Comparing verbal autopsy cause of death findings as determined by physician coding and probabilistic modelling: a public health analysis of 54 000 deaths in Africa and Asia

A Qualitative Method To Assess a History of Cerebral Malaria in Malian Children

The investigation of factors associated with susceptibility to severe malaria is best achieved using case-control studies. The presence of a history of severe malaria in controls could affect the quality of their phenotype and study findings and hence should be rigorously determined. Here, we assessed the performance of a qualitative questionnaire to identify a history of cerebral malaria in controls in a case-control study of severe malaria in Mali. We evaluated the archived medical records of 220 children diagnosed with severe diseases at health care centers in rural and urban settings in Mali from 2018 to 2019. Parents of enrolled children were then identified and interviewed using a structured questionnaire by an investigator blinded to the diagnosis. The diagnosis derived from the interview was then compared with the diagnosis from the medical records as the reference diagnosis. The sensitivity and specificity of the questionnaire to detect cerebral malaria in history were, respectively, 84% and 76%. The questionnaire was concordant with the medical record diagnosis in 60% (95% CI: 50-71%) of cases. For other clinical phenotypes of malaria (severe malaria anemia, uncomplicated malaria, and severe malaria anemia concurrent with cerebral malaria), sensitivity and specificity ranged from 42% to 85% and 88% to 96%, respectively. Positive and negative predictive values were, respectively, 75% and 85%. The questionnaire demonstrated suitable sensitivity and specificity to identify cerebral malaria in a participant's history. In sub-Saharan Africa, a region with suboptimal medical record archives, such a tool could be used in case-control studies of severe malaria to select controls.

Can verbal autopsies be used on a national scale? Key findings and lessons from South Africa's national cause-of-death validation study

BACKGROUND

Verbal autopsy (VA), though imperfect, serves as a vital tool to determine cause-of-death, particularly for out-of-facility deaths, but challenges persist in integrating VA into Civil Registration and Vital Statistics systems.

OBJECTIVE

To describe the challenges and successes of collecting a national sample of verbal autopsy interviews in South Africa to obtain the cause of death profile in 2017/18.

METHODS

We recruited next of kin from 27 randomly selected sub-districts (10.5%) across South Africa between September 2017 and April 2018. Trained fieldworkers conducted face-to-face interviews using the WHO2016 VA instrument, with physicians certifying underlying causes of death. Feasibility was evaluated based on response rates, participation, and data quality.

RESULTS

Of the total 36,976 deaths registered, only 26% were identified during recruitment, with a 55% overall response rate for VA interviews. Physician-reviewed VA data were deemed of good quality for assigning underlying causes of death in 83% of cases. By comparing cause-specific mortality fractions, physician-reviewed VA identified 22.3% HIV/AIDS and InterVA-5 identified 18.5%, aligning with burden of disease estimates, while Statistics South Africa reported 4.9% HIV/AIDS.

CONCLUSIONS

The study demonstrated the feasibility of using VA on a national scale, but immense challenges in identifying and recruiting next of kin highlight the importance of formalising VAs within the country's death notification system.

Validating the diagnostic accuracy of medical certification: a population-level comparison between verbal autopsy and Saudi medical records causes of death of deceased with type 2 diabetes

BACKGROUND

In contexts where certifying causes of death (COD) is inadequate - either in industrialized or non-industrialized countries - verbal autopsy (VA) serves as a practical method for determining probable COD, helping to address gaps in vital data.

OBJECTIVE

This study aimed to validate the diagnostic accuracy of medical certifications at a population level by comparing COD obtained from medical records against those derived from VA in Saudi Arabia.

METHOD

Death records from 2018 to 2021 were collected from a type 2 diabetes mellitus register at a major specialist hospital in Makkah. Three hundred and two VA interviews were completed with deceased patients' relatives, and the probable COD was determined using InterVA-5 software. Lin's concordance correlation coefficient was applied to examine similarities of the cause-specific mortality fractions (CSMFs) based on International Classification of Diseases chapters from both verbal autopsy causes of death (VACOD) and the physician review causes of death (PRCOD).

RESULTS

Overall, the findings demonstrated a moderate level of concordance of COD at the population between VACOD and PRCOD. However, the CSMFs for various COD categories derived from both sources showed a broad spectrum of absolute differences, with the largest disparities observed among the most prevalent COD categories.

CONCLUSION

PRCOD was found to overestimate population-level endocrine/metabolic and respiratory disease COD while underestimating circulatory disease, demonstrating medical certification challenges. Conversely, affirming previous literature on prevalent COD in Saudi Arabia, VA appears to deliver a plausible assessment, further strengthening its potential to integrate within the Saudi health system towards an augmented medical certification process.

Public health determinants of child malaria mortality: a surveillance study within Siaya County, Western Kenya

BACKGROUND

Malaria deaths among children have been declining worldwide during the last two decades. Despite preventive, epidemiologic and therapy-development work, mortality rate decline has stagnated in western Kenya resulting in persistently high child malaria morbidity and mortality. The aim of this study was to identify public health determinants influencing the high burden of malaria deaths among children in this region.

METHODS

A total of 221,929 children, 111,488 females and 110,441 males, under the age of 5 years were enrolled in the Kenya Medical Research Institute/Center for Disease Control Health and Demographic Surveillance System (KEMRI/CDC HDSS) study area in Siaya County during the period 2003-2013. Cause of death was determined by use of verbal autopsy. Age-specific mortality rates were computed, and cox proportional hazard regression was used to model time to malaria death controlling for the socio-demographic factors. A variety of demographic, social and epidemiologic factors were examined.

RESULTS

In total 8,696 (3.9%) children died during the study period. Malaria was the most prevalent cause of death and constituted 33.2% of all causes of death, followed by acute respiratory infections (26.7%) and HIV/AIDS related deaths (18.6%). There was a marked decrease in overall mortality rate from 2003 to 2013, except for a spike in the rates in 2008. The hazard of death differed between age groups with the youngest having the highest hazard of death HR 6.07 (95% CI 5.10-7.22). Overall, the risk attenuated with age and mortality risks were limited beyond 4 years of age. Longer distance to healthcare HR of 1.44 (95% CI 1.29-1.60), l ow maternal education HR 3.91 (95% CI 1.86-8.22), and low socioeconomic status HR 1.44 (95% CI 1.26-1.64) were all significantly associated with increased hazard of malaria death among children.

CONCLUSIONS

While child mortality due to malaria in the study area in Western Kenya, has been decreasing, a final step toward significant risk reduction is yet to be accomplished. This study highlights residual proximal determinants of risk which can further inform preventive actions.

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.

Data innovation in response to COVID-19 in Somalia: application of a syndromic case definition and rapid mortality assessment method

Background

During the COVID-19 pandemic, the importance of reliable public health data has been highlighted, as well as the multiple challenges in collecting it, especially in low income and conflict-affected countries. Somalia reported its first confirmed case of COVID-19 on 16 March 2020 and has experienced fluctuating infection levels since then.

Objectives

To monitor the impact of COVID-19 on beneficiaries of a long-term cash transfer programme in Somalia and assess the utility of a syndromic score case definition and rapid mortality surveillance tool.

Methods

Five rounds of telephone interviews were conducted from June 2020 – April 2021 with 1,046–1,565 households participating in a cash transfer programme. The incidence of COVID-19 symptoms and all-cause mortality were recorded. Carers of the deceased were interviewed a second time using a rapid verbal autopsy questionnaire to determine symptoms preceding death. Data were recorded on mobile devices and analysed using COVID Rapid Mortality Surveillance (CRMS) software and R.

Results

The syndromic score case definition identified suspected symptomatic cases that were initially confined to urban areas but then spread widely throughout Somalia. During the first wave, the peak syndromic case rate (311 cases/million people/day) was 159 times higher than the average laboratory confirmed case rate reported by WHO for the same period. Suspected COVID-19 deaths peaked at 14.3 deaths/million people/day, several weeks after the syndromic case rate. Crude and under-five death rates did not cross the respective emergency humanitarian thresholds (1 and 2 deaths/10,000 people/day).

Conclusion

Use of telephone interviews to collect data on the evolution of COVID-19 outbreaks is a useful additional approach that can complement laboratory testing and mortality data from the health system. Further work to validate the syndromic score case definition and CRMS is justified.

Mortality transition over a quarter century in rural South Africa: findings from population surveillance in Agincourt 1993-2018

Background

Mortality burden in South Africa since the mid-1990s has been characterized by a quadruple disease burden: HIV/AIDS and tuberculosis (TB); other communicable diseases (excluding HIV/AIDS and TB), maternal causes, perinatal conditions and nutritional deficiencies; non-communicable diseases (NCDs); and injuries. Causes from these broad groupings have persistently constituted the top 10 causes of death. However, proportions and rankings have varied over time, alongside overall mortality levels.

Objective

To provide evidence on the contributions of age and cause-of-death to changes in mortality levels in a rural South African population over a quarter century (1993–2018).

Methods

Using mortality and cause-of-death data from the Agincourt Health and Socio-Demographic Surveillance System (HDSS), we derive estimates of the distribution of deaths by cause, and hazards of death by age, sex, and time period, 1993–2018. We derive estimates of life expectancies at birth and years of life expectancy gained at age 15 if most common causes of death were deleted. We compare mortality indicators and cause-of-death trends from the Agincourt HDSS with South African national indicators generated from publicly available datasets.

