Mortality in rural Vietnam : Validity of routine reporting and experiences from a surveillance system

Background: Findings from many low-income countries, Vietnam included, show that mortality estimates based on death data collected in routine systems have low validity. Aims: The overall aim of this thesis is to examine and discuss the validity of mortality estimates based on data collected in different systems in a rural setting in Northern Vietnam and to identify factors that can influence the reporting and registration of deaths, all in order to make recommendations for improvement of death registration. Methods: The studies were conducted in a rural district where a Demographic Surveillance Site (DSS), called FilaBavi, started in 1999. Four studies were undertaken. One study (paper I) used a quantitative approach to examine and compare validity of mortality estimates collected in the FilaBavi and in the existing routine systems. Focus Group Discussions (FGD) were held with community members (paper II) and with persons responsible for the routine death reporting systems (paper III) to gain in-depth understanding of perceptions of death and to identify obstacles that impeded death reporting and registration in the official systems. FGDs and qualitative interviews were used to learn about experiences of surveyors and bereaved persons involved in death inquiries (paper IV). Main findings: Out of 471 deaths detected by all methods in 1999 and 2000, the FilaBavi quarterly household follow-ups detected 470 deaths (99.8%). The Commune Population Registry System missed 89 cases (19%), the majority being infants and elderly people’s deaths. The FilaBavi, re-census survey missed 19 deaths (4%), and the neighbourhood survey, where groups of local people listed deaths, over-reported (paper I). Community members regarded ‘elderly deaths’ as ‘natural’ and ‘deserved’, and consequently such deaths might not be notified. Infants were not considered as ‘fully human’, therefore infant deaths were not necessarily reported. ‘Young deaths’ were seen as either ‘good deaths’ or ‘bad deaths’. The latter, considered due to stigmatised diseases (e.g. HIV/AIDS, tuberculosis and leprosy) and suicide, were often concealed by the family. Other reasons for not reporting deaths were poor knowledge of the legal obligation to report deaths, lack of incentive to report and the absence of legal sanction in case of not reporting (paper II). Registrars also identified the inadequate functioning of administrative structures with respect to supervision, training and collaboration between systems as a reason for the low quality of death data in the routine system (paper III). There was a mutual reluctance between the surveyors at FilaBavi, for conducting death inquiries, and bereaved respondents, for receiving surveyors for such inquiries, especially in the case of child or untimely deaths, but both parties felt a sense of duty to do so. The choice of visiting time, the venue and the respondent, as well as the need for creating good rapport with the family were considered as important to facilitate the death inquiry and to minimise distress among the bereaved (paper IV). Conclusions: Mortality estimates based on information from routine systems at the commune level have low validity. FilaBavi performed better than any of these official systems. Several reasons for poor reporting and registration, which were possible to improve, were identified. Further, minimising discomfort and compassion stress is an important ethical issue in death inquiries. The experiences of death inquiry from the FilaBavi can also be useful in other DSS with similarities in cultural and administrative context