Verbal autopsy completion rate and factors associated with undetermined cause of death in a rural resource-poor setting of Tanzania

Exploring challenges and recommendations for verbal autopsy implementation in low-/middle-income countries: a cross-sectional study of Iringa Region-Tanzania

BACKGROUND

Verbal autopsy (VA) plays a vital role in providing cause-of-death information in places where such information is not available. Many low-/middle-income countries (LMICs) including Tanzania are still struggling to yield quality and adequate cause-of-death data for Civil Registration and Vital Statistics (CRVS).

OBJECTIVE

To highlight challenges and recommendations for VA implementation to support LMICs yield quality and adequate mortality statistics for informed decisions on healthcare interventions.

DESIGN

Cross-sectional study.

STUDY SETTING

Iringa region in Tanzania.

PARTICIPANTS

41 people including 33 community health workers, 1 VA national coordinator, 5 national task force members, 1 VA regional coordinator and 1 member of the VA data management team.

RESULTS

The perceived challenges of key informants include a weak death notification system, lengthy VA questionnaire, poor data quality and inconsistent responses, lack of clarity in the inclusion criteria, poor commitment to roles and responsibilities, poor coordination, poor financial mechanism and no or delayed feedback to VA implementers. Based on these findings, we recommend the following strategies for effective adaptation and use of VAs: (1) reinforce or implement legislative procedures towards the legal requirement for death notification. (2) Engage key stakeholders in the overall implementation of VAs. (3) Build capacity for data collection, monitoring, processing and use of VA data. (4) Improve the VA questionnaire and quality control mechanism for optimal use in data collection. (5) Create sustainable financing mechanisms and institutionalisation of VA implementation. (6) Integrating VA Implementation in CRVS.

CONCLUSION

Effective VA implementation demands through planning, stakeholder engagement, upskilling of local experts and fair compensation for interviewers. Such coordinated endeavours will overcome systemic, technical and behavioural challenges hindering VA's successful implementation.

Levels and trends in mortality and causes of death among women of reproductive age in Bangladesh: Findings from three national surveys

Background

Information on the mortality rate and proportional cause-specific mortality is essential for identifying diseases of public health importance, design programmes, and formulating policies, but such data on women of reproductive age in Bangladesh is limited.

Methods

We analysed secondary data from the 2001, 2010, and 2016 rounds of the nationally representative Bangladesh Maternal Mortality and Health Care Survey (BMMS) to estimate mortality rates and causes of death among women aged 15-49 years. We collected information on causes of death three years prior to each survey using a country-adapted version of the World Health Organization (WHO) verbal autopsy (VA) questionnaire. Trained physicians independently reviewed the VA questionnaire and assigned a cause of death using the International Classification of Diseases (ICD) codes. The analysis included mortality rates and proportional mortality showing overall and age-specific causes of death.

Results

The overall mortality rates for women aged 15-49 years decreased over time, from 190 per 100 000 years of observation in the 2001 BMMS, to 121 per 100 000 in the 2010 BMMS, to 116 per 100 000 in the 2016 BMMS. Age-specific mortality showed a similar downward pattern. The three diseases contributing the most to mortality were maternal causes (13-20%), circulatory system diseases (15-23%), and malignancy (14-24%). The relative position of these three diseases changed between the three surveys. From the 2001 BMMS to the 2010 BMMS and subsequently to the 2016 BMMS, the number of deaths from non-communicable diseases (e.g. cardiovascular diseases and malignancies) increased from 29% to 38% to 48%. Maternal causes led to the highest proportion of deaths among 20-34-year-olds in all three surveys (25-32%), while suicide was the number one cause of death for teenagers (19-22%). Circulatory system diseases and malignancy were the two leading causes of death for older women aged 35-49 years (40%-67%).

Conclusions

There was a gradual shift in the causes of death from communicable to non-communicable diseases among women of reproductive age in Bangladesh. Suicide as the primary cause of death among teenage girls demands urgent attention for prevention.