Results

Mortality and cause-of-death transition reveals that overall mortality levels have returned to pre-HIV epidemic levels. In recent years, the concentration of mortality has shifted towards older ages, and the mortality burden from cardiovascular diseases and other chronic NCDs are more prominent as people living with HIV/AIDS access ART and live longer. Changes in life expectancy at birth, distribution of deaths by age, and major cause-of-death categories in the Agincourt population follow a similar pattern to the South African population.

Conclusion

The Agincourt HDSS provides critical information about general mortality, cause-of-death, and age patterns in rural South Africa. Realigning and strengthening the South African public health and healthcare systems is needed to concurrently cater for the prevention, control, and treatment of multiple disease conditions.

Estimating causes of death where there is no medical certification: evolution and state of the art of verbal autopsy

Over the past 70 years, significant advances have been made in determining the causes of death in populations not served by official medical certification of cause at the time of death using a technique known as Verbal Autopsy (VA). VA involves an interview of the family or caregivers of the deceased after a suitable bereavement interval about the circumstances, signs and symptoms of the deceased in the period leading to death. The VA interview data are then interpreted by physicians or, more recently, computer algorithms, to assign a probable cause of death. VA was originally developed and applied in field research settings. This paper traces the evolution of VA methods with special emphasis on the World Health Organization's (WHO)'s efforts to standardize VA instruments and methods for expanded use in routine health information and vital statistics systems in low- and middle-income countries (LMICs). These advances in VA methods are culminating this year with the release of the 2022 WHO Standard Verbal Autopsy (VA) Toolkit. This paper highlights the many contributions the late Professor Peter Byass made to the current VA standards and methods, most notably, the development of InterVA, the most commonly used automated computer algorithm for interpreting data collected in the WHO standard instruments, and the capacity building in low- and middle-income countries (LMICs) that he promoted. This paper also provides an overview of the methods used to improve the current WHO VA standards, a catalogue of the changes and improvements in the instruments, and a mapping of current applications of the WHO VA standard approach in LMICs. It also provides access to tools and guidance needed for VA implementation in Civil Registration and Vital Statistics Systems at scale.

Causes and circumstances of maternal death: a secondary analysis of the Community-Level Interventions for Pre-eclampsia (CLIP) trials cohort

BACKGROUND

Incomplete vital registration systems mean that causes of death during pregnancy and childbirth are poorly understood in low-income and middle-income countries. To inform global efforts to reduce maternal mortality, we compared physician review and computerised analysis of verbal autopsies (interpreting verbal autopsies [InterVA] software), to understand their agreement on maternal cause of death and circumstances of mortality categories (COMCATs) in the Community-Level Interventions for Pre-eclampsia (CLIP) cluster randomised trials.

METHODS

The CLIP trials took place in India, Pakistan, and Mozambique, enrolling pregnant women aged 12-49 years between Nov 1, 2014, and Feb 28, 2017. 69 330 pregnant women were enrolled in 44 clusters (36 008 in the 22 intervention clusters and 33 322 in the 22 control clusters). In this secondary analysis of maternal deaths in CLIP, we included women who died in any of the 22 intervention clusters or 22 control clusters. Trained staff administered the WHO 2012 verbal autopsy after maternal deaths. Two physicians (and a third for consensus, if needed) reviewed trial surveillance data and verbal autopsies, and, in intervention clusters, community health worker-led visit data. They determined cause of death according to the WHO International Classification of Diseases-Maternal Mortality (ICD-MM). Verbal autopsies were also analysed by InterVA computer models (versions 4 and 5) to generate cause of death. COMCAT analysis was provided by InterVA-5 and, in India, by physician review of Maternal Newborn Health Registry data. Causes of death and COMCATs assigned by physician review, Inter-VA-4, and InterVA-5 were compared, with agreement assessed with Cohen's κ coefficient.

FINDINGS

Of 61 988 pregnancies with successful follow-up in the CLIP trials, 143 maternal deaths were reported (16 deaths in India, 105 in Pakistan, and 22 in Mozambique). The maternal death rate was 231 (95% CI 193-268) per 100 000 identified pregnancies. Most deaths were attributed to direct maternal causes (rather than indirect or undetermined causes as per ICD-MM classification), with fair to good agreement between physician review and InterVA-4 (κ=0·56 [95% CI 0·43-0·66]) or InterVA-5 (κ=0·44 [0·30-0·57]), and InterVA-4 and InterVA-5 (κ=0·72 [0·60-0·84]). The top three causes of death were the same by physician review, InterVA-4, and InterVA-5 (ICD-MM categories obstetric haemorrhage, non-obstetric complications, and hypertensive disorders); however, attribution of individual patient deaths to obstetric haemorrhage varied more between methods (physician review, 38 [27%] deaths; InterVA-4, 69 [48%] deaths; and InterVA-5, 82 [57%] deaths), than did attribution to non-obstetric causes (physician review, 39 [27%] deaths; InterVA-4, 37 [26%] deaths; and InterVA-5, 28 [20%] deaths) or hypertensive disorders (physician review, 23 [16%] deaths; InterVA-4, 25 [17%] deaths; and InterVA-5, 24 [17%] deaths). Agreement for all nine ICD-MM categories was fair for physician review versus InterVA-4 (κ=0·48 [0·38-0·58]), poor for physician review versus InterVA-5 (κ=0·36 [0·27-0·46]), and good for InterVA-4 versus InterVA-5 (κ=0·69 [0·59-0·79]). The most commonly assigned COMCATs by InterVA-5 were emergencies (68 [48%] of 143 deaths) and health systems (62 [43%] deaths), and by physician review (India only) were health systems (seven [44%] of 16 deaths) and inevitability (five [31%] deaths); agreement between InterVA-5 and physician review (India data only) was poor (κ=0·04 [0·00-0·15]).

INTERPRETATION

Our findings indicate that InterVA-5 is less accurate than InterVA-4 at ascertaining causes and circumstances of maternal death, when compared with physician review. Our results suggest a need to improve the next iteration of InterVA, and for researchers and clinicians to preferentially use InterVA-4 when recording maternal deaths.

FUNDING

University of British Columbia (grantee of the Bill & Melinda Gates Foundation).

Ultrasound in legal medicine-a missed opportunity or simply too late? A narrative review of ultrasonic applications in forensic contexts

Objectives

Conventional autopsies remain the gold standard of postmortem healthcare quality assurance and help gathering extended knowledge on diseases. In answer to constantly declining autopsy rates non- or minimally invasive autopsy methods were introduced. Ultrasound is a well-established tool for imaging commonly used in clinical practice. This narrative review aims to summarize the current literature regarding the feasibility and validity of ultrasound in a forensic context.

Material and methods

A PubMed database search was carried out. Abstracts were scanned for pre-defined ex- and inclusion criteria, followed by a snowball search procedure applied to the primarily included articles.

Results

Forty-five publications met our inclusion criteria. The selected articles concern the feasibility of ultrasound in pre- or postmortem settings, forensic age estimation, and minimally invasive approaches. For imaging, ultrasound was deemed a reliable tool for the examination of epiphyses und superficial wounds, with limitations regarding internal organs and image quality due to postmortem changes. Ultrasound-guided minimally invasive approaches yielded higher success rates for adequate tissue sampling. Many investigations were carried out in low- and middle-income countries focusing on infectious diseases.

Conclusion

Ultrasound seems a promising but underutilized imaging tool in legal medicine to date. Promising approaches on its feasibility have been conducted. Especially for minimally invasive methods, ultrasound offered significant improvements on qualified biopsy sampling and thus appropriate diagnostics. Moreover, ultrasonic evaluation of epiphyses for age estimation offered valuable results. Nevertheless, further assessment of ultrasonic feasibility in forensic contexts is needed.

Population impact and effectiveness of sequential 13-valent pneumococcal conjugate and monovalent rotavirus vaccine introduction on infant mortality: prospective birth cohort studies from Malawi

Background

Pneumococcal conjugate vaccine (PCV) and rotavirus vaccine (RV) are key tools for reducing common causes of infant mortality. However, measurement of population-level mortality impact is lacking from sub-Saharan Africa. We evaluated mortality impact and vaccine effectiveness (VE) of PCV13 introduced in November 2011, with subsequent RV1 roll-out in October 2012, in Malawi.

Methods

We conducted two independent community-based birth cohort studies. Study 1, in northern Malawi (40000population), evaluated population impact using change-point analysis and negative-binomial regression of non-traumatic 14–51-week infant mortality preintroduction (1 January 2004 to 31 September 2011) and postintroduction (1 October 2011 to 1 July 2019), and against three-dose coverage. Study 2, in central Malawi (465 000 population), was recruited from 24 November 2011 to 1 June 2015. In the absence of preintroduction data, individual three-dose versus zero-dose VE was estimated using individual-level Cox survival models. In both cohorts, infants were followed with household visits to ascertain vaccination, socioeconomic and survival status. Verbal autopsies were conducted for deaths.

Results

Study 1 included 20 291 live births and 216 infant deaths. Mortality decreased by 28.6% (95% CI: 15.3 to 39.8) post-PCV13 introduction. A change point was identified in November 2012. Study 2 registered 50 731 live births, with 454 deaths. Infant mortality decreased from 17 to 10/1000 live births during the study period. Adjusted VE was 44.6% overall (95% CI: 23.0 to 59.1) and 48.3% (95% CI: −5.9 to 74.1) against combined acute respiratory infection, meningitis and sepsis-associated mortality.

