deaths occurred in mothers having age at pregnancy less than 25 years. Overall 51.6% of the stillbirths were found in mothers having less than 21 years of age, which indicates that mothers in this age are often biologically ill-prepared for child-bearing.

Among the deliveries conducted by ANMs and dais (trained and untrained), obstructed labour due to malpresentation and complications of cord were the causes of stillbirths. Hence, training them for timely referral of these cases to higher levels could reduce the number of stillbirths.

Among early neonatal deaths, 44.2% of the deceased were not shown in any health facility and the majority of these were conducted by dais and others. This shows that seeking treatment at a health facility could reduce the early neonatal deaths. As majority, i.e., 59.6% of the early neonatal deaths occurred on the first day of life, intervention measures must be planned in this period to reduce deaths. In assigning the cause of death, in early neonatal deaths, prematurity was the leading cause of death (considered as birth after 27 but before 36 weeks of pregnancy). This period was as told by the respondent, and so may not be reliable. Secondly, in 7.6% deaths due to LBW, prematurity was also the underlying cause. In 22.5% of stillbirths and 15.3% of the neonatal deaths, neither blood was tested for anaemia nor BP was recorded. Thus, simple measures like blood testing for anaemia and recording of blood pressure can easily be done by the ANMs and trained dais and could prevent a large number of perinatal deaths.

More number of perinatal deaths were found in rural areas as compared to urban areas. One of the reasons could be the population covered in rural areas, which was more in comparison to that in urban areas ICMR task force group members. The other important factors were deliveries conducted by untrained personnel, improper antenatal care taken by the mothers and lack of proper facilities/infrastructure for management of premature and LBW babies at the primary healthcare level.


The authors are grateful to the following: ICMR for providing funding support; Dr Rajesh Kumar, Professor and Head, Department of Community Medicine, Post Graduate Institute of Medical Education and Research, Chandigarh; Dr Meghchandra Singh, Associate Professor, Department of Community Medicine, Maulana Azad Medical College, Delhi; Miss Namrata Agarwal, Senior Research Fellow and Mr Haresh Jadhav, Data entry operator for the project “Cause of Death by Verbal Autopsy”.

Contribution of authors

Ragini Kulkarni – Acquisition of data, analysis and interpretation, drafting the article. Sanjay Chauhan – Coordination of the study, data analysis and interpretation, revising it critically for important intellectual content. Bela Shah – Conception and design of study, final approval of the version to be published. Geetha R. Menon – Conception and design of study, revising it critically for important intellectual content. C. P. Puri – Final approval of the version to be published.


  1. Registrar general of India. Sample Registration System, Statistical report; 2003.
  2. National Family Health Survey – II. International Institute of Population Sciences: Mumbai; 1998-99
  3. Lawn JE, Cousens S, Zupan J; Lancet Neonatal Survival Steering Team. 4 million neonatal deaths: when? Where? Why? Lancet 2005;365:891-900.
  4. Soleman N, Chandramohan D, Shibuya K. Verbal autopsy: Current practices and challenge. Bull World Health Organ 2006;84:239-45.
  5. Joshi R. Perinatal and neonatal mortality in rural Punjab, working paper, August 2003.
  6. Chandramohan D. Verbal autopsy tools for adult deaths. [PhD Thesis] London School of Hygiene and tropical Medicine; 2001.
  7. Chandramohan D, Soleman N, Shibuya K, Porter J. Ethical issues in the application of verbal autopsies in mortality surveillance systems. Trop Med Int Health 2005;10:1087-9.
  8. Huong DL, Minh HV, Byass P. Applying verbal autopsy to determine cause of death in rural Vietnam. Scand J Public Health 2003;62:19-25.
  9. WHO Technical Consultation on Verbal Autopsy Tools, Talloires, France, 2-3 November 2004, Final report, Review of the literature and currently-used verbal autopsy tools, Department of Measurement and Health Information Systems Evidence and Information Systems, World Health Organization: Geneva; April 2005.
  10. Garene M, Oliver F. Assessing probable cause of death using a standardized questionnaire: A study in rural Sengal, Measurement and Analysis of mortality, New approaches. Clarendon Press: Oxford; 1989.
  11. LSHTM Verbal Autopsy workshop. Verbal autopsy tools for adult deaths. Workshop report, 11-15 January 1993. London School of Hygiene and Tropical Medicine (LSHTM); 1993.

Department of Operational Research, National Institute for Research in Reproductive Health (ICMR), Parel, Mumbai – 400 012, Maharashtra, India

Correspondence to:
Dr. Ragini Kulkarni,
Department of Operational Research, National Institute for Research in Reproductive Health (ICMR), J.M. Street, Parel, Mumbai – 400 012, Maharashtra, India.
E-mail: nirrhdor(at)
Received: 24.04.07
Accepted: 18.10.07

Access free medical resources from Wiley-Blackwell now!

About Indmedica - Conditions of Usage - Advertise On Indmedica - Contact Us

Copyright © 2005 Indmedica