Improving coverage of civil registration and vital statistics, Bangladesh

Abstract Problem Bangladesh has no national system for registering deaths and determining their causes. As a result, policy-makers lack reliable and complete data to inform public health decisions. Approach In 2016, the government of Bangladesh introduced a pilot project to strengthen the civil registration and vital statistics system and generate cause of death data in Kaliganj Upazila. Community-based health workers were trained to notify births and deaths to the civil registrar, and to conduct verbal autopsy interviews with family members of a deceased person. International experts in cause-of-death certification and coding trained master trainers on how to complete the international medical certificate of cause of death. These trainers then trained physicians and coders. Local setting Kaliganj Upazila has an estimated population of 304 600, and 5600 births and 1550 deaths annually. Health assistants and family welfare assistants make regular visits to households to track certain health outcomes. Relevant changes Following the start of the project in 2016, the number of births registered within 45 days rose from 873 to 4630 in 2018. The number of deaths registered within 45 days increased from 458 to 1404. During this period, health assistants conducted 7837 verbal autopsy interviews. Between January 2017 and December 2018, 105 master trainers and more than 7000 physicians were trained to complete the international medical certificate of cause of death and they completed more than 12 000 certificates. Lessons learnt Training community-based health workers, physicians and coders were successful approaches to improve death registration completeness and availability of cause-of-death data.


Introduction
In 2018, Bangladesh had a population of approximately 165 million, of these around 104 million (63%) people were living in rural areas and about 48 million (29%) people were younger than 14 years of age. 1 Since gaining independence in 1971, the country has achieved significant reductions in child and maternal mortality and life expectancy has increased from an estimated average of 46 years to over 70 years in 2015. 2 However, the government needs reliable and up-to-date information on natality and mortality to guide policy and planning at the national and subnational levels. The most efficient way of generating such data is a national civil registration system, which officially registers all births and deaths on a continuous basis and generates vital statistics.
In 2004, a new birth and death registration Act was promulgated and in 2010, an online birth and death registration was introduced. 3 However, by 2013 only 39 646 births had been registered, representing 1.3% of the expected 3 million births. Between the inception of the online registration system in October 2010 and February 2014, the cumulative total deaths registered amounted to 103 443, compared with the expected 761 000 deaths in 2013 alone. 4 An assessment of the civil registration and vital statistics system in 2013 identified multiple reasons for the low registration rates of deaths and births, including absence of mechanisms for reporting home births and deaths; inadequate understanding among health facility staff of the need to report births and deaths; lack of awareness in the population of the value of registration; cumbersome registration procedures; and overlap and inconsistencies between paper-based and electronic registration. 5 The country's seventh Five-Year Plan 2016-2020 calls for a stronger civil registration and a vital statistics system to produce timely and complete birth and death data for the entire country. 6 Here we describe a pilot project to increase birth and death registration and generate reliable information on causes of death.

Setting
In 2016, the government estimated that some 15% of all deaths take place in health facilities and 85% in homes, the latter with limited or no medical supervision. 7 Although deaths in public-sector health facilities are reported through the health management information system, these deaths are not notified to the civil registrar and therefore not officially registered. Moreover, the causes of most of these deaths are not determined according to international standards as defined by the World Health Organization (WHO).
The government introduced a pilot project to strengthen the civil registration and vital statistics system and generate cause of death data in Kaliganj Upazila. This administrative area has an estimated population of 304 600 (2011 census), and approximately 5600 births and 1550 deaths annually. 8 Problem Bangladesh has no national system for registering deaths and determining their causes. As a result, policy-makers lack reliable and complete data to inform public health decisions. Approach In 2016, the government of Bangladesh introduced a pilot project to strengthen the civil registration and vital statistics system and generate cause of death data in Kaliganj Upazila. Community-based health workers were trained to notify births and deaths to the civil registrar, and to conduct verbal autopsy interviews with family members of a deceased person. International experts in cause-of-death certification and coding trained master trainers on how to complete the international medical certificate of cause of death. These trainers then trained physicians and coders. Local setting Kaliganj Upazila has an estimated population of 304 600, and 5600 births and 1550 deaths annually. Health assistants and family welfare assistants make regular visits to households to track certain health outcomes. Relevant changes Following the start of the project in 2016, the number of births registered within 45 days rose from 873 to 4630 in 2018. The number of deaths registered within 45 days increased from 458 to 1404. During this period, health assistants conducted 7837 verbal autopsy interviews. Between January 2017 and December 2018, 105 master trainers and more than 7000 physicians were trained to complete the international medical certificate of cause of death and they completed more than 12 000 certificates. Lessons learnt Training community-based health workers, physicians and coders were successful approaches to improve death registration completeness and availability of cause-of-death data.
Lessons from the field Strengthening civil registration and vital statistics, Bangladesh Moyeen Uddin et al.