Performance evaluation of machine learning and Computer Coded Verbal Autopsy (CCVA) algorithms for cause of death determination: A comparative analysis of data from rural South Africa

Computer Coded Verbal Autopsy (CCVA) algorithms are commonly used to determine the cause of death (CoD) from questionnaire responses extracted from verbal autopsies (VAs). However, they can only operate on structured data and cannot effectively harness information from unstructured VA narratives. Machine Learning (ML) algorithms have also been applied successfully in determining the CoD from VA narratives, allowing the use of auxiliary information that CCVA algorithms cannot directly utilize. However, most ML-based studies only use responses from the structured questionnaire, and the results lack generalisability and comparability across studies. We present a comparative performance evaluation of ML methods and CCVA algorithms on South African VA narratives data, using data from Agincourt Health and Demographic Surveillance Site (HDSS) with physicians' classifications as the gold standard. The data were collected from 1993 to 2015 and have 16,338 cases. The random forest and extreme gradient boosting classifiers outperformed the other classifiers on the combined dataset, attaining accuracy of 96% respectively, with significant statistical differences in algorithmic performance (p < 0.0001). All our models attained Area Under Receiver Operating Characteristics (AUROC) of greater than 0.884. The InterVA CCVA attained 83% Cause Specific Mortality Fraction accuracy and an Overall Chance-Corrected Concordance of 0.36. We demonstrate that ML models could accurately determine the cause of death from VA narratives. Additionally, through mortality trends and pattern analysis, we discovered that in the first decade of the civil registration system in South Africa, the average life expectancy was approximately 50 years. However, in the second decade, life expectancy significantly dropped, and the population was dying at a much younger average age of 40 years, mostly from the leading HIV related causes. Interestingly, in the third decade, we see a gradual improvement in life expectancy, possibly attributed to effective health intervention programmes. Through a structure and semantic analysis of narratives where experts disagree, we also demonstrate the most frequent terms of traditional healer consultations and visits. The comparative approach also makes this study a baseline that can be used for future research enforcing generalization and comparability. Future study will entail exploring deep learning models for CoD classification.

Reporting of ethics in peer-reviewed verbal autopsy studies: a systematic review

Introduction

Verbal autopsy (VA) is a method that determines the cause of death by interviewing a relative of the deceased about the events occurring before the death, in regions where medical certification of cause of death is incomplete. This paper aims to review the ethical standards reported in peer-reviewed VA studies.

Methods

A systematic review of Medline and Ovid was conducted by two independent researchers. Data were extracted and analysed for articles based on three key areas: Institutional Review Board (IRB) clearance and consenting process; data collection and management procedures, including: time between death and interview; training and education of interviewer, confidentiality of data and data security; and declarations of funding and conflict of interest.

Results

The review identified 802 articles, of which 288 were included. The review found that 48% all the studies reported having IRB clearance or obtaining consent of participants. The interviewer training and education levels were reported in 62% and 21% of the articles, respectively. Confidentiality of data was reported for 14% of all studies, 18% did not report the type of respondent interviewed and 51% reported time between death and the interview for the VA. Data security was reported in 8% of all studies. Funding was declared in 63% of all studies and conflict of interest in 42%. Reporting of all these variables increased over time.

Conclusions

The results of this systematic review show that although there has been an increase in ethical reporting for VA studies, there still remains a large gap in reporting.

Main causes of death in Dande, Angola: results from Verbal Autopsies of deaths occurring during 2009-2012

BACKGROUND

The Dande Health and Demographic Surveillance System (HDSS) located in Bengo Province, Angola, covers nearly 65,500 residents living in approximately 19,800 households. This study aims to describe the main causes of deaths (CoD) occurred within the HDSS, from 2009 to 2012, and to explore associations between demographic or socioeconomic factors and broad mortality groups (Group I-Communicable diseases, maternal, perinatal and nutritional conditions; Group II-Non-communicable diseases; Group III-Injuries; IND-Indeterminate).