Conclusion

These data provide population-level evidence of infant mortality reduction following sequential PCV13 and RV1 introduction into an established immunisation programme in Malawi. These data support increasing coverage of vaccine programmes in high-burden settings.

Accuracy of verbal autopsy, clinical data and minimally invasive autopsy in the evaluation of malaria-specific mortality: an observational study

BACKGROUND

Global malaria mortality estimates are hindered by the low reliability of the verbal autopsy (VA) and the clinical records, the most common sources of information used to estimate malaria-specific mortality. We aimed to determine the accuracy of these tools, as well as of the minimally invasive autopsy (MIA), a needle-based postmortem sampling method, to identify malaria-specific mortality in a large series of deceased patients from Mozambique, using complete autopsy as the gold standard.

METHODS

Observational study that included 264 deaths, occurring at a tertiary level hospital in Mozambique, from 1 November 2013 to 31 March 2015 (17 months-long period). Clinical data were abstracted, a computer coded VA was completed using the clinical data as source of information, and an MIA followed by a complete autopsy were performed. Screening for malaria infection was conducted postmortem to all participants using molecular and histological techniques (PCR and immunohistochemistry).

FINDINGS

Malaria infection was considered the cause of death in 6/264 (2.3%) cases: 2/54 children (3.7%, both less than 5 years old) and 4/57 (7.0%) maternal deaths. The sensitivity and specificity of the VA, the clinical data and the MIA to identify malaria-specific deaths were 33.3% and 96.1%, 66.7% and 96.1%, and 100% and 100%, respectively. In addition, malaria was identified as a possible contributor in 14 additional patients who died of other diseases. These cases were also accurately identified by the MIA (sensitivity 82.4%, specificity 100%).

INTERPRETATION

The high sensitivity and specificity of the MIA in identifying malaria may help to improve current estimates of malaria-specific mortality in endemic areas.

Limitations to current methods to estimate cause of death: a validation study of a verbal autopsy model

Background: Accurate information on causes of death (CoD) is essential to estimate burden of disease, track global progress, prioritize cost-effective interventions, and inform policies to reduce mortality. In low-income settings, where a significant proportion of deaths take place at home or in poorly-resourced peripheral health facilities, data on CoD often relies on verbal autopsies (VAs). Validations of VAs have been performed against clinical diagnosis, but never before against an acceptable gold standard: the complete diagnostic autopsy (CDA). Methods: We have validated a computer-coded verbal autopsy method -the InterVA- using individual and population metrics to determine CoD against the CDA, in 316 deceased patients of different age groups who died in a tertiary-level hospital in Maputo, Mozambique between 2013 and 2015.   Results: We found a low agreement of the model across all age groups at the individual (kappa statistic ranging from -0.030 to 0.232, lowest in stillbirths and highest in adults) and population levels (chance-corrected cause-specific mortality fraction accuracy ranging from -1.00 to 0.62, lowest in stillbirths, highest in children). The sensitivity in identifying infectious diseases was low (0% for tuberculosis, diarrhea, and disseminated infections, 32% for HIV-related infections, 33% for malaria and 36% for pneumonia). Of maternal deaths, 26 were assigned to eclampsia but only four patients actually died of eclampsia. Conclusions: These findings do not lead to building confidence in current estimates of CoD. They also call to the need to implement autopsy methods where they may be feasible, and to improve the quality and performance of current VA techniques.

BMI and All-Cause Mortality in a Population-Based Cohort in Rural South Africa

OBJECTIVE

This study evaluates the association between BMI and all-cause and cause-specific mortality in South Africa.

METHODS

Prospective, population-based observational cohort data from rural South Africa were analyzed. BMI was measured in 2010. Demographic characteristics were recorded and deaths were verified with verbal autopsy interview. The InterVA-5 tool was used to assign causes of death. HIV testing was conducted annually. Cox proportional hazards models were fit to estimate the effect of BMI on all-cause and cause-specific mortality, accounting for the competing risk of death from other causes. Models were adjusted for sociodemographic characteristics and HIV status, and inverse probability weighting for survey nonparticipation was used.

RESULTS

The cohort consisted of 9,728 individuals. In adjusted models, those with BMI of 25.0 to 29.9 kg/m² or 30.0 to 34.9 kg/m² had a lower hazard of death (adjusted hazard ratio: 0.80; 95% CI: 0.69-0.92 and adjusted hazard ratio: 0.75; 95% CI: 0.60-0.93, respectively) compared with those with BMI of 18.5 to 24.9 kg/m² .

CONCLUSIONS

Individuals in South Africa who meet clinically defined criteria for overweight or obesity had a lower risk of all-cause mortality than those with a normal BMI. These findings were stronger for women and communicable conditions.

Time-critical conditions: assessment of burden and access to care using verbal autopsy in Agincourt, South Africa

Background

Time-critical conditions (TCC) are estimated to cause substantial mortality in low and middle-income countries. However, quantification of deaths and identification of contributing factors to those deaths are challenging in settings with poor health records.

Aim

To use verbal autopsy (VA) data from the Agincourt health and sociodemographic surveillance system in rural South Africa to quantify the burden of deaths from TCC and to evaluate the barriers in seeking, reaching and receiving quality care for TCC leading to death.

Methodology

Deaths from 1993 to 2015 were analysed to identify causality from TCC. Deaths due to TCC were categorised as communicable, non-communicable, maternal, neonatal or injury-related. Proportion of deaths from TCC by age, sex, condition type and temporal trends was described. Deaths due to TCC from 2012 to 2015 were further examined by circumstances of mortality (CoM) indicators embedded in VA. Healthcare access, at illness onset and during the final day of life, as well as place of death, was extracted from free text summaries. Summaries were also analysed qualitatively using a Three Delays framework to identify barriers to healthcare.

Results

Of 15 305 deaths, 5885 (38.45%) were due to TCC. Non-communicable diseases were the most prevalent cause of death from TCC (2961/5885 cases, 50.31%). CoM indicators highlighted delays in a quarter of deaths due to TCC, most frequently in seeking care. The most common pattern of healthcare access was to die outwith a facility, having sought no healthcare (409/1324 cases, 30.89%). Issues in receipt of quality care were identified by qualitative analysis.

Conclusion

TCCs are responsible for a substantial burden of deaths in this rural South African population. Delays in seeking and receiving quality care were more prominent than those in reaching care, and thus further research and solution development should focus on healthcare-seeking behaviour and quality care provision.

Implementing and scaling verbal autopsies: into the unknown

Please see related article: http://bmcmedicine.biomedcentral.com/articles/10.1186/s12916-020-01520-1.

Migration and first-year maternal mortality among HIV-positive postpartum women: A population-based longitudinal study in rural South Africa

BACKGROUND

In South Africa, within-country migration is common. Mobility affects many of the factors in the pathway for entry to or retention in care among people living with HIV. We characterized the patterns of migration (i.e., change in residency) among peripartum women from rural South Africa and their association with first-year postpartum mortality.

METHODS AND FINDINGS

All pregnant women aged ≥15 years were followed-up during pregnancy and the first year postpartum in a population-based longitudinal demographic and HIV surveillance program in KwaZulu-Natal, South Africa, from 2000 to 2016. During the household surveys (every 4-6 months), each household head was interviewed to record demographic components of the household, including composition, migration, and mortality. External migration was defined as moving (i.e., change in residency) into or out of the study area. For women of reproductive age, detailed information on new pregnancy and birth was recorded. Maternal death was ascertained via verbal autopsy and HIV status at delivery via annual HIV surveys. We fitted mixed-effects Cox regression models adjusting for multiple pregnancies per individual. Overall, 19,334 women had 30,291 pregnancies: 3,339 were HIV-positive, 10,958 were HIV-negative, and 15,994 had unknown HIV status at delivery. The median age was 24 (interquartile range: 20-30) years. During pregnancy and the first year postpartum, 64% (n = 19,344) and 13% (n = 3,994) did not migrate and resided within and outside the surveillance area, respectively. Of the 23% who had externally migrated at least once, 39% delivered outside the surveillance area. Overall, the mortality rate was 5.8 per 1,000 person-years (or 831 deaths per 100,000 live births) in the first year postpartum. The major causes of deaths were AIDS- or tuberculosis-related conditions both within 42 days of delivery (53%) and during the first year postpartum (62%). In this study, we observed that HIV-positive peripartum women who externally migrated and delivered outside the surveillance area had a hazard of mortality more than two times greater (hazard ratio = 2.74; 95% confidence interval 1.01-7.40, p-value = 0.047)-after adjusting for age, time period (before or after 2010), and sociodemographic status-compared to that of HIV-positive women who continuously resided within the surveillance area. Study limitations include lack of data on access to antiretroviral therapy (ART) care and social or clinical context at the destinations among mobile participants, which could lead to unmeasured confounding. Further information on how mobile postpartum women access and remain in care would be instructive.

CONCLUSIONS

In this study, we found that a substantial portion of peripartum women moved within the country around the time of delivery and experienced a significantly higher risk of mortality. Despite the scale-up of universal ART and declining trends in maternal mortality, there is an urgent need to derive a greater understanding of the mechanisms underlying this finding and to develop targeted interventions for mobile HIV-positive peripartum women.

