Background

The cause of death information in India is poor, particularly in rural areas due to the shortage of medical personnel and facilities. Most people do not die in health facilities, so cause of death data based on the “Medical Certification of Cause of Death” is insufficient to represent the whole country accurately. To overcome this gap, verbal autopsy (VA) was introduced as a part of the Sample Registration System in India. This was started in 2001 and has been providing excellent cause of death data in India since then. The latest report covers 2010-2013. This activity is co-ordinated by Registrar General of India for Births and Deaths Office, Ministry of Home Affairs, Government of India.

Current system of SRS based VA.

Sample Registration System (SRS) was established in the year 1969 with the objective to provide reliable estimates of birth rate, death rate and infant mortality rate at state level. SRS sample units are randomly selected to be representative of the population at the state level. The sample design in a uni-stage stratified sample random sample without replacement. As of 2014, SRS had 8861 sample units across the country with 4964 units being rural and 3897 being urban. These sample units are revised every ten years.At the time of initiation of a sampling unit, a baseline survey is conducted to obtain demographic details of the usual resident population of the sampling areas. Subsequently continuous enumeration is done by a resident or part-time enumerator for each of the sample unit who is usually an anganwadi worker/teacher and is paid honorarium for this work. An independent retrospective half yearly survey is carried out by the full time supervisors who are the regular staff of the state Directorate of census operations. One supervisor is usually assigned a set of 10-12 sample units for conducting half-yearly surveys.

The verbal autopsies are performed by the supervisors who have been trained by medical professional in the art of collection on causes of death.\

SRS currently uses three Verbal autopsy forms for its regular operations.

  1. Neonatal Form (0-28 days)
  2. Childhood Form (29 days to 14 years)
  3. Adults – Aged above 14 years.

In addition, it also fills maternal death forms. The supervisor interviews a close relative of the deceased and records responses for the close ended interview schedules as well as the narrative portion of the tool in which the relative describes in his or her own words events leading to death.

These filled VA forms are scanned and after quality checks are referred to independent trained physicians for assignment of cause of death. This occurs through a web enabled software developed for this purpose. These physicians undergo a web-based training in assigning causes of death before being empanelled. Once they enter the cause of death and code it as per ICD 10 codes, the data is transferred to RGI’s office where it is compiled, analysed and tabulated.

Critical Challenges

Sample Size

The study covers about 45,000-50,000 deaths in a year. The study’s use of the expanded Indian Sample Registration System allows for a large sampling framework, but despite the 7 597 sampling units and reported 2.4 million households the framework still only covers a fraction of India’s more than 1 billion people and more than 9.5 million annual deaths.
Inability to address rare deaths or special groups: The main focus of the MDS is adult mortality and the sampling framework captures a limited number of maternal deaths. This indicates the study may not have sufficient statistical power to thoroughly analyze maternal mortality and its connection with newborn mortality, especially in low mortality states like Kerala.

Tools Used

The verbal autopsy method operates with the assumption that most causes of death can be recognized by trained physicians based on descriptions of signs and symptoms provided by a close relative of the deceased. Identification is simple for causes of death with distinct symptoms such as tetanus and injuries but can be difficult to distinguish in cases which have symptoms common to many diseases, such as of the fever during malaria. Verbal autopsy is also not very useful in identifying causes of death in persons over 70 years of age.

Delay in Physician Coding

 Due to challenges in language and dialect, the physician coding tends to get delayed resulting in delayed preparation of reports. Some experts have suggested moving to algorithm based system for quicker assignment of cause of death.

There is insufficient medical expertise within RGI office (which is a part of the Ministry of Home Affairs) and this had to be sought externally(Center for Global Health Research,Canada). It established the basic infrastructure and tools for training, quality control and has published many publications in high impact international journals to provide this activity an international recognition. It also created a network of 15 Indian Institutions to work on this project.The initial two phases of this study were 1998-2003 and 2004-1014 to cover “million deaths” – hence the label of Million Death Study attached to this exercise. Now that this phase is getting over, Registrar General of India  decided that this project has to be led by an Indian Institution.
Based on exploratory discussion, the All India Institute of Medical Sciences, New Delhi was approached to undertake this activity. AIIMS New Delhi has done some pioneering work in the area of verbal autopsy and its faculty members have also been involved in the global collaborations to simplify verbal autopsies and uses of computer assisted programs for arriving at cause of death.