METHODS

Verbal Autopsies (VA) were performed after death identification during routine HDSS visits. Associations between broad groups of CoD and sex, age, education, socioeconomic position, place of residence and place of death, were explored using chi-square tests and fitting logistic regression models.

RESULTS

From a total of 1488 deaths registered, 1009 verbal autopsies were performed and 798 of these were assigned a CoD based on the 10(th) revision of the International Classification of Diseases (ICD-10). Mortality was led by CD (61.0 %), followed by IND (18.3 %), NCD (11.6 %) and INJ (9.1 %). Intestinal infectious diseases, malnutrition and acute respiratory infections were the main contributors to under-five mortality (44.2 %). Malaria was the most common CoD among children under 15 years old (38.6 %). Tuberculosis, traffic accidents and malaria led the CoD among adults aged 15-49 (13.5 %, 10.5 % and 8.0 % respectively). Among adults aged 50 or more, diseases of the circulatory system (23.2 %) were the major CoD, followed by tuberculosis (8.2 %) and malaria (7.7 %). CD were more frequent CoD among less educated people (adjusted odds ratio, 95 % confidence interval for none vs. 5 or more years of school: 1.68, 1.04-2.72).

CONCLUSION

Infectious diseases were the leading CoD in this region. Verbal autopsies proved useful to identify the main CoD, being an important tool in settings where vital statistics are scarce and death registration systems have limitations.

Multiple-bias analysis as a technique to address systematic error in measures of abortion-related mortality

Background

The UN Millennium Development Goals (MDGs) and Sustainable Development Goals (SDGs) have brought heightened global attention to the measurement of maternal mortality. It is imperative that new and novel approaches be used to measure maternal mortality and to better understand existing data. In this paper we present one approach: an epidemiologic framework for identifying the identification and quantification of systematic error (multiple-bias analysis), outline the necessary steps for investigators interested in conducting multiple-bias analyses in their own data, and suggest approaches for reporting such analyses in the literature.

Methods

To conceptualize the systematic error present in studies of abortion-related deaths, we propose a bias framework. We posit that selection bias and misclassification are present in both verbal autopsy studies and facility-based studies. The multiple-bias analysis framework provides a relatively simple, quantitative strategy for assessing systematic error and resulting bias in any epidemiologic study.

Results

In our worked example of multiple-bias analysis on a study reporting 20.6 % of maternal deaths to be abortion related, after adjustment for selection bias, misclassification, and random error, the median increased, on average, to 0.308, approximately 20 % greater than the reported proportion of abortion-related deaths.

Conclusions

Reporting results of multiple-bias analyses in estimates of abortion-related mortality, predictors of unsafe abortion, and other reproductive health questions that suffer from similar biases would not only improve reporting practices in the field, but might also provide a more accurate understanding of the range of potential impact of policies and programs that target the underlying causes of unsafe abortion and abortion-related mortality.

Electronic supplementary material

The online version of this article (doi:10.1186/s12963-016-0075-3) contains supplementary material, which is available to authorized users.

Seasonal and temporal trends in all-cause and malaria mortality in rural Burkina Faso, 1998-2007

BACKGROUND

High mortality levels in sub-Saharan Africa are still a major public health problem. Children are the most affected group with malaria as one of the major causes of death in this region. To plan health interventions, reliable empirical information on cause-specific mortality patterns is essential, yet such data are often not available in developing countries. Health and Demographic Surveillance Systems (HDSS) implementing the verbal autopsy (VA) method provide such data on a longitudinal basis. Physician Coded VA is usually used to determine cause of death, but recently a computerized method, Interpreting VA (InterVA) was alternatively introduced. This study investigates the effect of season on all-cause and malaria mortality analysing cause of death data from 1998 to 2007 obtained by the Nouna HDSS in rural Burkina Faso and derived by InterVA.

