Formative research

During the formative research phase, a participatory action research (PAR) approach was conducted as a collaborative process involving the research team as well as the research participants comprising local political leaders, representatives from the provincial planning departments, district and local level personnel from the CRVS department and health sector and village headmen.16 17 Qualitative methods were used to compile information from the PAR process.11 Stakeholder groups from each location were first involved in participatory discussions using a structured CRVS rapid assessment framework developed by the WHO.18 The WHO rapid CRVS assessment addresses 11 broad themes covering 25 specific questions related to the design and performance of local CRVS systems. The objectives of the assessment were to evaluate characteristics, strengths and limitations of existing CRVS procedures, identify roles and responsibilities of each stakeholder group, and design specific system strengthening interventions. The specific questions examine existing laws and regulations, registration practices, mechanisms for ascertainment and coding of causes of death and data compilation and use. The framework assigns ratings for each domain as well as an overall assessment. At the discussions, stakeholder groups shared their knowledge and experience regarding CRVS system characteristics for each theme/question, leading to consensus opinions on system performance for each item, along with recommendations on specific interventions to strengthen implementation. These discussions led to the development of a revised process map for reporting births, deaths and causes of death at the local level, with clearly identified roles and responsibilities for institutions and relevant personnel. The formative research phase also involved development of guidelines and operational manuals for field implementation of the revised data recording and compilation procedures. The findings for each of the above themes are presented in the following sections.

CRVS assessment

The assessment identified that the CRVS systems in both districts were scored in the ‘weak’ category, with most aspects of the system requiring attention, particularly in regards to information on causes of death. Table 1 outlines the broad limitations of existing systems at district and local level and also includes several specific recommendations that illustrate the scope and nature of system strengthening activities that would be required.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Multisectoral Memorandum of Understanding between the Department of Population Administration, Ministry of Health and National Development Planning Agency for implementation of the CRVS system in Indonesia. CRVS, Civil Registration and Vital Statistics.

In regard to the legal framework, CRVS regulations at the central level have not been widely disseminated, although several provinces/districts have issued their own local Regulations that specify the obligation of citizens to register births and deaths. However, even in these locations, there is a need for active dissemination to all local stakeholders, in order to facilitate compliance. There is also a need to build awareness regarding aspects of the National CRVS strategy related to strengthening cause of death reporting as well as vital statistics production.

A key limitation noted during the assessment was the lack of awareness among stakeholders of their individual roles and responsibilities towards implementation of CRVS as well as the absence of any formal collective commitment by the lead agencies of participating institutions that could facilitate intersectoral collaboration. To this end, the participatory discussions were followed up by the production of a formal Memorandum of Understanding on implementation of CRVS activities, which endorses the commitment and collaboration between the principal lead agencies (figure 1). In general, it was agreed that there is a need for District CRVS Committees to guide implementation.

In regard to data compilation, it was noted that the health sector in various locations operates electronic systems for reporting births and deaths, and that Community Health Centre staff use their mobile phones to report such events at the village level. The assessment also identified the need to strengthen mechanisms for ascertainment of causes of death as well as for integrate the health sector and the SIAK (CRVS) systems, to ensure timely, complete and reliable data compilation across both sectors. This requires intersectoral collaboration for system integration and interoperability, and also the need to create alternate mechanisms for data submission from remote areas with limited internet access. In addition, guidelines are required for local level fund allocation to health sector institutions (eg, Community Health Centre, hospitals, district health offices) to enable their functions in data compilation, including causes of death. Finally, the assessment clearly identified the need for intensive capacity building for personnel from all stakeholder institutions, to ensure data quality.