Approach
In 2016, the government established a national civil registration and vital statistics coordinating group to oversee efforts to increase birth and death registration. The group consisted of stakeholders in health, civil registration, statistics, local government, information and technology, justice, legal affairs and the Cabinet Division, which is responsible for interministerial coordination. With support from the Bloomberg Data for Health Initiative, the group introduced several priority activities: (i) technical support for strengthening the office of the registrar general; (ii) training of health assistants and family welfare assistants to increase notification and registration of births and deaths in the pilot area; (iii) implementation of verbal autopsy to ascertain causes of deaths in the pilot area; (iv) utilization of the international form of the death certificate and medical certification of cause of death in hospitals across the country; and (iv) training and capacity development for statistical coding and analysis of mortality statistics.
In the pilot area, local government, health and relevant administrative officials and community representatives discussed the importance of birth and death registration and agreed to the intervention in the area. The health ministry designated health assistants and family welfare assistants as intermediaries between families and the civil registrar. These assistants make regular visits to households in their areas of operation to track the health of women and children (including pregnancy outcomes, immunization and child growth and development). Assistant health inspectors trained the assistants to record births and deaths during their household visits and to notify the events to the local registration office for official registration within 45 days of occurrence. The assistants helped the family to complete the registration forms and counselled them where to collect the associated certificates. They also counselled families about the importance of visiting the registration office to complete the registration and collect the relevant certificates.
In addition, when a death was identified, the health assistants arranged to meet with the family after the mourning period to conduct a verbal autopsy interview (Fig. 1). Small cash payments were provided to the health assistants and to their supervisors to compensate for the additional transport costs.
The health assistant used SmartVA, a tool developed by the institute for Health Metrics and Evaluation at the University of Washington. The tool consists of a verbal autopsy questionnaire and the Tariff2 automated diagnostic algorithm to identify probable cause of death. The health assistants uploaded completed interviews to a central server that ran the diagnostic algorithms and shared the probable cause of death with the health management information system. On ethical and confidentiality grounds, and in agreement with the communities involved, individual cause of death information from the verbal autopsy was not shared with either the family members or with the health assistants.
To complement the cause of death statistics from the verbal autopsy, the health ministry introduced WHO's International Form of Medical Certification of Cause of Death (2016 version). 9 The ministry used a cascade training approach, by first creating a cadre of master trainers to be responsible for the training of physicians nationwide. Starting in January 2017, experts in cause of death certification and coding from the University of Melbourne and the regional WHO Collaborating Centre in Sri Lanka conducted workshops for master trainers using training materials based on international standards adapted to local circumstances. Master trainers, who were physicians in 11 hospitals in urban and rural areas, learnt to complete the medical certificate of cause of death in accordance with international standards. At the same time, the health ministry established a new cadre of health information officers to carry out statistical coding of the underlying cause of death in accordance with the International statistical classification of diseases and related health problems, 10th revision (ICD). 10 Since ICD statistical coding is a new area of work in Bangladesh, the coding is being introduced through WHO's Start-up Mortality List. 11

Relevant changes
In the pilot area, the timeliness birth and death notification and registration have  Between January 2017 and December 2018, 105 master trainers and over 7000 physicians have been trained to complete the international medical certificate of cause of death. The physicians have completed more than 12 000 certificates and the data have been entered into the district health information system database. 12 The health ministry is now gradually rolling out medical certification of cause of death to all public and private hospitals. Analysis of the newly available cause of death data is currently under way.

Lessons learnt
Here we show that the Kaliganj model is an effective and efficient strategy for increasing birth and death notification and registration, as well as for generating cause-of-death data in settings with little or no data on cause of death patterns (Box 1). Health assistants and family welfare assistants rapidly became adept at using electronic devices to collect ver-bal autopsy interviews and this enabled the production of cause-of-death data within two to three months of death registration. This success suggests that this model is a feasible strategy for longterm, large-scale mortality surveillance and can be integrated within the civil registration system. 13 Replication of the interventions in additional administrative areas could significantly improve national birth and death registration completeness.
The government has invested financial and in-kind domestic resources to extend the pilot approach to a purposive sample of all subdistricts of Gazipur district and one sub-district of Mymensingh district (Trishal), covering almost 2 million people. In this expanded area, a total of 660 verbal autopsy interviewers and supervisors had been trained by the end of 2018 and over 18 453 verbal autopsies had been conducted. Analysis of cause of death distributions for the registered deaths is ongoing. Discussions are under way regarding the most efficient and sustainable sampling strategy for taking verbal autopsy implementation to scale nationwide. 14 The government has positioned civil registration and vital statistics as central to its goal of enhancing service delivery across the life course, simplifying administration and strengthening statistical systems. The use of health assistants and family welfare assistants as notification agents has led to significant increases in birth and death registration and generated data on deaths by age, sex and causes in settings where most deaths occur at home without medical supervision. Complementing these data with information on causes of deaths that occur in hospitals will provide empirical data on the evolution of the epidemiological transition in Bangladesh.
The Kaliganj model calls for multiple interventions across many sectors and levels, a high degree of interministerial collaboration, resources, technical knowledge and information, and technology support. A key finding from the pilot phase is the importance of sustained, high-level political leadership and commitment across sectors, particularly between health, the office of the registrar general and the Bangladesh Bureau of Statistics. The cabinet division has been pivotal by taking on a major role in policy advocacy, stakeholder coordination and progress monitoring. ■

. Summary of main lessons learnt
• Community-based health workers can, with appropriate training and support, successfully identify births and deaths, and notify them to the local civil registrar for official registration.
• Community-based health workers can conduct verbal autopsy interviews with family members using hand-held devices and upload the data to a central server that generates plausible causeof-death distributions.
• In health facilities, physicians can be trained to correctly complete the international form of the medical certificate of cause of death.