Mortality due to road injuries in the states of India: the Global Burden of Disease Study 1990-2017

Background

A systematic understanding of population-level trends in deaths due to road injuries at the subnational level over time for India's 1·4 billion people, by age, sex, and type of road user is not readily available; we aimed to fill this knowledge gap.

Methods

As part of the Global Burden of Diseases, Injuries, and Risk Factors Study, we estimated the rate of deaths due to road injuries in each state of India from 1990 to 2017 based on several verbal autopsy data sources. We calculated the number of deaths and death rate for road injuries by type of road user, and assessed the age and sex distribution of these deaths over time. Based on the trends of the age-standardised death rate from 1990 to 2017, we projected the age-standardised death rate to 2030 to assess if the states of India would meet the Sustainable Development Goal (SDG) target to halve the death rate for road injuries from 2015 by 2020 or 2030. We calculated 95% uncertainty intervals (UIs) for the point estimates.

Findings

In 2017, 218 876 deaths (95% UI 201 734 to 231 141) due to road injuries occurred in India, with an age-standardised death rate for road injuries of 17·2 deaths (15·7 to 18·1) per 100 000 population, which was much higher in males (25·7 deaths [23·5 to 27·4] per 100 000) than in females (8·5 deaths [7·2 to 9·1] per 100 000). The number of deaths due to road injuries in India increased by 58·7% (43·6 to 74·7) from 1990 to 2017, but the age-standardised death rate decreased slightly, by 9·2% (0·6 to 18·3). In 2017, pedestrians accounted for 76 729 (35·1%) of all deaths due to road injuries, motorcyclists accounted for 67 524 (30·9%), motor vehicle occupants accounted for 57 802 (26·4%), and cyclists accounted for 15 324 (7·0%). India had a higher age-standardised death rate for road injury among motorcyclists (4·9 deaths [3·9–5·4] per 100 000 population) and cyclists (1·2 deaths [0·9–1·4] per 100 000 population) than the global average. Road injury was the leading cause of death in males aged 15 to 39 years in India in 2017, and the second leading cause in this age group for both sexes combined. The overall age-standardised death rate for road injuries varied by up to 2·6 times between states in 2017. Wide variations were seen between the states in the percentage change in age-standardised death rate for road injuries from 1990 to 2017, ranging from a reduction of 38·2% (22·3 to 51·7) in Delhi to an increase of 17·0% (0·6 to 34·7) in Odisha. If the trends estimated up to 2017 were to continue, no state in India or India overall would achieve the SDG 2020 target in 2020 or even in 2030.

Interpretation

India's contribution to the global number of deaths due to road injuries is increasing, and the country is unlikely to meet the SDG targets if the trends up to 2017 continue. India needs to implement evidence-based road safety interventions, promote strong policies and traffic law enforcement, have better road and vehicle design, and improve care for road injuries at the state level to meet the SDG goal.

Funding

Bill & Melinda Gates Foundation and Indian Council of Medical Research, Department of Health Research, Ministry of Health and Family Welfare, Government of India.

Neonatal azithromycin administration to prevent infant mortality: study protocol for a randomised controlled trial

INTRODUCTION

Biannual mass azithromycin distribution to children aged 1-59 months has been shown to reduce all-cause mortality. Children under 28 days of age were not treated in studies evaluating mass azithromycin distribution for child mortality due to concerns related to infantile hypertrophic pyloric stenosis (IHPS). Here, we report the design of a randomised controlled trial to evaluate the efficacy and safety of administration of a single dose of oral azithromycin during the neonatal period.

METHODS AND ANALYSIS

The Nouveaux-nés et Azithromycine: une Innovation dans le Traitement des Enfants (NAITRE) study is a double-masked randomised placebo-controlled trial designed to evaluate the efficacy of a single dose of azithromycin (20 mg/kg) for the prevention of child mortality. Newborns (n=21 712) aged 8-27 days weighing at least 2500 g are 1:1 randomised to a single, directly observed, oral dose of azithromycin or matching placebo. Participants are followed weekly for 3 weeks after treatment to screen for adverse events, including IHPS. The primary outcome is all-cause mortality at the 6-month study visit.

ETHICS AND DISSEMINATION

This study was approved by the Institutional Review Boards at the University of California, San Francisco in San Francisco, USA (Protocol #18-25027) and the Comité National d'Ethique pour la Recherche in Ouagadougou, Burkina Faso (Protocol #2018-10-123). The findings of this trial will be presented at local, regional and international meetings and published in open access peer-reviewed journals.

TRIAL REGISTRATION NUMBER

NCT03682653; Pre-results.

Validation of verbal autopsy methods for assessment of child mortality in sub-Saharan Africa and the policy implication: a rapid review

Reliable data on the cause of child death is the cornerstone for evidence-informed health policy making towards improving child health outcomes. Unfortunately, accurate data on cause of death is essentially lacking in most countries of sub-Saharan Africa due to the widespread absence of functional Civil Registration and Vital Statistics (CRVS) systems. To address this problem, verbal autopsy (VA) has gained prominence as a strategy for obtaining Cause of Death (COD) information in populations where CRVS are absent. This study reviewed publications that investigated the validation of VA methods for assessment of COD. A MEDLINE PubMed search was undertaken in June 2018 for studies published in English that investigated the validation of VA methods in sub-Saharan Africa from 1990-2018. Of the 17 studies identified, 9 fulfilled the study inclusion criteria from which additional five relevant studies were found by reviewing their references. The result showed that Physician-Certified Verbal Autopsy (PCVA) was the most widely used VA method. Validation studies comparing PCVA to hospital records, expert algorithm and InterVA demonstrated mixed and highly varied outcomes. The accuracy and reliability of the VA methods depended on level of healthcare the respondents have access to and the knowledge of the physicians on the local disease aetiology and epidemiology. As the countries in sub-Saharan Africa continue to battle with dysfunctional CRVS system, VA will remain the only viable option for the supply of child mortality data necessary for policy making.

Modelling the relationship between malaria prevalence as a measure of transmission and mortality across age groups

Background

Parasite prevalence has been used widely as a measure of malaria transmission, especially in malaria endemic areas. However, its contribution and relationship to malaria mortality across different age groups has not been well investigated. Previous studies in a health and demographic surveillance systems (HDSS) platform in western Kenya quantified the contribution of incidence and entomological inoculation rates (EIR) to mortality. The study assessed the relationship between outcomes of malaria parasitaemia surveys and mortality across age groups.

Methods

Parasitological data from annual cross-sectional surveys from the Kisumu HDSS between 2007 and 2015 were used to determine malaria parasite prevalence (PP) and clinical malaria (parasites plus reported fever within 24 h or temperature above 37.5 °C). Household surveys and verbal autopsy (VA) were used to obtain data on all-cause and malaria-specific mortality. Bayesian negative binomial geo-statistical regression models were used to investigate the association of PP/clinical malaria with mortality across different age groups. Estimates based on yearly data were compared with those from aggregated data over 4 to 5-year periods, which is the typical period that mortality data are available from national demographic and health surveys.

Results

Using 5-year aggregated data, associations were established between parasite prevalence and malaria-specific mortality in the whole population (RR_malaria = 1.66; 95% Bayesian Credible Intervals: 1.07–2.54) and children 1–4 years (RR_malaria = 2.29; 1.17–4.29). While clinical malaria was associated with both all-cause and malaria-specific mortality in combined ages (RR_all-cause = 1.32; 1.01–1.74); (RR_malaria = 2.50; 1.27–4.81), children 1–4 years (RR_all-cause = 1.89; 1.00–3.51); (RR_malaria = 3.37; 1.23–8.93) and in older children 5–14 years (RR_all-cause = 3.94; 1.34–11.10); (RR_malaria = 7.56; 1.20–39.54), no association was found among neonates, adults (15–59 years) and the elderly (60+ years). Distance to health facilities, socioeconomic status, elevation and survey year were important factors for all-cause and malaria-specific mortality.

Conclusion

Malaria parasitaemia from cross-sectional surveys was associated with mortality across age groups over 4 to 5 year periods with clinical malaria more strongly associated with mortality than parasite prevalence. This effect was stronger in children 5–14 years compared to other age-groups. Further analyses of data from other HDSS sites or similar platforms would be useful in investigating the relationship between malaria and mortality across different endemicity levels.

Electronic supplementary material

The online version of this article (10.1186/s12936-019-2869-9) contains supplementary material, which is available to authorized users.

Automated versus physician assignment of cause of death for verbal autopsies: randomized trial of 9374 deaths in 117 villages in India

BACKGROUND

Verbal autopsies with physician assignment of cause of death (COD) are commonly used in settings where medical certification of deaths is uncommon. It remains unanswered if automated algorithms can replace physician assignment.

METHODS

We randomized verbal autopsy interviews for deaths in 117 villages in rural India to either physician or automated COD assignment. Twenty-four trained lay (non-medical) surveyors applied the allocated method using a laptop-based electronic system. Two of 25 physicians were allocated randomly to independently code the deaths in the physician assignment arm. Six algorithms (Naïve Bayes Classifier (NBC), King-Lu, InSilicoVA, InSilicoVA-NT, InterVA-4, and SmartVA) coded each death in the automated arm. The primary outcome was concordance with the COD distribution in the standard physician-assigned arm. Four thousand six hundred fifty-one (4651) deaths were allocated to physician (standard), and 4723 to automated arms.