Stakeholder analysis

In addition to attending to the CRVS system assessment tasks, participants also conducted a detailed analysis of the expected roles and responsibilities of key stakeholders at the central, regional and local levels. Table 2 presents the findings from the stakeholder analysis, outlining the roles of the various agencies. In broad terms, these roles and responsibilities are aligned with the details noted in the AKPSH Strategy. The specific roles of health sector institutions cover the key areas of reporting of births, deaths and causes of death, data management and analysis, all of which cover a range of technical as well as administrative functions. In particular, there is a need for close involvement of the Centre for Data and Information Technology to strengthen data compilation and processing. The role of the Health Insurance department was also discussed, and the need for its integration with the CRVS system in order to strengthen the completeness and accuracy of information for the purposes of both departments. Consensus was also developed around the roles of the National Bureau of Statistics in providing population data for fertility and mortality analysis as well as in production of annual demographic and vital statistics reports for national and international dissemination.

Death reporting process

A second key output from the participatory discussions was the development of a generic business process model for implementation of the reporting of births, deaths and causes of death at the local level, as shown in figure 2. Under Indonesian law, all births and deaths (at home or in health facilities) among residents are reported by households through community volunteers (RT/RW, that is, rukun tetangga/rukun warga) and local administrators to the nearest subdistrict or district Administration office, depending on availability of service. In parallel, these events are also monitored and reported by the village health nurse (bidan desa) to the local Community Health Centre. There was consensus among participants that to strengthen data completeness at the village level, health sector death records should be periodically tallied with those from the village notification system. Similar integration could also be possible between the District Health Office and the District Administration Office, for data verification and consolidation, with relevant updates to both sets of records. Subsequently, for each home death in the updated list of records, trained staff would then carry out VA interviews with household members of the deceased, using an electronic software programme. These Verbal Autopsy (VA) interviews could then be analysed to derive the cause(s) of death either by using specially designed computer algorithms or by trained physicians, following which the data could be submitted to the District Health Office. For deaths in health facilities, a medical certificate of cause of death is completed by the attending physician, and these are compiled and submitted by all hospitals to the local District Health Office on a periodic basis. In parallel, all unnatural deaths would be independently processed through the police and forensic investigation procedures, and the causes of these deaths could also be submitted to the District Health Office. The cause of death records from all three sources, that is, medically certified deaths, home deaths with VA derived causes and police system unnatural deaths could be compiled and assigned relevant codes from the International Classification of Diseases, Tenth Revision (ICD-10).19 Finally, the coded data will be analysed to derive periodic tabulations of deaths by age, sex and cause of death and then compiled at the relevant District Health Office. It was envisaged that with adequate commitment from all local stakeholders, such a process for reporting and compilation could be implemented successfully and would yield good-quality mortality statistics at the district level, which could be submitted to the national statistics office

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Generic district level health sector procedures for compiling cause-specific mortality statistics in Indonesia. *RT/TW=rukun tetangga/rukun warga; community social service volunteers. **Bidan desa=village midwife. ICD, International Classification of Diseases; SIAK, Sistem Information Administrasi Kependudukan; VA, Verbal Autopsy.

Field guideline development

Following the stakeholder discussions described above a draft set of field guidelines were developed for district/city governments to implement the revised cause of death reporting process, in conjunction with the local CRVS activities. The guidebook contains the roles of various stakeholders, steps that must be taken by the district/city government to organise the implementation, the work processes that would ensure system efficiency and fund requirements for the district/city to organise system implementation. The development of these guidelines involved several rounds of consultation, review and necessary modifications to ensure that all aspects for system strengthening are addressed. Some of the key modifications included the involvement of the forensic collegiate to strengthen reporting of deaths from injuries, the involvement of the Ministry of Communication and Information in management of the information systems, assigning responsibilities to the village head in reporting data on births and deaths and providing population documents in health service facilities. This guideline serves as a guide and reference for local governments to be able to independently implement the national AKPSH strategy.