METHODS

Monthly mortality rates were calculated for different age groups (infants, children, adolescents, adults, elderly). Seasonal and temporal trends were modelled with parametric Poisson regression adjusted for sex, area of residence and year of death.

RESULTS

Overall, 7,378 deaths occurred corresponding to a mortality rate of 11.9/1,000 with highest rates in infants (56.8/1,000) and children (22.0/1,000). Young children were most affected by malaria. Malaria mortality patterns in children showed significantly higher rates during the rainy season and a stagnant long-term trend. The seasonal trend is well described parametrically with a sinusoidal function. InterVA assigned about half as many deaths to malaria than physicians.

CONCLUSIONS

Malaria mortality remains highly seasonal in rural Burkina Faso. The InterVA method appears to determine reasonably well seasonal mortality patterns, which should be considered for the planning of health resources and activities.

Trends in Causes of Adult Deaths among the Urban Poor: Evidence from Nairobi Urban Health and Demographic Surveillance System, 2003-2012

What kills people around the world and how it varies from place to place and over time is critical in mapping the global burden of disease and therefore, a relevant public health question, especially in developing countries. While more than two thirds of deaths worldwide are in developing countries, little is known about the causes of death in these nations. In many instances, vital registration systems are nonexistent or at best rudimentary, and even when deaths are registered, data on the cause of death in particular local contexts, which is an important step toward improving context-specific public health, are lacking. In this paper, we examine the trends in the causes of death among the urban poor in two informal settlements in Nairobi by applying the InterVA-4 software to verbal autopsy data. We examine cause of death data from 2646 verbal autopsies of deaths that occurred in the Nairobi Urban Health and Demographic Surveillance System (NUHDSS) between 1 January 2003 and 31 December 2012 among residents aged 15 years and above. The data is entered into the InterVA-4 computer program, which assigns cause of death using probabilistic modeling. The results are presented as annualized trends from 2003 to 2012 and disaggregated by gender and age. Over the 10-year period, the three major causes of death are tuberculosis (TB), injuries, and HIV/AIDS, accounting for 26.9, 20.9, and 17.3% of all deaths, respectively. In 2003, HIV/AIDS was the highest cause of death followed by TB and then injuries. However, by 2012, TB and injuries had overtaken HIV/AIDS as the major causes of death. When this is examined by gender, HIV/AIDS was consistently higher for women than men across all the years generally by a ratio of 2 to 1. In terms of TB, it was more evenly distributed across the years for both males and females. We find that there is significant gender variation in deaths linked to injuries, with male deaths being higher than female deaths by a ratio of about 4 to 1. We also find a fifteen percentage point increase in the incidences of male deaths due to injuries between 2003 and 2012. For women, the corresponding deaths due to injuries remain fairly stable throughout the period. We find cardiovascular diseases as a significant cause of death over the period, with overall mortality increasing steadily from 1.6% in 2003 to 8.1% in 2012, and peaking at 13.7% in 2005 and at 12.0% in 2009. These deaths were consistently higher among women. We identified substantial variations in causes of death by age, with TB, HIV/AIDS, and CVD deaths lowest among younger residents and increasing with age, while injury-related deaths are highest among the youngest adults 15-19 and steadily declined with age. Also, deaths related to neoplasms and respiratory tract infections (RTIs) were prominent among older adults 50 years and above, especially since 2005. Emerging at this stage is evidence that HIV/AIDS, TB, injuries, and cardiovascular disease are linked to approximately 73% of all adult deaths among the urban poor in Nairobi slums of Korogocho and Viwandani in the last 10 years. While mortality related to HIV/AIDS is generally declining, we see an increasing proportion of deaths due to TB, injuries, and cardiovascular diseases. In sum, substantial epidemiological transition is ongoing in this local context, with deaths linked to communicable diseases declining from 66% in 2003 to 53% in 2012, while deaths due to noncommunicable causes experienced a four-fold increase from 5% in 2003 to 21.3% in 2012, together with another two-fold increase in deaths due to external causes (injuries) from 11% in 2003 to 22% in 2012. It is important to also underscore the gender dimensions of the epidemiological transition clearly visible in the mix. Finally, the elevated levels of disadvantage of slum dwellers in our analysis relative to other population subgroups in Kenya continue to demonstrate appreciable deterioration of key urban health and social indicators, highlighting the need for a deliberate strategic focus on the health needs of the urban poor in policy and program efforts toward achieving international goals and national health and development targets.