RESULTS

The two arms were nearly identical in demographics and key symptom patterns. The average concordances of automated algorithms with the standard were 62%, 56%, and 59% for adult, child, and neonatal deaths, respectively. Automated algorithms showed inconsistent results, even for causes that are relatively easy to identify such as road traffic injuries. Automated algorithms underestimated the number of cancer and suicide deaths in adults and overestimated other injuries in adults and children. Across all ages, average weighted concordance with the standard was 62% (range 79-45%) with the best to worst ranking automated algorithms being InterVA-4, InSilicoVA-NT, InSilicoVA, SmartVA, NBC, and King-Lu. Individual-level sensitivity for causes of adult deaths in the automated arm was low between the algorithms but high between two independent physicians in the physician arm.

CONCLUSIONS

While desirable, automated algorithms require further development and rigorous evaluation. Lay reporting of deaths paired with physician COD assignment of verbal autopsies, despite some limitations, remains a practicable method to document the patterns of mortality reliably for unattended deaths.

TRIAL REGISTRATION

ClinicalTrials.gov , NCT02810366. Submitted on 11 April 2016.

An integrated approach to processing WHO-2016 verbal autopsy data: the InterVA-5 model

BACKGROUND

Verbal autopsy is an increasingly important methodology for assigning causes to otherwise uncertified deaths, which amount to around 50% of global mortality and cause much uncertainty for health planning. The World Health Organization sets international standards for the structure of verbal autopsy interviews and for cause categories that can reasonably be derived from verbal autopsy data. In addition, computer models are needed to efficiently process large quantities of verbal autopsy interviews to assign causes of death in a standardised manner. Here, we present the InterVA-5 model, developed to align with the WHO-2016 verbal autopsy standard. This is a harmonising model that can process input data from WHO-2016, as well as earlier WHO-2012 and Tariff-2 formats, to generate standardised cause-specific mortality profiles for diverse contexts. The software development involved building on the earlier InterVA-4 model, and the expanded knowledge base required for InterVA-5 was informed by analyses from a training dataset drawn from the Population Health Metrics Research Collaboration verbal autopsy reference dataset, as well as expert input.

RESULTS

The new model was evaluated against a test dataset of 6130 cases from the Population Health Metrics Research Collaboration and 4009 cases from the Afghanistan National Mortality Survey dataset. Both of these sources contained around three quarters of the input items from the WHO-2016, WHO-2012 and Tariff-2 formats. Cause-specific mortality fractions across all applicable WHO cause categories were compared between causes assigned in participating tertiary hospitals and InterVA-5 in the test dataset, with concordance correlation coefficients of 0.92 for children and 0.86 for adults. The InterVA-5 model's capacity to handle different input formats was evaluated in the Afghanistan dataset, with concordance correlation coefficients of 0.97 and 0.96 between the WHO-2016 and the WHO-2012 format for children and adults respectively, and 0.92 and 0.87 between the WHO-2016 and the Tariff-2 format respectively.

CONCLUSIONS

Despite the inherent difficulties of determining "truth" in assigning cause of death, these findings suggest that the InterVA-5 model performs well and succeeds in harmonising across a range of input formats. As more primary data collected under WHO-2016 become available, it is likely that InterVA-5 will undergo minor re-versioning in the light of practical experience. The model is an important resource for measuring and evaluating cause-specific mortality globally.

Community health workers trained to conduct verbal autopsies provide better mortality measures than existing surveillance: Results from a cross-sectional study in rural western Uganda

Background

In much of sub-Saharan Africa, health facilities serve as the primary source of routine vital statistics. These passive surveillance systems, however, are plagued by infrequent and unreliable reporting and do not capture events that occur outside of the formal health sector. Verbal autopsies (VA) have been utilized to estimate the burden and causes of mortality where civil registration and vital statistics systems are weak, but VAs have not been widely employed in national surveillance systems. In response, we trained lay community health workers (CHW) in a rural sub-county of western Uganda to conduct VA interviews in order to assess the feasibility of leveraging CHW to measure the burden of disease in resource limited settings.

Methods and findings

Trained CHWs conducted a cross-sectional survey of the 36 villages comprising the Bugoye sub-county to identify all deaths occurring in the prior year. The sub county has an estimated population of 50,249, approximately one-quarter of whom are children under 5 years of age (25.3%). When an eligible death was reported, CHWs administered a WHO 2014 VA questionnaire, the results of which were analyzed using the InterVA-4 tool. To compare the findings of the CHW survey to existing surveillance systems, study staff reviewed inpatient registers from neighboring referral health facilities in an attempt to match recorded deaths to those identified by the survey. Overall, CHWs conducted high quality VA interviews on direct observation, identifying 230 deaths that occurred within the sub-county, including 77 (33.5%) among children under five years of age. More than half of the deaths (123 of 230, 53.5%) were reported to have occurred outside a health facility and thus would not be captured by passive surveillance. More than two-thirds (73 of 107, 68.2%) of facility deaths took place in one of three nearby hospitals, yet only 35 (47.9%) were identified on our review of inpatient registers. Consistent with previous VA studies, the leading causes of death among children under five years of age were malaria (19.5%), prematurity (19.5%), and neonatal pneumonia (15.6%). while among adults, HIV/AIDS-related deaths illness (13.6%), pulmonary tuberculosis (11.4%) and malaria (8.6%) were the leading causes of death. No child deaths identified from inpatient registers listed HIV/AIDS as a cause of death despite 8 deaths (10.4%) attributed to HIV/AIDS as determined by VA.

Conclusions

Lay CHWs are able to conduct high quality VA interviews to capture critical information that can be analyzed using standard methodologies to provide a more complete estimate of the burden and causes of mortality. Similar approaches can be scaled to improve the measurement of vital statistics in order to facilitate appropriate public health interventions in rural areas of sub-Saharan Africa.

Integrating community-based verbal autopsy into civil registration and vital statistics (CRVS): system-level considerations

Background: Reliable and representative cause of death (COD) statistics are essential to inform public health policy, respond to emerging health needs, and document progress towards Sustainable Development Goals. However, less than one-third of deaths worldwide are assigned a cause. Civil registration and vital statistics (CRVS) systems in low- and lower-middle-income countries are failing to provide timely, complete and accurate vital statistics, and it will still be some time before they can provide physician-certified COD for every death.

Proposals: Verbal autopsy (VA) is a method to ascertain the probable COD and, although imperfect, it is the best alternative in the absence of medical certification. There is extensive experience with VA in research settings but only a few examples of its use on a large scale. Data collection using electronic questionnaires on mobile devices and computer algorithms to analyse responses and estimate probable COD have increased the potential for VA to be routinely applied in CRVS systems. However, a number of CRVS and health system integration issues should be considered in planning, piloting and implementing a system-wide intervention such as VA. These include addressing the multiplicity of stakeholders and sub-systems involved, integration with existing CRVS work processes and information flows, linking VA results to civil registration records, information technology requirements and data quality assurance.

Conclusions: Integrating VA within CRVS systems is not simply a technical undertaking. It will have profound system-wide effects that should be carefully considered when planning for an effective implementation. This paper identifies and discusses the major system-level issues and emerging practices, provides a planning checklist of system-level considerations and proposes an overview for how VA can be integrated into routine CRVS systems.

Nations within a nation: variations in epidemiological transition across the states of India, 1990-2016 in the Global Burden of Disease Study

BACKGROUND

18% of the world's population lives in India, and many states of India have populations similar to those of large countries. Action to effectively improve population health in India requires availability of reliable and comprehensive state-level estimates of disease burden and risk factors over time. Such comprehensive estimates have not been available so far for all major diseases and risk factors. Thus, we aimed to estimate the disease burden and risk factors in every state of India as part of the Global Burden of Disease (GBD) Study 2016.

METHODS

Using all available data sources, the India State-Level Disease Burden Initiative estimated burden (metrics were deaths, disability-adjusted life-years [DALYs], prevalence, incidence, and life expectancy) from 333 disease conditions and injuries and 84 risk factors for each state of India from 1990 to 2016 as part of GBD 2016. We divided the states of India into four epidemiological transition level (ETL) groups on the basis of the ratio of DALYs from communicable, maternal, neonatal, and nutritional diseases (CMNNDs) to those from non-communicable diseases (NCDs) and injuries combined in 2016. We assessed variations in the burden of diseases and risk factors between ETL state groups and between states to inform a more specific health-system response in the states and for India as a whole.

FINDINGS

DALYs due to NCDs and injuries exceeded those due to CMNNDs in 2003 for India, but this transition had a range of 24 years for the four ETL state groups. The age-standardised DALY rate dropped by 36·2% in India from 1990 to 2016. The numbers of DALYs and DALY rates dropped substantially for most CMNNDs between 1990 and 2016 across all ETL groups, but rates of reduction for CMNNDs were slowest in the low ETL state group. By contrast, numbers of DALYs increased substantially for NCDs in all ETL state groups, and increased significantly for injuries in all ETL state groups except the highest. The all-age prevalence of most leading NCDs increased substantially in India from 1990 to 2016, and a modest decrease was recorded in the age-standardised NCD DALY rates. The major risk factors for NCDs, including high systolic blood pressure, high fasting plasma glucose, high total cholesterol, and high body-mass index, increased from 1990 to 2016, with generally higher levels in higher ETL states; ambient air pollution also increased and was highest in the low ETL group. The incidence rate of the leading causes of injuries also increased from 1990 to 2016. The five leading individual causes of DALYs in India in 2016 were ischaemic heart disease, chronic obstructive pulmonary disease, diarrhoeal diseases, lower respiratory infections, and cerebrovascular disease; and the five leading risk factors for DALYs in 2016 were child and maternal malnutrition, air pollution, dietary risks, high systolic blood pressure, and high fasting plasma glucose. Behind these broad trends many variations existed between the ETL state groups and between states within the ETL groups. Of the ten leading causes of disease burden in India in 2016, five causes had at least a five-times difference between the highest and lowest state-specific DALY rates for individual causes.