Implementation research

In each district, data were collected for all deaths that had been reported over the previous 12 months. In this analysis, we report data collected for Malang District during the period from 1 January to 31 December 2019, and for Kudus District during 1 January to 31 December 2020. As had been noted during the CRVS assessment, the weakest aspects of CRVS in districts/ cities are data collection, diagnosis of causes of death and data management. Hence, large-scale training programmes were conducted for all field staff in each district, covering their respective duties in regard to CRVS operations. At the field level, Community Health Centre staff were first instructed on the processes to ensure that all records of deaths in their respective areas are duly noted in the health sector death registers, including periodic collation of death records with village registration personnel (RT/RW). For all home deaths, local paramedical staff were trained to conduct VA interviews using the SmartVA data computerised data collection tool which also has an inbuilt computerised diagnostic software programme to derive the probable cause of death SMARTVA.20 In parallel, general practitioners and specialist doctors from all government hospitals and private hospitals in the district / city were training in medical certification of cause of death for events that occurred in their respective facilities.21 Medical record staff from all hospitals were also trained in selection and coding of the underlying cause for each death, according to the standard guidelines prescribed by the World Health Organization (WHO).21 In each district, data management training was imparted to personnel from the information technology departments at the Health Office and Civil Registry Office, guiding them in to integrating all death records from both sources, applying relevant data cleaning and verification protocols, prior to data analysis.

Study site preparation and management

The field implementation programme was led by a Project Team from the Center for Health Financing and Decentralisation Policy. Table 3 shows the key characteristics of the two study sites. The preparatory activities included a series of preliminary coordination meetings with the district Administration and Health Offices (and at sub district level, where relevant), to introduce the overall programme and elicit cooperation for field implementation and conduct capacity building programmes. During field work, the Project Team conducted several monitoring and evaluation vists to each district, including field trips to villages, hospitals and Community Health Centre for on the job training, field supervision and troubleshooting support. For this project, all VA questionnaires, VA cause of death forms and hospital MCCDs were compiled and submitted to the Project Team office, where the data were verified, coded and analysed. Underlying causes for medically certified deaths were coded and aggregated to WHO Mortality tabulation list of 103 cause categories.22 The VA derived causes of death were mapped to this tabulation list, and the aggregated data set of home and health facility deaths was analysed to derive the magnitude and rank order of leading causes of death by sex and age.

Demographic analysis

The annual data on deaths by age and sex for each district were initially evaluated to assess the completeness of death reporting, according to the Adair Lopez (AL) method.23 For each sex, the AL model estimates death registration completeness as a function of its relationship with the registered crude death rate, an estimate of the true under-five mortality rate for the study population and the proportion of individuals aged above 65 years in the study population. Table 4 shows the input parameters used for each study site and the results from the assessment. For this analysis, the under-five mortality estimate from the reported for each of the study sites from the Indonesian Demographic and Health Survey (IDHS) in 2017 was used.24

As shown in table 4, both study sites demonstrated relatively high levels of completeness of death registration, which indicates that the overall processes for death reporting and collaboration between the CRVS and health sectors at the local level seem to be functioning adequately.

These completeness estimates were then used to adjust the reported age-specific death rates by sex at ages above 5 years for each study site, to derive adjusted life tables and related summary mortality measures.19 For ages below 5 years, the study data were substantially lower that the under-five mortality rates reported in IDHS 2017,which were used as adjustment factors for life table estimation.20 This table 5 shows the reported and adjusted mortality measures by sex and age for each study site. The adjustments for completeness have resulted in a reduction in life expectancies of about 3–4 years in Malang, and about 2–3 years in Kudus.

Causes of death

Table 6 shows that non-communicable diseases predominate the leading causes of death in both field sites, although there are variations in magnitude, particularly for ischaemic heart disease, and cerebrovascular disease. Diabetes mellitus as well as chronic obstructive lung disease was also observed to be among the leading causes, along with tuberculosis, pneumonia and kidney diseases. However, cancers as well as road traffic accidents were not identified as leading causes. In both districts, the causes of death from VA interviews for home deaths were analysed using the SmartVA diagnostic software.25 In Malang, the home deaths accounted for 75% of the total reported, whereas in Kudus, only 53% of the reported events had occurred at home. Given the uncertainty in VA cause of death assignment, there is a higher proportion of deaths from ill-defined causes in the study sample for Malang.