Health & Demographic Surveillance System Profile: The Ifakara Rural and Urban Health and Demographic Surveillance System (Ifakara HDSS)

The Ifakara Rural HDSS (125,000 people) was set up in 1996 for a trial of the effectiveness of social marketing of bed nets on morbidity and mortality of children aged under 5 years, whereas the Ifakara Urban HDSS (45,000 people) since 2007 has provided demographic indicators for a typical small urban centre setting. Jointly they form the Ifakara HDSS (IHDSS), located in the Kilombero valley in south-east Tanzania. Socio-demographic data are collected twice a year. Current malaria work focuses on phase IV studies for antimalarials and on determinants of fine-scale variation of pathogen transmission risk, to inform malaria elimination strategies. The IHDSS is also used to describe the epidemiology and health system aspects of maternal, neonatal and child health and for intervention trials at individual and health systems levels. More recently, IHDSS researchers have studied epidemiology, health-seeking and national programme effectiveness for chronic health problems of adults and older people, including for HIV, tuberculosis and non-communicable diseases. A focus on understanding vulnerability and designing methods to enhance equity in access to services are cross-cutting themes in our work. Unrestricted access to core IHDSS data is in preparation, through INDEPTH iSHARE [www.indepth-ishare.org] and the IHI data portal [http://data.ihi.or.tz/index.php/catalog/central].

Trends in non-communicable disease mortality among adult residents in Nairobi's slums, 2003-2011: applying InterVA-4 to verbal autopsy data

BACKGROUND

About 80% of deaths from non-communicable diseases (NCDs) occur in developing countries such as Kenya. However, not much is known about the burden of NCDs in slums, which account for about 60% of the residences of the urban population in Kenya. This study examines trends in NCD mortality from two slum settings in Nairobi.

DESIGN

We use verbal autopsy data on 1954 deaths among adults aged 35 years and older who were registered in the Nairobi Urban Health and Demographic Surveillance System between 2003 and 2011. InterVA-4, a computer-based program, was used to assign causes of death for each case. RESULTS are presented as annualized cause-specific mortality rates (CSMRs) and cause-specific mortality fractions (CSMFs) by sex.

RESULTS

The CSMRs for NCDs did not appear to change significantly over time for both males and females. Among males, cardiovascular diseases (CVDs) and neoplasms were the leading NCDs--contributing CSMFs of 8 and 5%, respectively, on average over time. Among females, CVDs contributed a CSMF of 14% on average over time, while neoplasms contributed 8%. Communicable diseases and related conditions remained the leading causes of death, contributing a CSMF of over 50% on average in males and females over time.

CONCLUSIONS

Our findings are consistent with the Global Burden of Disease 2010 study which shows that communicable diseases remain the dominant cause of death in Africa, although NCDs were still significant contributors to mortality. We recommend an integrated approach towards disease prevention that focuses on health systems strengthening in resource-limited settings such as slums.

The contribution of reduction in malaria as a cause of rapid decline of under-five mortality: evidence from the Rufiji Health and Demographic Surveillance System (HDSS) in rural Tanzania

Background

Under-five mortality has been declining rapidly in a number of sub-Saharan African settings. Malaria-related mortality is known to be a major component of childhood causes of death and malaria remains a major focus of health interventions. The paper explored the contribution of malaria relative to other specific causes of under-five deaths to these trends.