INTERPRETATION

Per capita disease burden measured as DALY rate has dropped by about a third in India over the past 26 years. However, the magnitude and causes of disease burden and the risk factors vary greatly between the states. The change to dominance of NCDs and injuries over CMNNDs occurred about a quarter century apart in the four ETL state groups. Nevertheless, the burden of some of the leading CMNNDs continues to be very high, especially in the lowest ETL states. This comprehensive mapping of inequalities in disease burden and its causes across the states of India can be a crucial input for more specific health planning for each state as is envisioned by the Government of India's premier think tank, the National Institution for Transforming India, and the National Health Policy 2017.

FUNDING

Bill & Melinda Gates Foundation; Indian Council of Medical Research, Department of Health Research, Ministry of Health and Family Welfare, Government of India; and World Bank.

Determining the Cause of Death: Mortality Surveillance Using Verbal Autopsy in Indonesia

In the absence of a vital registration and health information systems, Indonesia does not have complete, accurate, and continuous data to summarize the mortality statistics of the population, nor determine the exact cause of death. Verbal autopsies performed in a community-based mortality surveillance have been used to provide information on the cause of deaths in such context. However, physician review of verbal autopsy can be expensive, time-consuming, and give inconsistent results, raising concern about its reliability. We used the Purworejo Health and Demographic Surveillance System's (HDSS) mortality data collected between 2000 and 2002 and assigned causes of death for all age groups using Interpreting Verbal Autopsy-4, analytic software that applies a probabilistic model. A total of 1,999 deaths were identified among 55,581 individuals surveyed in 14,409 households; 830 deaths were able to be recorded using the standardized World Health Organization (WHO) verbal autopsy questionnaire. We calculated the proportion of different causes of death and its incidence rate (IR) ratios with 95% confidence interval (CI) to compare the IR per person-years-observation (PYO). The IR of stroke was 126.7 per 100,000 PYO (95% CI: 109.7, 143.7); acute respiratory infection including pneumonia was 70.8 per 100,000 PYO (95% CI: 58.1, 83.5); and the IR of other and unspecified cardiac diseases was 57.7 per 100,000 PYO (95% CI: 46.2, 69.2). Stroke was indicated as the leading cause of death among elderly people aged 50 years and above. Meanwhile, pneumonia as a communicable disease was found to be the most common cause of death among both 0-14-year-old children and elderly people.

The long road to elimination: malaria mortality in a South African population cohort over 21 years

Background

Malaria elimination is on global agendas following successful transmission reductions. Nevertheless moving from low to zero transmission is challenging. South Africa has an elimination target of 2018, which may or may not be realised in its hypoendemic areas.

Methods

The Agincourt Health and Demographic Surveillance System has monitored population health in north-eastern South Africa since 1992. Malaria deaths were analysed against individual factors, socioeconomic status, labour migration and weather over a 21-year period, eliciting trends over time and associations with covariates.

Results

Of 13 251 registered deaths over 1.58 million person-years, 1.2% were attributed to malaria. Malaria mortality rates increased from 1992 to 2013, while mean daily maximum temperature rose by 1.5 °C. Travel to endemic Mozambique became easier, and malaria mortality increased in higher socioeconomic groups. Overall, malaria mortality was significantly associated with age, socioeconomic status, labour migration and employment, yearly rainfall and higher rainfall/temperature shortly before death.

Conclusions

Malaria persists as a small but important cause of death in this semi-rural South African population. Detailed longitudinal population data were crucial for these analyses. The findings highlight practical political, socioeconomic and environmental difficulties that may also be encountered elsewhere in moving from low-transmission scenarios to malaria elimination.

Bayesian spatio-temporal modeling of mortality in relation to malaria incidence in Western Kenya

INTRODUCTION

The effect of malaria exposure on mortality using health facility incidence data as a measure of transmission has not been well investigated. Health and demographic surveillance systems (HDSS) routinely capture data on mortality, interventions and other household related indicators, offering a unique platform for estimating and monitoring the incidence-mortality relationship in space and time.

METHODS

Mortality data from the HDSS located in Western Kenya collected from 2007 to 2012 and linked to health facility incidence data were analysed using Bayesian spatio-temporal survival models to investigate the relation between mortality (all-cause/malaria-specific) and malaria incidence across all age groups. The analysis adjusted for insecticide-treated net (ITN) ownership, socio-economic status (SES), distance to health facilities and altitude. The estimates obtained were used to quantify excess mortality due to malaria exposure.

RESULTS

Our models identified a strong positive relationship between slide positivity rate (SPR) and all-cause mortality in young children 1-4 years (HR = 4.29; 95% CI: 2.78-13.29) and all ages combined (HR = 1.55; 1.04-2.80). SPR had a strong positive association with malaria-specific mortality in young children (HR = 9.48; 5.11-37.94), however, in older children (5-14 years), it was associated with a reduction in malaria specific mortality (HR = 0.02; 0.003-0.33).

CONCLUSION

SPR as a measure of transmission captures well the association between malaria transmission intensity and all-cause/malaria mortality. This offers a quick and efficient way to monitor malaria burden. Excess mortality estimates indicate that small changes in malaria incidence substantially reduce overall and malaria specific mortality.

Space-time patterns in maternal and mother mortality in a rural South African population with high HIV prevalence (2000-2014): results from a population-based cohort

Background

International organs such as, the African Union and the South African Government view maternal health as a dominant health prerogative. Even though most countries are making progress, maternal mortality in South Africa (SA) significantly increased between 1990 and 2015, and prevented the country from achieving Millennium Development Goal 5. Elucidating the space-time patterns and risk factors of maternal mortality in a rural South African population could help target limited resources and policy guidelines to high-risk areas for the greatest impact, as more generalized interventions are costly and often less effective.

Methods

Population-based mortality data from 2000 to 2014 for women aged 15–49 years from the Africa Centre Demographic Information System located in the Umkhanyakude district of KwaZulu-Natal Province, South Africa were analysed. Our outcome was classified into two definitions: Maternal mortality; the death of a woman while pregnant or within 42 days of cessation of pregnancy, regardless of the duration and site of the pregnancy, from any cause related to or exacerbated by the pregnancy or its management but not from unexpected or incidental causes; and ‘Mother death’; death of a mother whilst child is less than 5 years of age. Both the Kulldorff and Tango spatial scan statistics for regular and irregular shaped cluster detection respectively were used to identify clusters of maternal mortality events in both space and time.

Results

The overall maternal mortality ratio was 650 per 100,000 live births, and 1204 mothers died while their child was less than or equal to 5 years of age, of a mortality rate of 370 per 100,000 children. Maternal mortality declined over the study period from approximately 600 per 100,000 live births in 2000 to 400 per 100,000 live births in 2014. There was no strong evidence of spatial clustering for maternal mortality in this rural population. However, the study identified a significant spatial cluster of mother deaths in childhood (p = 0.022) in a peri-urban community near the national road. Based on our multivariable logistic regression model, HIV positive status (Adjusted odds ratio [aOR] = 2.5, CI 95%: [1.5–4.2]; primary education or less (aOR = 1.97, CI 95%: [1.04–3.74]) and parity (aOR = 1.42, CI 95%: [1.24–1.63]) were significant predictors of maternal mortality.

Conclusions

There has been an overall decrease in maternal and mother death between 2000 and 2014. The identification of a clear cluster of mother deaths shows the possibility of targeting intervention programs in vulnerable communities, as population-wide interventions may be ineffective and too costly to implement.

Measuring mortality due to HIV-associated tuberculosis among adults in South Africa: Comparing verbal autopsy, minimally-invasive autopsy, and research data

BACKGROUND

The World Health Organization (WHO) aims to reduce tuberculosis (TB) deaths by 95% by 2035; tracking progress requires accurate measurement of TB mortality. International Classification of Diseases (ICD) codes do not differentiate between HIV-associated TB and HIV more generally. Verbal autopsy (VA) is used to estimate cause of death (CoD) patterns but has mostly been validated against a suboptimal gold standard for HIV and TB. This study, conducted among HIV-positive adults, aimed to estimate the accuracy of VA in ascertaining TB and HIV CoD when compared to a reference standard derived from a variety of clinical sources including, in some, minimally-invasive autopsy (MIA).

METHODS AND FINDINGS

Decedents were enrolled into a trial of empirical TB treatment or a cohort exploring diagnostic algorithms for TB in South Africa. The WHO 2012 instrument was used; VA CoD were assigned using physician-certified VA (PCVA), InterVA-4, and SmartVA-Analyze. Reference CoD were assigned using MIA, research, and health facility data, as available. 259 VAs were completed: 147 (57%) decedents were female; median age was 39 (interquartile range [IQR] 33-47) years and CD4 count 51 (IQR 22-102) cells/μL. Compared to reference CoD that included MIA (n = 34), VA underestimated mortality due to HIV/AIDS (94% reference, 74% PCVA, 47% InterVA-4, and 41% SmartVA-Analyze; chance-corrected concordance [CCC] 0.71, 0.42, and 0.31, respectively) and HIV-associated TB (41% reference, 32% PCVA; CCC 0.23). For individual decedents, all VA methods agreed poorly with reference CoD that did not include MIA (n = 259; overall CCC 0.14, 0.06, and 0.15 for PCVA, InterVA-4, and SmartVA-Analyze); agreement was better at population level (cause-specific mortality fraction accuracy 0.78, 0.61, and 0.57, for the three methods, respectively).