Benefits from formative and implementation research

The formative research undertaken for CRVS assessment, stakeholder analysis and development of the business process model for reporting deaths and causes of death as well as compiling statistics at the District Health Office prepared the foundation for the implementation research programme in the districts of Malang and Kudus. Furthermore, the formative research guided the feasibility in implementation, through local adaptation of international standard protocols for medical certification and VA procedures. In addition, the activity developed training resources and conducted capacity building programmes for village-level health staff, and in Community Health Centre, hospitals, and the District Health Offices, which is essential for sustainability of the overall programme.26 Indonesia has made a commitment under the United Nations Economic and Social Commission for Asia Pacific (UNESCAP) CRVS decade 2015–2024 initiative to accelerate the coverage of CRVS.27 The experiences and materials from these initial formative and implementation research activities have established a sound foundation for further expansion of system strengthening activities towards this target.

PAR experiences

Initial PAR activities helped develop a common understanding among all stakeholders of the importance of and need for local data, which helped focus attention on CRVS strengthening in Indonesia. Also, the PAR established central level intersectoral commitment with general agreement of roles and responsibilities of various agencies involved in CRVS operations. However, the implementation experience identified that there were persistent uncertainties as to which agency is actually responsible for the overall AKPSH programme at the central level. Also, while it is clear that the Health Office is responsible at district level, implementation tasks are variously assigned to different subunits such as Community Health programme, Healthcare Services Unit or the Health Data sections. There is a need for follow-up PAR activities to develop clarity and consensus on the protocols for field implementation, and these should be documented in standard instructions and regulations from the Ministry of Health and other relevant CRVS stakeholders.

Importance of intersectoral collaboration

The stakeholder analysis presented in table 1 demonstrates the multisectoral nature of CRVS operations. As per the AKPSH Strategy, this activity was led by the Health sector and implemented in collaboration with the official CRVS programme operated by the Administration sector. This was supported by the intersectoral Memorandum of Understanding (figure 2), which along with the AKPSH strategy is clear evidence of the importance given to CRVS development in Indonesia. There is a need for similar commitment from other stakeholders listed in table 1. Also, within the health sector, there is need for further consultation to select the specific department to lead the activity with coordination and communication with other health programmes and units.

Need for decentralised coordination

As is seen from table 3, there are a considerably large number of villages and urban wards in each district, which necessitated efficient coordination and data management to yield high levels of data completeness. This was possible because of the technical support provided Center for Health Financing and Decentralization Policy. Moving forward. there is a definite need to establish district-level CRVS coordination committees to provide necessary management and technical oversight. The district CRVS committee should include representatives from all stakeholders including Health, Administration Service, Communication & Information, Planning and Development, Population Control and Family Planning, Head of Government, and local hospitals, among others. At the subdistrict level too, there should be coordination between the Community Health Centre, village heads, police and local community volunteer organisations, to improve efficiency and participation in the reporting of births, deaths and their causes. Also, to support field implementation, it is necessary to formulate local government regulations at the district/city level, which mandate CRVS procedures. Specific instructions and procedures are also required to enable transmission of mortality data between the existing Hospital Information System and the District/City Health Office.