Methods

This paper uses longitudinal demographic surveillance data to examine trends and causes of death of under-five mortality in Rufiji, whose population has been followed for over nine years (1999–2007). Causes of death, determined by the verbal autopsy technique, are analysed with Arriaga’s decomposition method to assess the contribution of declining malaria-related mortality relative to other causes of death as explaining a rapid decline in overall childhood mortality.

Results

Over the 1999–2007 period, under-five mortality rate in Rufiji declined by 54.3%, from 33.3 to 15.2 per 1,000 person-years. If this trend is sustained, Rufiji will be a locality that achieves MDG4 target. Although hypotrophy at birth remained the leading cause of death for neonates, malaria remains as the leading cause of death for post-neonates followed by pneumonia. However, declines in malaria death rates accounted for 49.9% of the observed under-five mortality decline while all perinatal causes accounted for only 19.9%.

Conclusion

To achieve MDG 4 in malaria endemic settings, health programmes should continue efforts to reduce malaria mortality and more efforts are also needed to improve newborn survival.

InterVA versus Spectrum: how comparable are they in estimating AIDS mortality patterns in Nairobi's informal settlements?

Background

The Spectrum computer package is used to generate national AIDS mortality estimates in settings where vital registration systems are lacking. Similarly, InterVA-4 (the latest version of the InterVA programme) is used to estimate cause-of-mortality data in countries where cause-specific mortality data are not available.

Objective

This study aims to compare trends in adult AIDS-related mortality estimated by Spectrum with trends from the InterVA-4 programme applied to data from a Health and Demographic Surveillance System (HDSS) in Nairobi, Kenya.

Design

A Spectrum model was generated for the city of Nairobi based on HIV prevalence data for Nairobi and national antiretroviral therapy coverage, underlying mortality, and migration assumptions. We then used data, generated through verbal autopsies, on 1,799 deaths that occurred in the HDSS area from 2003 to 2010 among adults aged 15–59. These data were then entered into InterVA-4 to estimate causes of death using probabilistic modelling. Estimates of AIDS-related mortality rates and all-cause mortality rates from Spectrum and InterVA-4 were compared and presented as annualised trends.

Results

Spectrum estimated that HIV prevalence in Nairobi was 7%, while the HDSS site measured 12% in 2010. Despite this difference, Spectrum estimated higher levels of AIDS-related mortality. Between 2003 and 2010, the proportion of AIDS-related mortality in Nairobi decreased from 63 to 40% according to Spectrum and from 25 to 16% according to InterVA. The net AIDS-related mortality in Spectrum was closer to the combined mortality rates when AIDS and tuberculosis (TB) deaths were included for InterVA-4.

Conclusion

Overall trends in AIDS-related deaths from both methods were similar, although the values were closer when TB deaths were included in InterVA. InterVA-4 might not accurately differentiate between TB and AIDS deaths.

An innovative approach to measuring maternal mortality at the community level in low-resource settings using mid-level providers: a feasibility study in Tigray, Ethiopia

This paper proposes a new, community-based approach to the measurement of maternal mortality, and presents results from a feasibility study in 2010-11 of that approach in rural Tigray, Ethiopia. The study was implemented in three health posts and one health centre with a total catchment area of approximately 22,000 people. Priests, traditional birth attendants and community-based reproductive health agents were responsible for locating and reporting all births and deaths in their areas and assisted mid-level providers in locating key informants for verbal autopsy. Community-based health workers were trained to report all births and deaths to the local health post, where vital registries were kept. Once a month, each health post compiled a list of all deaths of women aged 12-49, which were registered in government logbooks. Nurses and nurse-midwives were trained to administer verbal autopsies on these deaths, and assign primary cause of death using WHO ICD-10 classifications. The study drew on the theory of task-shifting, shifting the task of cause-of-death attribution from physicians to mid-level providers. It aimed to build a sustainable methodology for maximizing existing local health care infrastructure and human capacity, leading to community-based solutions to improve maternal health. While the approach has not yet been implemented outside the initial study area, the results are promising as regards its feasibility.