CONCLUSIONS

Current VA methods underestimate mortality due to HIV-associated TB. ICD and VA methods need modifications that allow for more specific evaluation of HIV-related deaths and direct estimation of mortality due to HIV-associated TB.

Verbal autopsy-assigned causes of death among adults being investigated for TB in South Africa

Background

Adults being investigated for TB in South Africa experience high mortality, yet causes of death (CoD) are not well defined. We determined CoD in this population using verbal autopsy (VA), and compared HIV- and TB-associated CoD using physician-certified verbal autopsy (PCVA) and InterVA-4 software.

Methods

All contactable consenting caregivers of participants who died during a trial comparing Xpert MTB/RIF to smear microscopy were interviewed using the WHO VA tool. CoD were assigned using PCVA and InterVA-4. Kappa statistic (K) and concordance correlation coefficient (CCC) were calculated for comparison.

Results

Among 231 deaths, relatives of 137 deceased were interviewed. Of the 137 deceased 76 (55.4%) were males, median age 41 years (IQR 33–50). PCVA assigned 70 (51.1%) TB immediate CoD (44 [62.8%] pulmonary TB; 26 [37.1%] extra-pulmonary TB); 21 (15.3%) HIV/AIDS-related; and 46 (33.5%) other CoD. InterVA-4 assigned 48 (35.0%) TB deaths; 49 (35.7%) HIV/AIDS-related deaths; and 40 (29.1%) other CoD. Agreement between PCVA and InterVA-4 CoD was slight at individual level (K=0.20; 95% CI 0.10–0.30) and poor at population level (CCC 0.67; 95% CI 0.38–0.99).

Conclusions

TB and HIV are leading CoD among adults being investigated for TB. PCVA and InterVA agreement at individual level was slight and poor at population level. VA methodology needs further development where TB and HIV are common.

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.

Innovations in health and demographic surveillance systems to establish the causal impacts of HIV policies

PURPOSE OF REVIEW

Health and demographic surveillance systems (HDSS), in conjunction with HIV treatment cohorts, have made important contributions to our understanding of the impact of HIV treatment and treatment-related interventions in sub-Saharan Africa. The purpose of this review is to describe and discuss innovations in data collection and data linkage that will create new opportunities to establish the impacts of HIV treatment, as well as policies affecting the treatment cascade, on population health and economic and social outcomes.

RECENT FINDINGS

Novel approaches to routine collection of biomarkers, behavioural data, spatial data, social network information, migration events and mobile phone records can significantly strengthen the potential of HDSS to generate exposure and outcome data for causal analysis of HIV treatment impact and policies affecting the HIV treatment cascade. Additionally, by linking HDSS data to health service administration, education and welfare service records, researchers can substantially broaden opportunities to establish how HIV treatment affects health and economic outcomes when delivered through public sector health systems and at scale.

SUMMARY

As the HIV treatment scaleup in sub-Saharan Africa enters its second decade, it is becoming increasingly important to understand the long-term causal impacts of large-scale HIV treatment and related policies on broader population health outcomes, such as noncommunicable diseases, as well as on economic and social outcomes, such as family welfare and children's educational attainment. By collecting novel data and linking existing data to public sector records, HDSS can create near-unique opportunities to contribute to this research agenda.

Factors Associated with Early Mortality in HIV-Positive Men and Women Investigated for Tuberculosis at Ethiopian Health Centers

Introduction

Despite increasing access to antiretroviral treatment (ART) in low-income countries, HIV-related mortality is high, especially in the first months following ART initiation. We aimed to evaluate the impact of TB coinfection on early mortality and to assess gender-specific predictors of mortality in a cohort of Ethiopian adults subjected to intensified casefinding for active TB before starting ART.

Material and Methods

Prospectively recruited ART-eligible adults (n = 812, 58.6% female) at five Ethiopian health centers were followed for 6 months. At inclusion sputum culture, Xpert MTB/RIF, and smear microscopy were performed (158/812 [19.5%] had TB). Primary outcome was all-cause mortality. We used multivariate Cox models to identify predictors of mortality.

Results

In total, 37/812 (4.6%) participants died, 12 (32.4%) of whom had TB. Karnofsky performance score (KPS) and mid-upper arm circumference (MUAC) were associated with mortality in the whole population. However, the associations were different in men and women. In men, only MUAC remained associated with mortality (adjusted hazard ratio [aHR] 0.71 [95% CI 0.57–0.88]). In women, KPS <80% was associated with mortality (aHR 10.95 [95% CI 2.33–51.49]), as well as presence of cough (aHR 3.98 [95% CI 1.10–14.36]). Cough was also associated with mortality for TB cases (aHR 8.30 [95% CI 1.06–65.14]), but not for non-TB cases.

Conclusions

In HIV-positive Ethiopian adults managed at health centers, mortality was associated with reduced performance score and malnutrition, with different distribution with regard to gender and TB coinfection. These robust variables could be used at clinic registration to identify persons at increased risk of early mortality.

Remotely Sensed Environmental Conditions and Malaria Mortality in Three Malaria Endemic Regions in Western Kenya

Background

Malaria is an important cause of morbidity and mortality in malaria endemic countries. The malaria mosquito vectors depend on environmental conditions, such as temperature and rainfall, for reproduction and survival. To investigate the potential for weather driven early warning systems to prevent disease occurrence, the disease relationship to weather conditions need to be carefully investigated. Where meteorological observations are scarce, satellite derived products provide new opportunities to study the disease patterns depending on remotely sensed variables. In this study, we explored the lagged association of Normalized Difference Vegetation Index (NVDI), day Land Surface Temperature (LST) and precipitation on malaria mortality in three areas in Western Kenya.

Methodology and Findings

The lagged effect of each environmental variable on weekly malaria mortality was modeled using a Distributed Lag Non Linear Modeling approach. For each variable we constructed a natural spline basis with 3 degrees of freedom for both the lag dimension and the variable. Lag periods up to 12 weeks were considered. The effect of day LST varied between the areas with longer lags. In all the three areas, malaria mortality was associated with precipitation. The risk increased with increasing weekly total precipitation above 20 mm and peaking at 80 mm. The NDVI threshold for increased mortality risk was between 0.3 and 0.4 at shorter lags.

Conclusion

This study identified lag patterns and association of remote- sensing environmental factors and malaria mortality in three malaria endemic regions in Western Kenya. Our results show that rainfall has the most consistent predictive pattern to malaria transmission in the endemic study area. Results highlight a potential for development of locally based early warning forecasts that could potentially reduce the disease burden by enabling timely control actions.

Naive Bayes classifiers for verbal autopsies: comparison to physician-based classification for 21,000 child and adult deaths

BACKGROUND

Verbal autopsies (VA) are increasingly used in low- and middle-income countries where most causes of death (COD) occur at home without medical attention, and home deaths differ substantially from hospital deaths. Hence, there is no plausible "standard" against which VAs for home deaths may be validated. Previous studies have shown contradictory performance of automated methods compared to physician-based classification of CODs. We sought to compare the performance of the classic naive Bayes classifier (NBC) versus existing automated classifiers, using physician-based classification as the reference.

METHODS

We compared the performance of NBC, an open-source Tariff Method (OTM), and InterVA-4 on three datasets covering about 21,000 child and adult deaths: the ongoing Million Death Study in India, and health and demographic surveillance sites in Agincourt, South Africa and Matlab, Bangladesh. We applied several training and testing splits of the data to quantify the sensitivity and specificity compared to physician coding for individual CODs and to test the cause-specific mortality fractions at the population level.

RESULTS

The NBC achieved comparable sensitivity (median 0.51, range 0.48-0.58) to OTM (median 0.50, range 0.41-0.51), with InterVA-4 having lower sensitivity (median 0.43, range 0.36-0.47) in all three datasets, across all CODs. Consistency of CODs was comparable for NBC and InterVA-4 but lower for OTM. NBC and OTM achieved better performance when using a local rather than a non-local training dataset. At the population level, NBC scored the highest cause-specific mortality fraction accuracy across the datasets (median 0.88, range 0.87-0.93), followed by InterVA-4 (median 0.66, range 0.62-0.73) and OTM (median 0.57, range 0.42-0.58).

CONCLUSIONS

NBC outperforms current similar COD classifiers at the population level. Nevertheless, no current automated classifier adequately replicates physician classification for individual CODs. There is a need for further research on automated classifiers using local training and test data in diverse settings prior to recommending any replacement of physician-based classification of verbal autopsies.

Mass HIV Treatment and Sex Disparities in Life Expectancy: Demographic Surveillance in Rural South Africa

BACKGROUND

Women have better patient outcomes in HIV care and treatment than men in sub-Saharan Africa. We assessed--at the population level--whether and to what extent mass HIV treatment is associated with changes in sex disparities in adult life expectancy, a summary metric of survival capturing mortality across the full cascade of HIV care. We also determined sex-specific trends in HIV mortality and the distribution of HIV-related deaths in men and women prior to and at each stage of the clinical cascade.