Strengthening cause of death determination for home deaths

In regard to the causes of death, the project demonstrated success in deriving a cause-specific mortality profile by sex for each district, with some differences in the magnitude of observed patterns across the two sites. One reason could be a difference in socioeconomic profile, with Kudus being more urban while Malang has a relatively more rural environment. The relatively higher proportion of deaths from ill-defined causes (21%) in Malang district was because 75% of the deaths had occurred at home and were investigated using VA procedures. There are known challenges in cause of death ascertainment from VA, due to the potential for respondents not being fully aware of all clinical details associated with the death as well as the likelihood of recall bias during the VA interview.28 In this project, the recall periods were within a ‘3 to 6’-month window after the event that is acceptable. Nevertheless, additional attention needs to be paid to the quality of VA interviews in order to improve the information content recorded in the questionnaire, which will enable more detailed diagnoses. Attempts could also be made to link deaths at home to information from previous hospitalisation or domiciliary clinical care during the terminal illness, which could improve the quality of information available to assign a probable cause of death. The VA questionnaire records information on healthcare access during the terminal illness, which could be potentially followed back to retrieve clinical information from medical records. The Project Team for CRVS must also develop official versions of technical manuals for VA, medical certification of cause of death and data management and analysis, which can be used to guide capacity building activities and field implementation.

Utility of quantitative data analysis

Dissemination of standard mortality indicators as per international norms from pilot study data demonstrates ‘proof of concept’ of the activity as well as availability of capacity for such quantitative analysis. Also, reporting of data completeness is essential for evaluating success of the implementation project. The AL method is reliable since it is based on empirical model based on historical evidence from populations with high-quality mortality data and uses simple and readily available data inputs. A limitation is that the resultant completeness estimate for all ages above 5 years does not enable interpretation of age variations.19 Nevertheless, it offers a useful general understanding of data quality. Quantitative results also indicate areas for additional data quality research, such as studies to evaluate the reliability and validity of reported causes of death, to assess their direct utility for health policy and monitoring.29 Such assessment could be done through epidemiological studies that compare study findings with causes of death noted on clinical records for a sample of cases.30 The findings from such validation studies could be used to correct these mortality profiles as well as to guide interventions to improve future data quality.31 There is also a need for a standard information technology software programme to enable data compilation, transmission, archival and analysis. This should be designed to generate timely and reliable mortality statistics by age, sex and cause on a routine basis at district, province and national levels, at least from an appropriate set of sample population clusters.32

Lessons for promoting sustainability of operations

This activity established the feasibility of implementing international practices for mortality data recording and compilation as well as demonstrated success through production of plausible mortality statistics for the study population. However, sustainability of these operations requires specific technical, human and financial resources, particularly for the health sector. For this initial activity, funding was derived from the Center for Decentralization Policy institutional research budget, which is not sustainable for scaled up operations across Indonesia. There is need for a standard funding mechanism from the Ministry of Health to cover the health sector costs under the AKPSH strategy, and for such costs to be provided as part of the routine district-level health operational costs. At the field level, there is a need for specific funding allocation to each Community Health Centre, to enable local level liaison with the RT/RW and other village-level personnel from the Administration sector for identifying death events, and for implementing the VA interviews for cause of death ascertainment. Health sector fund allocations are also required for technical support for capacity building, field supervision and monitoring and evaluation, and data analysis and dissemination. Similar attention to technical and budgetary support is also required for the local CRVS operations within the administrative sector.

Translating lessons into actions

The lessons learnt from these implementing experiences have informed a summary list of recommendations that are essential for further development of the mortality statistics programme in Indonesia, as documented in figure 3. These include a range of administrative reforms to institutionalise operations and strengthen coordination of activities, technological solutions to enable data sharing and integration and capacity building activities to improve data quality and analysis. Once a clear administrative, regulatory and technical basis for the AKPSH Strategy is developed, it can act as a strong foundation to rapidly scale up implementation across all districts/cities to expand CRVS coverage to meet targets of the UNESCAP CRVS decade 2015–2024. The lessons and experiences from this implementation research are useful for other countries in South East Asia and also aiming to meet the UNESCAP targets. The lessons also offer relevant guidance for all developing countries to address the urgent needs for data on cause-specific mortality for surveillance of COVID-19 and for other future potential pandemics.33

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

Recommendations for vital statistics development in Indonesia. AKPSH, Strategi Nasional Percepatan Administrasi Kependudukan untuk Pengembangan Statistik Hayati; CRVS, Civil Registration and Vital Statistics; UNSDGs, United Nations Sustainable Development Goals.