METHODS AND FINDINGS

Data were collected on all deaths occurring from 2001 to 2011 in a large population-based surveillance cohort (52,964 women and 45,688 men, ages 15 y and older) in rural KwaZulu-Natal, South Africa. Cause of death was ascertained by verbal autopsy (93% response rate). Demographic data were linked at the individual level to clinical records from the public sector HIV treatment and care program that serves the region. Annual rates of HIV-related mortality were assessed for men and women separately, and female-to-male rate ratios were estimated in exponential hazard models. Sex-specific trends in adult life expectancy and HIV-cause-deleted adult life expectancy were calculated. The proportions of HIV deaths that accrued to men and women at different stages in the HIV cascade of care were estimated annually. Following the beginning of HIV treatment scale-up in 2004, HIV mortality declined among both men and women. Female adult life expectancy increased from 51.3 y (95% CI 49.7, 52.8) in 2003 to 64.5 y (95% CI 62.7, 66.4) in 2011, a gain of 13.2 y. Male adult life expectancy increased from 46.9 y (95% CI 45.6, 48.2) in 2003 to 55.9 y (95% CI 54.3, 57.5) in 2011, a gain of 9.0 y. The gap between female and male adult life expectancy doubled, from 4.4 y in 2003 to 8.6 y in 2011, a difference of 4.3 y (95% CI 0.9, 7.6). For women, HIV mortality declined from 1.60 deaths per 100 person-years (95% CI 1.46, 1.75) in 2003 to 0.56 per 100 person-years (95% CI 0.48, 0.65) in 2011. For men, HIV-related mortality declined from 1.71 per 100 person-years (95% CI 1.55, 1.88) to 0.76 per 100 person-years (95% CI 0.67, 0.87) in the same period. The female-to-male rate ratio for HIV mortality declined from 0.93 (95% CI 0.82-1.07) in 2003 to 0.73 (95% CI 0.60-0.89) in 2011, a statistically significant decline (p = 0.046). In 2011, 57% and 41% of HIV-related deaths occurred among men and women, respectively, who had never sought care for HIV in spite of the widespread availability of free HIV treatment. The results presented here come from a poor rural setting in southern Africa with high HIV prevalence and high HIV treatment coverage; broader generalizability is unknown. Additionally, factors other than HIV treatment scale-up may have influenced population mortality trends.

CONCLUSIONS

Mass HIV treatment has been accompanied by faster declines in HIV mortality among women than men and a growing female-male disparity in adult life expectancy at the population level. In 2011, over half of male HIV deaths occurred in men who had never sought clinical HIV care. Interventions to increase HIV testing and linkage to care among men are urgently needed.

Cause-specific neonatal mortality: analysis of 3772 neonatal deaths in Nepal, Bangladesh, Malawi and India

OBJECTIVE

Understanding the causes of death is key to tackling the burden of three million annual neonatal deaths. Resource-poor settings lack effective vital registration systems for births, deaths and causes of death. We set out to describe cause-specific neonatal mortality in rural areas of Malawi, Bangladesh, Nepal and rural and urban India using verbal autopsy (VA) data.

DESIGN

We prospectively recorded births, neonatal deaths and stillbirths in seven population surveillance sites. VAs were carried out to ascertain cause of death. We applied descriptive epidemiological techniques and the InterVA method to characterise the burden, timing and causes of neonatal mortality at each site.

RESULTS

Analysis included 3772 neonatal deaths and 3256 stillbirths. Between 63% and 82% of neonatal deaths occurred in the first week of life, and males were more likely to die than females. Prematurity, birth asphyxia and infections accounted for most neonatal deaths, but important subnational and regional differences were observed. More than one-third of deaths in urban India were attributed to asphyxia, making it the leading cause of death in this setting.

CONCLUSIONS

Population-based VA methods can fill information gaps on the burden and causes of neonatal mortality in resource-poor and data-poor settings. Local data should be used to inform and monitor the implementation of interventions to improve newborn health. High rates of home births demand a particular focus on community interventions to improve hygienic delivery and essential newborn care.

Local perceptions of causes of death in rural South Africa: a comparison of perceived and verbal autopsy causes of death

Background

Understanding how lay people perceive the causes of mortality and their associated risk factors is important for public health. In resource-limited settings, where verbal autopsy (VA) is used as the most expedient method of determining cause of death, it is important to understand how pre-existing concepts of cause of death among VA-informants may influence their VA-responses and the consequential impact on cause of death assessment. This study describes the agreement between VA-derived causes of death and informant-perceived causes and associated influential factors, which also reflects lay health literacy in this setting.

Method

Using 20 years of VA data (n=11,228) from the Agincourt Health and Demographic Surveillance System (HDSS) site in rural South Africa, we explored the agreement between the causes of death perceived by the VA-informants and those assigned by the automated Inter-VA tool. Kappa statistics and concordance correlation coefficients were applied to measure agreement at individual and population levels, respectively. Multivariable regression models were used to explore factors associated with recognised lay perceptions of causes of mortality.

Results

Agreement between informant-perceived and VA-derived causes of death at the individual level was limited, but varied substantially by cause of death. However, agreement at the population level, comparing cause-specific mortality fractions was higher, with the notable exception of bewitchment as a cause. More recent deaths, those in adults aged 15–49 years, deaths outside the home, and those associated with external causes showed higher concordance with InterVA.

Conclusion

Overall, informant perception of causes of death was limited, but depended on informant characteristics and causes of death, and to some extent involved non-biomedical constructs. Understanding discordance between perceived and recognised causes of death is important for public health planning; low community understanding of causes of death may be detrimental to public health. These findings also illustrate the importance of using rigorous and standardised VA methods rather than relying on informants’ reported causes of death.

How much do the physician review and InterVA model agree in determining causes of death? A comparative analysis of deaths in rural Ethiopia

Background

Despite it is costly, slow and non-reproducible process, physician review (PR) is a commonly used method to interpret verbal autopsy data. However, there is a growing interest to adapt a new automated and internally consistent method called InterVA. This study evaluated the level of agreement in determining causes of death between PR and the InterVA model.

Methods

Verbal autopsy data for 434 cases collected between September 2009 and November 2012, were interpreted using both PR and the InterVA model. Cohen’s kappa statistic (κ) was used to compare the level of chance corrected case-by-case agreement in the diagnosis reached by the PR and InterVA model.

Results

Both methods gave comparable cause specific mortality fractions of communicable diseases (36.6 % by PR and 36.2 % by the model), non-communicable diseases (31.1 % by PR and 38.2 % by the model) and accidents/injuries (12.9 % by PR and 10.1 % by the model). The level of case-by-case chance corrected concordance between the two methods was 0.33 (95 % CI for κ = 0.29–0.34). The highest and lowest agreements were seen for accidents/injuries and non-communicable diseases; with κ = 0.75 and κ = 0.37, respectively.

Conclusion

If the InterVA were used in place of the existing PR process, the overall diagnosis would be fairly similar. The methods had better agreement in important public health diseases like; TB, perinatal causes, and pneumonia/sepsis; and lower in cardiovascular diseases and neoplasms. Therefore, both methods need to be validated against a gold-standard diagnosis of death.

Electronic supplementary material

The online version of this article (doi:10.1186/s12889-015-2032-7) contains supplementary material, which is available to authorized users.

Verbal Autopsy: Evaluation of Methods to Certify Causes of Death in Uganda

To assess different methods for determining cause of death from verbal autopsy (VA) questionnaire data, the intra-rater reliability of Physician-Certified Verbal Autopsy (PCVA) and the accuracy of PCVA, expert-derived (non-hierarchical) and data-driven (hierarchal) algorithms were assessed for determining common causes of death in Ugandan children. A verbal autopsy validation study was conducted from 2008-2009 in three different sites in Uganda. The dataset included 104 neonatal deaths (0-27 days) and 615 childhood deaths (1-59 months) with the cause(s) of death classified by PCVA and physician review of hospital medical records (the 'reference standard'). Of the original 719 questionnaires, 141 (20%) were selected for a second review by the same physicians; the repeat cause(s) of death were compared to the original,and agreement assessed using the Kappa statistic.Physician reviewers' refined non-hierarchical algorithms for common causes of death from existing expert algorithms, from which, hierarchal algorithms were developed. The accuracy of PCVA, non-hierarchical, and hierarchical algorithms for determining cause(s) of death from all 719 VA questionnaires was determined using the reference standard. Overall, intra-rater repeatability was high (83% agreement, Kappa 0.79 [95% CI 0.76-0.82]). PCVA performed well, with high specificity for determining cause of neonatal (>67%), and childhood (>83%) deaths, resulting in fairly accurate cause-specific mortality fraction (CSMF) estimates. For most causes of death in children, non-hierarchical algorithms had higher sensitivity, but correspondingly lower specificity, than PCVA and hierarchical algorithms, resulting in inaccurate CSMF estimates. Hierarchical algorithms were specific for most causes of death, and CSMF estimates were comparable to the reference standard and PCVA. Inter-rater reliability of PCVA was high, and overall PCVA performed well. Hierarchical algorithms performed better than non-hierarchical algorithms due to higher specificity and more accurate CSMF estimates. Use of PCVA to determine cause of death from VA questionnaire data is reasonable while automated data-driven algorithms are improved.